On organizations with many thousands of bank movements, the background processing that finalizes imported invoices and reconciles movements could stop before completing: some movements were not associated and some invoices stayed queued longer than expected.

The processing has been made more robust: movement reconciliation now runs in full even at large volumes, and finalization has the time it needs to complete each batch. Imported invoices are finalized and movements reconciled reliably, in waves within a few minutes of import.

No action is required: recovery is automatic.

The clinical dashboard run detail has a new Delivered tab, next to Doctors and Therapists. It answers a different question from the other two: not who refers the patients, but how much each professional produced, cost and returned, month by month.

Each month has three side-by-side columns - Delivered, Cost, Margin - so each professional's trend reads at a glance, month by month:

• Delivered: the value of the activities carried out.
• Cost: how much that professional invoiced us, from the accrual ledger, spread across the months by the competence period of the received invoices (not payments, not payroll).
• Margin: the difference between delivered and cost.

Clicking a number opens a breakdown showing what it is made of: the list of activities that make the delivered value for that month (date, patient, activity, amount) and the invoice quotas that make the cost, with the resulting margin.

To connect delivered and cost, each professional is linked to their suppliers in the registry. Exact name matches are linked automatically at import / AI processing (e.g. "Pepe Francesca" ↔ "Francesca Pepe"): no manual step for the obvious ones. For the rest, from the row a panel opens with suggestions to confirm, a manual search, and a New supplier button to create the supplier on the fly and link it without leaving the dialog. A professional can have several linked suppliers - costs sum across all of them.

The margin is computed only in the months with an invoice from the linked supplier, so a month still without invoices does not show an inflated margin. A month with no amounts stays empty rather than zero.

The table is filterable and sortable by column and exportable. Whoever set business units on suppliers can group the rows by them. Cost and margin are visible only to those with access to accounting data. Columns are only the months present in the data. The tab appears only when the imported file contains activity prices; on existing analyses the calculation runs when the page opens, with no re-import.

In the Profitability module, when opening an item row (Edit item row), the Healthcare staff variable costs field now has a € / % switch:

• € (default): enter the amount in euros, as before.
• %: enter a percentage of the average selling price and the app computes the euro cost for you, showing a preview of the value.

Handy when the staff cost is defined as a share of the selling price: no more recomputing it by hand. If the average selling price changes (manual edit or refresh from the clinical dashboard), the percentage-based cost realigns automatically.

File import also recognizes a staff-cost percentage column (e.g. a "% personale" header): imported rows are created directly in percentage mode, with the cost computed from the price.

When opening a document's detail to link its subject, you could previously pick only from subjects already in the registry: adding a new one meant leaving the screen, creating it in the registry, and coming back.

The subject field now uses the search selector: type to filter existing subjects and, if the subject is missing, a "Create new" entry appears that opens the creation form pre-filled with the typed name. On save the subject is created and linked right away to the document, without leaving the screen.

When importing a foreign supplier's purchase invoice as a PDF, the automatic extraction could swap the roles and set your own organization (shown on the document as the recipient, "Bill To") as the issuer instead of the supplier. This happened especially when the organization was the most prominent party or the only one with an Italian VAT number.

Extraction now knows which organization is importing its own documents and always assigns the counterpart to the other party on the invoice - the supplier. If, in a residual case, the extraction still returns the organization itself, the invoice is flagged and the counterpart is left empty, so you can set the correct supplier before finalizing instead of recording your organization as its own supplier.

XML electronic invoices were not affected: they already handled issuer and recipient correctly.

In the import dialogs (profitability rows, clinical data, master data) the Retry button appeared even when the import had succeeded - in particular when processing runs in the background and rows appear once finished. In that case there is nothing to retry and the button was confusing.

Now Retry appears only when something went wrong (error or cancellation). On a successful import only Close remains.

In the Clinical dashboard, Configuration tab, acting on a service (excluding/including with the eye icon, or changing its category) could accidentally select the entire services grid.

Fixed: each row now has a stable identity, so the selection stays on what you actually chose.

In the Clinical dashboard, Configuration tab, excluding a service from statistics is now independent of its category.

Previously exclusion was an entry in the classification menu: mapping a service to "Excluded" overwrote the type you had set, and re-including the service lost the category. Now they are two distinct data:

• The category menu is only for the clinical type (first visit, physiotherapy, follow-up...).
• Exclusion is toggled with an eye icon next to the service name (or with the Exclude/Include selected actions for multiple rows). An excluded service is shown struck through and enters no count.

Excluding no longer changes the category, and re-including restores it exactly as it was. Changes take effect with Recalculate statistics.

On Price list analysis (item profitability), importing a price list from a large Excel file (several MB) could leave the dialog stuck on "Sending for analysis..." and never actually start in the background.

Now the file is uploaded and processing (analysis + row extraction) continues as a background job, with progress in the bottom notification: the dialog frees up immediately and the rows appear in the configuration once finished. It also works with files of several MB, which previously exceeded the request limit and hung.

Two fixes on Price list analysis (item profitability).

Excel import in the background. When importing a price list from a long file, processing could stay stuck on "in progress" and never finish (the computation exceeded the request time limit). The import now runs in the background, like the app's other imports: it starts immediately, progress shows in the bottom notification, and the rows appear in the configuration once processing finishes, without blocking the page.

Duplicate working again. Duplicating a configuration could fail when it contained name-only rows (entries without a linked master item): the copy aborted with a generic error. Duplication now correctly copies the name of these rows too, so the copy succeeds and keeps all the price-list entries.

The Clinical dashboard now lets you exclude some services from the statistics. Certain rows in the file (for example rentals, consumables or administrative items) are not clinical activity to measure and were skewing the counts: they can now be taken out of the calculations.

In the Configuration tab, each service's classification menu has an Excluded from statistics option at the bottom. Choosing it keeps that service out of every KPI: total visits, unique patients, started-physio, drop-out and amounts.

To work in bulk, two mass actions are available: select the rows with the checkboxes and use Exclude selected or Include selected. Changes take effect once you press Recalculate statistics.

Exclusion is reversible at any time: the rows stay in the analysis, they are just skipped by the calculations, and the Data quality panel shows how many visits were excluded. Reclassify with AI also respects manually set exclusions, without bringing them back into the counts.

In the Documents list the Accounts column is now clearer. Instead of a bare icon there is a labeled badge:

• green Complete when all accounts on the ledger rows are set;
• amber N to set when at least one row has no account;
• - for documents not finalized yet.

The badge is tappable (no more mouse hover required, so it works on mobile too): it opens the list of ledger accounts - truncated with an expandable "+N more" when there are many - and an Open document shortcut to go and set them.

The column can also be filtered by status: from the column filter you pick Complete and/or Incomplete, to spot at a glance the documents whose accounts still need setting (an alternative to the "Incomplete" toggle already in the header).

When reprocessing received invoices with AI processing, the supplier's default account is now respected: if a default account has been set for that supplier, its lines are classified to that account.

This keeps classification consistent per supplier: after reprocessing restarts from scratch (without reusing the previous classification of the same invoices), the supplier default ensures the same supplier always lands on the same account. This is the same behavior already active on Movements. If a supplier has no default account, nothing changes.

In the in-grid account picker - the Account column of Movements and the Account column in document lines (Lines view and detail) - long account names no longer overflow the cell: the name truncates within the column and the clear (X) button stays always visible, pinned to the right, so the account can always be cleared.

When using AI processing on invoices, the reclassification of account and competence dates for the selected invoices now restarts from scratch.

Previously, reprocessing could re-propose the classification already present on the selected invoices, because it used them as its own reference history: correcting a batch of misclassified invoices therefore had limited effect. Now the invoices being reprocessed no longer serve as their own reference, while the history of the other invoices still guides the choice. The result: selecting the invoices and running AI processing actually redoes the classification, aligned with the current rules and values.

This is the same behavior already active in AI processing on Movements.

In the Movements list the Account column can now be edited directly from the row: pick the ledger account from the chart of accounts without opening the detail, just like in the documents detail.

Editing from the row applies to Without-invoice movements (the only ones that point to a chart account). A still-missing account is highlighted in amber, and as soon as you set it the count of movements to fix goes down. On With-invoice movements, transfers and balance rows the account stays read-only, because it cannot be assigned from there.

In the Medical dashboard some instrumental and imaging exams were sometimes classified as physiotherapy during import. As these services are often bundled with many visits, the Patients started on physiotherapy figure was inflated (many patients appeared to have started physiotherapy while they had only had an exam).

Import now recognises that an imaging or instrumental exam - X-ray, MRI, CT, ultrasound and echo-Doppler, bone densitometry, mammography, Holter, ECG, electromyography, spirometry - and items bundled with a visit are not physiotherapy, and files them under follow-up visit. The started on physiotherapy count and the physiotherapy sessions therefore reflect only genuine physiotherapy services.

The fix applies to new imports (new analysis and add-data). Services whose category had already been corrected by hand are left unchanged.

The documents Lines view now has an Account column, editable directly in the grid.

Selecting the Account cell lets you pick the ledger account from the chart of accounts without opening the document, exactly as from its detail. After saving, the app may offer to set that account as the supplier default and to apply it to the other documents.

The account is editable on lines of finalized documents (those that already have ledger entries); on draft lines the column shows "Account missing" and becomes editable once the document is finalized. The other columns in the view stay read-only.

The guided tutorials for the Documents and Movements pages have been rebuilt from scratch.

They now follow the real order of the page, top to bottom and along the workflow, leaving no feature uncovered: the filters and their options, the bulk actions (Process/Fill with AI, Change status, Delete, Print), import, the automatic rules and opening an item.

In Documents the tutorial also steps into the document detail (Summary, Schedule, VAT and Ledger tabs, print and finalization). In Movements it covers the whole reconciliation workbench, up to the Confirm step that closes a reconciliation, which was never explained before.

The tutorial can be replayed at any time from the help button at the bottom right.

On a movement's form, linking (or creating on the fly) a subject that has a default accounting account makes the movement adopt that account. This now happens instantly on link.

• Before, the apply was deferred by a fraction of a second: if you saved right away after linking or creating the subject, the save still recorded the previous account and the movement stayed on the wrong one (typical case: an employee's salary account linked to a salary transfer).
• Now the subject's account is set immediately, before you can save; it stays editable by hand.
• If a movement was already linked to the subject before the subject had a default account, just re-select the subject in the Customer/Supplier field to apply it (or set the account on the record and choose to propagate it to existing documents).

A subject's default accounting account - the cost counterpart the movement adopts when you link the subject - could be set only on subjects flagged as Supplier. Anyone who needed to assign an account to an employee (for instance their salary account) couldn't find the field, and setting it by temporarily flagging the subject as a supplier meant the value was discarded on save.

• The Default accounting account field is now available on any subject, with no need to tick Supplier.
• The picker still shows only cost accounts (negative sign): it's meant for the counterpart on outgoing movements, such as salaries.
• Linking the subject to an outgoing movement - from the Customer/Supplier field, including creating the record on the fly from the movement detail - now applies the subject's account automatically to the movement, overwriting any account already there. It stays editable by hand.
• For customers the default revenue account still lives on the item, not on the subject.

When re-processing an already-classified bank charge movement (e.g. multichannel / home banking fee, commissions) with "Process with AI", the fix required two passes: the first set the account (bank charges), the second the counterpart, attributing the movement to the bank's subject.

• Now a single run of "Process with AI" settles both the account and the bank counterpart together.
• The technical reason: the bank-charges account is now defined per individual bank, and the bank-counterpart link only fires once the account is already the right one. On the re-processing path the two steps didn't converge in the same run; now they do.
• Newly imported movements were already correct in one pass: this fix concerns re-processing existing movements.

The multichannel / home banking service fee - e.g. "MULTICHANNEL SERVICE FEE HOME BANKING ... USER No. ..." - was not recognized by any system rule: it fell into the automatic guess, which mistook it for a supplier and created a made-up counterpart ("Multichannel Service"). It is instead always a bank charge.

• A movement whose description contains "multichannel" or "home banking" is now classified as a bank charge "No invoice", with no counterpart suggested.
• The account used is the bank-charges account configured for the movement's bank, as for other bank commissions and account fees.
• The rule applies to outgoing movements.
• The fix covers newly imported movements: movements already classified (and counterparts already created) may need to be corrected by hand.
• The classification can always be edited by hand on the individual movement.

The Documents page now has two views, selectable from the Documents / Lines toggle next to the title.

The new Lines (transposed) view shows, in a single grid, every line of every filtered document, grouped by document. Each row carries the document context - Date, Counterpart, Number - together with the same columns as the detail: description, competence from/to, quantity, unit price, discount, VAT, net, gross and item.

The same filters as the list apply (Issued/Received, search, dates, type, status, causes, accounts, payment). A row's pencil opens the document it belongs to. The default view stays Documents; the choice is remembered across sessions.

The 'ATM' system rule only recognized incoming movements (deposits) as internal transfers, leaving withdrawals uncovered. But an ATM withdrawal is also an internal transfer between own accounts (from the bank to cash on hand), not a cost.

• A withdrawal whose description contains 'ATM' - e.g. "ATM withdrawal ... card operation" - is now classified as "Internal transfer", no longer suggested as a cost or "From invoice".
• The rule now applies in both directions: incoming deposits and outgoing withdrawals.
• It was also applied to existing withdrawals still marked "To classify"; those already classified by hand are left untouched.
• The classification can always be edited by hand on the individual movement.

A series of fixes keeps the balance sheet and its detail views consistent with each other:

• Cell detail and layer filter: clicking a cell with a subset of accounting layers selected now lists only the chosen layers - previously it showed all layers and the total did not match the cell.
• Cell detail and reference date: on the current period the detail stops at the reference date, like the matrix values - previously it also included later entries.
• Per-customer/supplier detail: expanding several accounts in quick succession could leave some rows stuck loading; they now always load.
• Break-even with custom range: while loading (or after an error) the cards show the matrix-range values, with no more incorrect interim combinations. The custom range now also respects the selected layers.
• More accurate warnings: the count of entries to reclassify respects the displayed period; a temporary loading error no longer appears as "No account chart"; truncated lists are now flagged.
• Exports: fixed the Excel index link for charts with an apostrophe in the name; the PDF export validates its parameters more strictly.

The clinical dashboard has been restructured to efficiently handle analyses with tens of thousands of visits, even as they grow over time:

• Instant open: the detail page loads KPIs and reports immediately; per-doctor, per-therapist and per-patient breakdowns load in the background, with a short loading state only on first open.
• Faster operations: adding data, recomputes and reclassifications now write only the actual changes instead of rewriting the whole archive - they stay fast regardless of the analysis size.
• More accurate classification: automatic classification now recognizes instrumental diagnostics (ECG, ultrasounds, X-rays, MRI), medical procedures (infiltrations, dressings) and non-rehabilitative consultations (psychology, nutrition, speech therapy) and no longer counts them as physiotherapy sessions. Physiotherapy conversion KPIs and therapist reports now reflect real values. To apply the new rules to an existing analysis, use "Reclassify with AI" from the Configuration tab.

In a few places of the application labels and messages appeared untranslated: instead of the correct text a raw technical identifier was shown. The most visible case was the "Success" option in the severity filter of the notifications history.

These texts are now shown correctly in the user's language, both in Italian and English.

Opening a clinical dashboard analysis is now much faster: the page shows statistics and reports immediately, while the full list of individual visits is fetched only when needed (opening a "total visits" or "unique patients" drill-down).

Operations that recompute the whole analysis no longer run while you wait on screen - they start as a background job, just like importing a new analysis:

• Recalculate statistics, Reclassify with AI, Rename activity and Add data start the work and return control immediately; a notification signals completion and the statistics refresh on their own.
• On very large analyses these actions no longer fail with "unexpected error, try again".

After an import, invoices are finalized automatically in the background (ledger generation and assisted classification). During periods of high system load this process could stop before completing the batch.

Processing now respects a per-cycle time limit: it completes the documents it can handle in time and automatically resumes with the remaining ones on the next cycle. The outcome is the same, but without interruptions and with steadier progress. No action is required.

In the column filters of the lists (documents, transactions, ledger, subjects and others) it is possible to combine two conditions with AND/OR, for example "does not contain X" and "contains Y". With two combined conditions the list returned an error ("Data loading failed. Retry.") instead of applying the filter.

Combined filters now work in every list: both conditions are applied correctly in AND and OR mode. Single-condition filters were not affected and keep working as before.

A series of technical improvements reduces loading and response times across the application, with no functional changes:

• Clinical Statistics: the analyses list now loads only the data needed by the grid instead of each analysis' full result; opening, filtering and exporting are noticeably faster for organizations with many analyses.
• Reconciliation: match confirmation and the aggregate reconciliation proposal (POS/cards) run their checks in parallel, with less waiting.
• Dimensional balance and break-even: the heaviest ledger reads now run in parallel with the chart of accounts loading, reducing report generation time on wide ranges.
• Reports and monitoring: several detail panels (bank cost composition, cash flow breakdown, cost matrix, consultant activity) fetch their data with fewer database round-trips.
• Pages with complex forms (KPI catalog, fixed assets, profitability): entry forms are loaded only on first open, lightening the initial page load.

The account where bank fees and charges are recorded is now configured on each bank: in Settings, opening a collection resource of type bank account, the new Bank costs account field is available.

Previously there was a single "Bank fees" account for the whole organization, set as an attribute in the chart of accounts. Now each bank has its own, so charges from different banks can be kept separate. Existing banks automatically keep the account used until now.

Good to know: a fee movement on a bank without a costs account set - or on cash, or on a manual import with no bank - stays To classify and must be handled by hand. After the update it is therefore worth checking that each bank has its costs account set. The "Bank costs" attribute is no longer selectable in the chart of accounts.

An internal transfer is a movement of money between own accounts (for example a cash withdrawal or a bank deposit): it has no customer or supplier as counterparty.

• Setting a movement's type to "Internal transfer" removes the subject/counterparty automatically, and the subject column stays empty.
• It applies everywhere: the classification badge in the movements grid, the single-movement form, automatic rules and import.
• The rule was also applied to existing internal transfers: those still carrying a subject were cleaned up.
• Changing the type away from "Internal transfer" lets a subject be set again as usual.

Incoming movements whose description contains 'ATM' (typically cash deposits at an automated teller) are not revenue: they are an internal transfer between own accounts, i.e. a movement of money with no economic impact.

• An incoming movement with 'ATM' in the description is classified as "Internal transfer", no longer suggested as revenue or "From invoice".
• A new deterministic system rule, identical for every organization, applies ahead of the AI suggestion: the classification is immediate and predictable.
• The rule was also applied to existing movements still marked "To classify"; those already classified by hand are left untouched.
• The classification can always be edited by hand on the individual movement.

Until now a rent transfer whose description said only "Locazione" (for example "Locazione Giugno 2026"), without the word "canone" before it, was not recognized as a rent cost. If the beneficiary was a company (Srl, SpA...), the movement ended up suggested as "From invoice" to reconcile, even when the rent is not invoiced.

The automatic classification now also recognizes the word "locazione" on its own:

• A movement whose description says "Locazione ..." (or "Canone di locazione ...") is suggested as a rent cost (Without invoice), ready to be posted to the income statement.
• It applies both on import and when re-processing movements with "Process with AI".
• Bank charges are unchanged: a "Canone c/c" or "Canone tenuta conto" is still treated as a bank fee, not as rent.

Already-imported movements do not change on their own: to apply the recognition to existing ones, re-process them (for example with "Process with AI") or classify them by hand. The automatic suggestion can always be edited.

An e-commerce card payment (a description like "Payment E-Commerce of ... Card ... EUR ... SUPPLIER") used to stay "To classify", because a card is not a bank transfer and was not recognized as an invoice payment.

The automatic classification now suggests it as "From invoice":

• An online card purchase is suggested as "From invoice", ready to be reconciled with the supplier invoice (auto-reconciliation included), instead of staying "To classify".
• It applies both on import and when re-processing movements with "Process with AI".
• The recognition is targeted at the "e-commerce" description only: a generic POS payment or another card debit (which often have no invoice) stay "To classify".
• Only the automatic suggestion changes: the classification can always be edited by hand.

Already-imported movements do not change on their own: to apply it to existing ones, re-process them (for example with "Process with AI") or classify them by hand.

An outgoing transfer with no clear signal used to stay "To classify", leaving it to the user to figure out its nature. But when the beneficiary is a company (the description or counterpart carries a legal form such as Srl, SpA, Snc, Sas...), that transfer usually settles a supplier invoice.

The automatic classification now recognizes this:

• A transfer to a company with no other signal is suggested as "From invoice", ready to be reconciled with the matching invoice (auto-reconciliation included), instead of staying "To classify".
• Transfers to individuals stay "To classify": a director's fee, a salary or a refund to a person is not treated as an invoice payment.
• Cases that do not originate from an invoice are excluded, even toward a company: financing installments, mortgages and loans, shareholder contributions, internal transfers.
• Genuine recurring costs are unchanged: rent, salaries, taxes, withholdings, bank fees and charges, insurance premiums are still recognized when the description says so.
• Only the automatic suggestion changes: the classification can always be edited by hand.

When you change a supplier's default accounting account from its record and save, Numbers now shows a prompt asking whether to apply the account to the supplier's other documents too.

The supplier record is also reachable from a finalized document's detail, via the Subject link in the counterpart card: from there you set the default account and, on save, you choose whether to keep just the default or re-apply it to all the supplier's already-recorded documents, with the same past, future and scope options (all documents, non-reconciled only, only those without an account) as the dialog reachable from a single document.

The prompt only appears when the account actually changes to a real value; saving without propagating updates only the supplier's default.

On ledgers with many rows, opening was slow: before showing anything, the Ledger waited for both the rows and the Totals sum computed over the entire filtered period (tens of thousands of rows), taking several seconds.

Now the rows appear immediately and the Total row at the bottom fills in a moment later, without blocking the view. On a ledger with ~16,000 rows, time-to-data dropped from ~4 seconds to under 1 second.

In addition, when an organization has no fiscal years configured, the Ledger still starts filtered on the current year instead of loading the whole history: previously it scanned the entire ledger and was very slow. The period stays editable by hand.

If a list failed to load (for example a temporary server error), the grid stayed empty with no message, indistinguishable from "no data": this happened above all on the Ledger with many rows, where it looked like nothing was loading.

Now, when a load request fails, an alert appears ("Failed to load data. Try again.") and the grid retries the block. The error is visible and recoverable instead of looking like an empty list. This applies to all lists loaded in blocks: Ledger, Documents, Transactions, Subjects and the other grids.

When an invoice carries a pension-fund contribution line (or an INPS rivalsa / contributo integrativo), that line now automatically inherits the prevalent line's account - the main service, the highest-amount row.

The contribution is an accessory of the service and has no economic nature of its own: making it follow the main service's account keeps it off a separate or unclassified account. The rule is deterministic and applies to both received and issued invoices; stamp-duty (bollo) lines still post to their dedicated "stamp duty" account.

The characteristic revenue accounts - services, product and goods sales, active commissions, active rents - are now posted only from invoices and reached by a bank movement only through reconciliation.

Characteristic revenue is born from a fiscal document: if a collection were automatically classified as standalone revenue, the revenue would be counted twice - once from the collection and once from the invoice it settles. These accounts are now excluded from automatic movement classification (AI suggestions, history, rules) and can no longer be picked by hand on a single movement. On invoice lines they remain fully available.

Other income that normally does not go through an invoice stays classifiable from movements too: grants and subsidies, financial income and tax credits, extraordinary income and gains/losses.

Report figures are only correct when everything is classified: unclassified rows stay out of the totals and skew the numbers, often unnoticed.

Now Ledger, Documents and Transactions show a banner at the top when something in the last 12 months is unclassified, including how many items are involved. The Show button filters the list to those items right away, so they can be sorted out without hunting for them.

What counts as unclassified:

• Ledger: ledger rows with no account assigned.
• Transactions: movements still to classify, or without an invoice but missing an account, for the selected account/cash resource.
• Documents: invoices not yet finalized, or with at least one line missing an account.

The banner disappears on its own once there is nothing left unclassified.

On the Employees page, Create from payslips now accepts multiple files at once:

• Multiple PDFs: you can select several payslips together.
• ZIP archive: you can upload one or more ZIPs containing the payslips, including in subfolders.
• Mixed: loose PDFs and ZIPs together; payslips inside and outside the archive end up in the same list.

Payslips are extracted one by one (with progress) and, when more than one is found, a summary opens to choose which employees to create. Protected files ask for the password one at a time; an unreadable file is reported and skipped without stopping the others.

The bank and financial accounts - bank fees, POS fees, interest expense (loans and leasing), partner repayments, financial income and tax credits - are now posted only from bank movements and never from invoice lines.

These amounts come from the bank statement or a bank credit, not from a fiscal document: if they landed on an invoice line, the cost or income was recorded in the wrong place. These accounts are now excluded from automatic invoice classification (suggestions, history, rules) and can no longer be picked by hand on an invoice line, while they stay fully available on bank movements. This is the same behavior already in place for employee personnel cost accounts.

The Transactions page now opens with the date filter already set from January 1st of the current year (on the transaction date), instead of from the 1st of the current month as before. The behavior is now aligned with the Documents page. No end date is set, so the range stays open up to today.

The filter remains fully editable: you can use the Month/Quarter/Year presets or the ✕ to clear it. Anyone who already had a saved range keeps their choice: the new default only applies on first access.

The employee personnel cost accounts - "Employee Operating Staff", "Administrative Staff", "Severance (TFR)" - are now posted only from bank movements (net salary payments, F24) and never from invoice lines.

Salaries aren't invoiced: if an invoice adjacent to personnel (for example a payroll consultant or an expense reimbursement) ended up on a personnel account, the cost was counted twice - once from the invoice and once from the payment movement. These accounts are now excluded from automatic invoice classification (AI suggestions, history, rules) and from finalization, and can no longer be picked by hand on an invoice line. On bank movements they remain fully available.

Contractors and consultants, partner/director fees and commission-based fees stay classifiable from invoices, because they do issue invoices.

Typing a date was more tedious than needed: you always had to type the full year too.

Now, in date and date-and-time fields, you can omit the year: typing just day and month (e.g. 1503 or 15/03 in the Italian format) fills in the current year and shows 15/03/2026. Two-digit year (150326) and the full date (15032026) still work.

The displayed format and the saved value do not change.

On mobile, date fields used the browser's native picker, which has its own styling and sizes and clashed with the other fields in the console.

Now, as already happens for the date-and-time field, date fields use on mobile too the same component as the desktop: a typable "dd/mm/yyyy" field in the configured format, with a calendar popup. The saved value does not change.

When the application was updated (a new release) while the sign-in page stayed open in the browser, the sign-in attempt could fail with a generic error message, forcing a manual reload.

The page now reloads automatically in this case and shows the notice "The application has been updated. Please sign in again.". The same recovery applies to password sign-in, verification code, email link and password reset.

The console's time and date-and-time fields no longer use the browser's native picker (which had different styling and sizes from the other fields). They are now custom components consistent with the rest:

• The date-and-time field is a single box "dd/mm/yyyy HH:mm" in the configured format. Fill it by typing, with the up/down arrows (they act on day, month, year, hour or minute depending on where the cursor is), or by opening a popup with the calendar and time selection (clickable hours and minutes, plus a field for the exact minute).
• The time field respects the user format (24h HH:mm or 12h hh:mm with AM/PM).
• The saved value does not change.

The New hire comparison on the Employees page has been extended:

• Choice of starting value: besides the gross annual salary (RAL), the comparison can start from the person's target monthly net. The gross/fee is derived for each form at equal take-home pay (net × yearly payments for subordinate forms, net × 12 for the flat-rate VAT freelancer), with the estimated equivalent gross shown. Creating the employee from a form sets the derived gross directly.
• Internship aligned to the net: starting from the monthly net, the internship too follows the target value (an internship's net equals the allowance paid); in gross mode it uses the typical regional minimum. If the allowance becomes salary-like, the app clearly warns it is a disguised employee, reclassifiable as permanent with penalties.
• More complete fixed-term notes: besides the 1.4% NASpI surcharge (the only cost difference versus the permanent contract at equal gross salary), the additional 0.5% due at each renewal is now flagged.

The Employees page has a new New hire button that opens a comparison of the contract forms for a person not yet on board.

• You enter the age and the target compensation (annual gross or fee, hours), tuning benefits, meal vouchers and reimbursements.
• The age filters the available forms: for example the professional apprenticeship appears only between 18 and 29.
• For each form - internship, apprenticeship, flat-rate VAT freelancer, fixed-term, permanent - it shows the annual employer cost, the estimated worker net, efficiency, rules, duration constraints and risks (e.g. the bogus self-employment risk).
• From an available form you can create the employee directly, with the contract and compensation package already set.

Indicative estimates for management purposes: rates, reliefs, internship minimums (regional) and flat-rate scheme rules depend on the collective agreement, territory and current legislation. The values are editable.

The employee card has a new Expense reimbursement (€/month) field for flat itemized reimbursements or travel allowances, with a Taxable checkbox for cases exceeding the exemption or concerning travel within the municipality.

• Employer cost: the reimbursement appears as a dedicated line. Within the exemption (for travel outside the municipality, up to 46.48 euro/day) it is pure cash, with no contributions; if taxable it joins the contribution base like salary.
• Cash flow forecast: the reimbursement is a monthly outflow to the employee and is projected together with net salaries.
• Raise simulator: the flat expense reimbursement is now one of the levers, and among the suggestions - when a raise is simulated - it surfaces the exempt reimbursement that gives the employee the same net while costing the company only the net instead of the full gross. The suggestion reminds that it must correspond to real expenses, not a salary substitute.

Indicative estimates for management purposes: actual rates, thresholds and exemptions depend on the collective agreement, personal situation and current legislation.

The dashboard cash flow forecast now accounts for personnel cost. In addition to receivable and payable invoice due dates, it projects forward the recurring outflows estimated from the employee registry:

• the net salaries on each month's payday (with the 13th month in December and the 14th in June for those with 14 payments);
• the F24 (employee and employer contributions + income tax) around the 16th of the month.

TFR is excluded (not a cash outflow until termination). The Cash flow card has an Include personnel cost toggle to switch these items on or off, and the summary shows how much of the expected outflow is personnel.

No reconciliation with bank movements is needed: it is an estimate derived from the registry; already-paid salaries stay in the current balance and are not double counted.

Three additions to the Employees page:

• Hours, part-time and hourly cost: you can enter the weekly hours (and the full-time reference). The form recognizes part-time employees (with the corresponding percentage) and computes the employer hourly cost, shown in the grid next to the employer cost.
• Password-protected payslip: if the payslip PDF is protected, the password is requested to read it and extract the data.
• Multi-payslip PDF: if a single PDF contains the payslips of several employees, a summary lists all detected employees and you can create them all at once (optionally deselecting the unwanted ones).

In the employee form the Dependent children field is now a number instead of a simple yes/no checkbox. The value is also extracted as a number from the payslip (e.g. "FIGLI A CARICO 2").

For the fringe benefit exemption threshold (which by law doubles from 1,000 to 2,000 euro when there is at least one dependent child) what matters is whether there are children; the count is still recorded to reflect the employee's actual situation and is used in the optimization analysis.

Several refinements to the Employees page:

• Per-employee local surtax rate: regional and municipal surtaxes depend on residence; there is now a dedicated field (with an indicative default) to enter the actual municipal rate, for a more accurate estimated net.
• Monthly, "smart" optimization analysis: it can be regenerated at most once a month and only if the employee's data has changed since the last analysis; with identical data the button stays disabled.
• Safer payslip upload when editing: uploading a payslip for an existing employee no longer automatically overwrites already-entered values (gross salary, rates).

On the Employees page each row now has a star-icon button that opens an AI-generated compensation optimization analysis.

• It proposes legal, personalized strategies for the individual employee: fringe benefits within the threshold (doubled with dependent children), meal vouchers, corporate welfare, supplementary pension, tax-relieved result bonuses and other levers.
• It starts from the employee's real data (gross salary, contract, current benefits, dependent children, estimated employer cost) and indicates the expected benefit for the business and the employee.
• The analysis can be regenerated once a week per employee; the latest version stays available in the meantime.

Suggestions are indicative and should be validated with a labor consultant before applying them.

When the payslip indicates dependent children, the extraction now automatically checks the Dependent children option in the employee form. The option raises the fringe benefit exemption threshold used in the employer-cost calculation and the simulations.

Improved automatic data extraction from the payslip in the employee form:

• More reliable extraction: some payslips were not read, or the gross annual salary (RAL) was missing. This is now fixed: values are extracted in full, including the RAL.
• Employer-cost data: when the payslip includes the employer-cost section, the employer contribution rate, the INAIL rate and the annual fringe benefit / meal voucher amount are now filled in automatically.

On the Employees page the New employee button is now a menu with two paths:

• Create manually: opens the empty form, as before;
• Create from payslip: immediately asks for the payslip PDF, runs the automatic extraction and opens the form pre-filled with the extracted values (period, gross, net, estimated annual salary, collective agreement, level, hire date).

Uploading a payslip remains available from the dedicated section of the form, both when creating and when editing.

New Employees page in the operational menu, dedicated to personnel costs:

• Payroll registry: for each employee or contractor you record contract, collective agreement, level, yearly payments, gross salary and contribution rates.
• Fill-in from the payslip: uploading the payslip PDF automatically extracts the values (period, gross, net, contributions, estimated annual salary) and pre-fills the form. The file is not stored; the extraction history remains available.
• Employer cost: for each employee the grid shows the actual annual cost for the business (gross salary + contributions + INAIL + severance + benefits and other costs) together with the estimated monthly net.
• Raise and benefit simulator: for each employee you can simulate salary raises, fringe benefits, meal vouchers and welfare, comparing the additional employer cost with the additional net received and the efficiency of each lever.
• Optimization suggestions: the simulator highlights more efficient alternatives (for instance an exempt fringe benefit replicating the net value of a raise at a lower cost) and exemption threshold overruns.

Employee-side estimates are indicative and managerial: they are not a substitute for payroll processing.

Several indicators in the application refresh at regular intervals: the running-imports badge, the notifications bell, the trash count and the alerts page.

These periodic updates now stop automatically when the browser tab is in the background and resume - with an immediate refresh - as soon as you return to the tab. The result is less network traffic and lower battery usage when the application is not in the foreground, with no perceptible difference while using it: on return the data is up to date right away.

We reduced the number of database requests on several frequent operations, which now respond faster:

• Balance sheet: the three data sets that make up the page are now loaded in parallel instead of sequentially.
• Reconciliation: suggestions for a movement or a due date compute the already-allocated share in parallel, cutting the wait.
• Movements: creating and editing a movement avoid a redundant read right after the write.
• Payments: recording a payment on a due date reuses already-read data instead of re-reading it.

All figures and behaviors remain identical: only the response speed changes.

Several pages of the application now open faster.

The heaviest components - data grids and charts - are now downloaded by the browser only when actually needed, instead of being part of the initial page load. Among others, the Reports, Dashboard, Cashflow, Profitability and change Log pages and the document detail view benefit. On password setup pages, the strength indicator now prepares in the background without slowing the page down.

Automatic subject reconciliation after a document import is also faster: updates are applied in bulk rather than one at a time. Results are identical, only the speed changes.

Fixed a case where some movements downloaded from the bank were not processed with AI (subject, account, accrual period, classification, collection channel) and stayed "To classify" until processing was started manually with Process with AI.

It happened mostly on the first connection, when the download covers a long period: if the download operation stopped before finishing, the already-imported movements were not queued for AI and on later downloads were recognized as already present, so they were never picked up again.

AI processing is now queued as the movements arrive, so nothing is lost even if the download is long or interrupted. No action is required.

The indicators at the top of the Cash flow page have been corrected and optimized.

• More accurate current balance: the calculation now accounts for the full movement history of the selected resources. With many movements the figure could be understated because only a subset of the data was read.
• Corrected DSO (days sales outstanding): the indicator now uses only the current organization's invoices. Previously it could include unrelated data, skewing the displayed value.
• Faster loading: the indicators are now computed in parallel instead of sequentially, reducing the page's load time.

No action required: the values update automatically the next time the page is opened.

The Balance sheet page now loads faster.

The charts of accounts - the base one and all active reclassifications - were loaded sequentially, one after another: with several reclassifications configured the wait added up. They are now loaded in parallel, noticeably reducing the page's opening time.

The balance sheet figures and all reclassification tabs are identical to before: only the speed changes.

Improvements to the Aggregate reconciliation dialog (POS credits, online payouts and card statements):

• Date from the description: when the movement description contains a date (e.g. "INCASSO POS ... DEL 22/05/26 ..."), that is the real collection day and is preselected as the Collection date, which drives the due-date window. It stays editable.
• Perfect matches: when the movement balances in more than one way, a picker next to Balancing lists the combinations of schedules that match exactly; choosing one applies that selection right away.
• Highlighted rows: selected schedules are now highlighted with a tinted background and a colored border, not just by the checkbox.

In date fields (on desktop) the keyboard up and down arrows let you adjust the value quickly without retyping it.

• Cursor on a field: with the cursor on the day, month, or year, up/down increases or decreases only that field. The value rolls within its own range without moving the others (e.g. 31 January + 1 day rolls back to 1 January).
• Whole value selected: when the entire field is selected, up/down shifts the whole date by one day, carrying across months and years as usual.
• Empty field: the first press inserts today's date.

The arrows respect any minimum and maximum limits set on the field.

The Bank costs report has a new Value-date cost section.

The bank sets a movement's value date on a different day than the operation date: that gap is a financial cost (the money is not available, or is debited earlier).

• You enter an annual rate and the section shows the estimated cost over the period, computed from the movements' value days: for credits the days between operation and value, for debits between value and operation.
• The calculation uses all movements in the period; those without a value date do not contribute.
• The section respects the report's period filter.

The value date is entered by hand on the movement or comes from bank imports and statements.

The Bank costs report has a new Collection costs by channel section: a summary of the fees retained by the collection channels (POS, online circuits) over the selected period, with each channel's share.

• Costs populate when you reconcile an aggregate collection applying the channel's commission: the recorded commission is attributed to the channel.
• The section respects the report's period filter.
• Collection channels are configured in Settings.

The "Bank cost comparison" panel (the savings estimate against an alternative bank offer) has been removed.

In the reconciler (opened from Transactions and Schedule) the Flip button at the center of the header becomes a round, icon-only control with the two-arrow icon and no label. It works exactly as before - it swaps the two columns while keeping the selected item on the left - and the explanatory tooltip still appears on hover.

Movements can now record the value date, i.e. the day the bank actually makes the amount available (or debits it), distinct from the operation date.

• Manual entry: the movement popup offers an optional Value date field, next to the operation date.
• Automatic import: the value date is read from imported bank statements and from bank-connected accounts, when available.
• In the grid: the Value date column is available in the movements list (hidden by default, enabled from the column options) and is included in the export.

The field is optional: when left empty it stays blank and does not affect totals, reconciliation, or competence.

On issued invoices, linking a line to an item already in the catalog keeps working for every organization: text recognition and the AI semantic match are unchanged.

What changes is the creation of a brand-new item from the AI suggestion, when no catalog item matches. It now happens only when both conditions hold:

1. the organization is a medical center;
2. the line is a healthcare service (visits, sessions, therapies, exams, clinical consultations and the like).

For other organizations, or for lines that are not healthcare services, the line still links to a catalog item when one exists, but no item is created automatically.

The Configuration → Automatic rules page is now organized into three tabs:

• Rules - the usual automatic rules, unchanged.
• Suppliers - the list of suppliers with their counterpart (a cost account).
• Items - the list of items with their counterpart (a revenue account).

In the Suppliers and Items tabs you can assign the default counterpart quickly:

• Inline: pick the account directly in the Counterpart column.
• In bulk: select several rows and apply the same account to all with Assign counterpart.
• Full record: open the supplier's or item's record with all fields.

Assigning from the grid sets only the default account on the record: already-recorded invoices, transactions and entries are not modified.

On bank-connected accounts, the movements list now shows an opening balance row at the start of each month, before the first movement.

• Real balance: the value is anchored to the actual account balance provided by the bank and reconstructed month by month, so it matches the bank statement.
• Updates on its own: the balance rows are recomputed at every bank synchronization, with no manual action.
• Does not pollute totals: the balance rows stay markers and are not counted in totals, reports, or the movements export.

Note: the bank only exposes the current account balance, so past months' balances are reconstructed backward from the movements; any movements not yet visible at the bank may slightly shift the oldest balances.

Progressive-loading lists (subjects, invoices, transactions, ledger, schedule and similar) now show a Refresh button in the toolbar.

• Pressing it reloads data from the server, handy to immediately see rows changed by another user or by a running process, without reloading the whole page.
• If you scrolled down the list, the refresh keeps your scroll position and current page: you are not sent back to the top.
• The button is available both on desktop (next to "Columns") and on mobile (next to "Import / Export").

The Dashboard button in the transactions toolbar and the charts/KPI view it opened (income/expense summary, balance trend, top accounts) have been removed.

• The transactions page now shows the grid directly, without the grid/dashboard toggle.
• Statistics and indicators remain available in the dedicated sections of the app.

A movement's classification answers the question "does it originate from an invoice?", which can be decided right away: reconciliation - linking it to the specific invoice - is a later, optional step.

• Movements that originate from an invoice are now marked With invoice already at classification time, before reconciliation: outgoing transfers, incoming takings (transfers and credits), POS/Stripe takings, invoice references in the description (including the "Invoice N dated ..." format), amounts that match an open schedule, and aggregated takings (POS/card).
• To classify is left only for the genuine unknown. Incoming credits that do not come from an invoice (refunds, tax refunds, interest) are not marked With invoice.
• These movements stay reconcilable: on the reconciliation page the Category filter shows both With invoice and To classify by default, i.e. all the movements to match; the link-status filter hides the ones already reconciled.
• No effect on the accounts: a With-invoice movement does not generate any economic ledger entry, exactly as before (the economics come from the invoice, never from the movement).

Several lists in the application show data in a table with many columns on a large screen. On phones the same lists become cards, and in various cases the cards only showed part of the fields: data present on desktop (amounts, statuses, classifications, tax codes, accounts, KPIs and more) did not appear on mobile.

What changes:

• The mobile cards of the lists now mirror the columns shown on desktop, keeping the same badges, colors and actions.
• Aligned lists include: transactions, documents/invoices, registry, ledger, schedule, fixed assets, work orders, profitability, the medical dashboard, plus several administration panels.
• Actions already available on the cards are unchanged; the missing informational fields were added.

When a bank account is connected for the first time via Open Banking, the first sync now retrieves the widest history window the bank allows, up to about 24 months of transactions. Previously, for some institutions, the first import was limited to the last 3 months.

If the bank retains a shorter history, only the available transactions are imported. Subsequent automatic syncs are unchanged: they keep updating the most recent transactions.

Automatic synchronization of connected banks has been restored.

• Downloads on its own again: the periodic check imports new movements from connected banks again, without having to start the sync manually every time.
• AI processing on downloaded movements: movements just imported from the bank are now automatically processed with AI (subject, account, accrual period, classification, collection channel), the same result as the Process with AI button. Processing is more robust: if interrupted it resumes on its own and completes the movements left behind.

No action is required: already connected banks resume syncing and new movements get classified.

Two improvements to the invoice detail and report grids.

Account editable inline:

• In the Lines tab of a finalized invoice, the Ledger column shows the account of each line and you can now change it directly in the cell: one click opens the account picker, you choose, and the change is saved, without opening the ledger entry edit dialog.
• Lines without an account stay highlighted until one is assigned.
• It also works on mobile: in the line cards the account is shown and editable, alongside the same fields as desktop (competence, item). The account picker is wider and shows full names.
• Advanced editing of the entry (competence, subject, saving the account as a default to propagate to other documents) is still available from the Ledger tab and the Ledger page.

Consistent date filter:

• Some date columns (for example Competence From/To in the invoice lines and the first/last session dates in reports) used a plain text search box instead of the standard calendar.
• All these columns now use the same calendar filter as the other dates, with range operators (greater than or equal, less than or equal, between two dates).

Automatic detection of the collection channel on incoming movements is now more reliable.

• If one of the channel's keywords (e.g. "INCASSO POS") appears in the movement description, the channel is tagged automatically just by processing with AI, instead of being left empty.
• A channel already set by hand is not overwritten when re-processing.

When saving a company registry record, the VAT number was verified through the European VIES service and, when VIES did not confirm it, the field was shown as invalid (red). However VIES also returns "invalid" in entirely legitimate cases: valid Italian VAT numbers not registered for intra-EU operations, or when the service is temporarily unavailable. This made the field feel like a blocker.

What changes:

• Formal validity of an Italian VAT number now relies on the local checksum, which is deterministic and reliable.
• The VIES result is only informational: if VIES confirms the VAT number the field is green ("verified via VIES"); if it does not confirm it, or the service is unavailable, the field is amber and indicates it can still be saved.
• The red state appears only for a genuine formal error (failed checksum, e.g. a typo).
• Saving the registry record is never blocked by the VIES check result.

Refinements to the transaction popup.

• Movement type and Collection channel now sit side by side, with their labels aligned at the same height.
• The "Extend to similar movements" button is next to the Movement type label.
• The "Aggregate movement" field is no longer set manually from the popup: it stays auto-detected (and in the grid the "Aggregate" column shows the channel, or the POS/Card type when there is no channel).
• Fix: the collection channel now stays selected when reopening an already-saved movement (previously the dropdown reopened empty even though it had been saved).

Improved handling of the collection channel on transactions.

In the transaction popup:

• The Collection channel field now sits next to the Movement type and is an always-available dropdown.
• It appears only for invoice-collecting movements (Movement type "To classify" or "Invoice settlement"); for standalone movements and internal transfers it is hidden (there is no invoice there).
• The "+ Add channel..." item and the edit button open the channel popup right here: you create/fix a channel without going to Settings, and the new one is selected automatically.

In the Transactions grid: there is now a single "Aggregate" column showing the channel when set (e.g. the network name), otherwise the POS / Card badge as before - so the information is not lost even on aggregate movements without a channel.

When importing movements from a file or statement, the Channel column for incoming credits is now filled in right away: the collection channel is recognized from the description by matching it against the keywords of the channels configured under Settings > Fiscal > Collection Channels.

• Immediate recognition at import time, with no need to run AI processing.
• Incoming movements only; if no channel matches, the column stays empty.
• A manually set channel is not overwritten by a new import.

The Transactions grid has a new Channel column: for incoming credits, AI guesses which collection channel the movement came from, choosing among the channels configured under Settings > Fiscal > Collection Channels.

• Guessed by AI: while processing movements (import, bank sync, "Process with AI"), the channel is recognized from the description among the configured ones. Incoming movements only.
• Set manually: from the transaction popup you can choose or correct the channel ("Collection channel" field, shown on incoming credits).
• Filterable and sortable: the Channel column filters by name and sorts like the others.

If no channels are configured, the column stays empty: add them under Settings > Fiscal > Collection Channels.

A newly registered user had every already-published release note flagged as "unread": on first login the What's new panel opened with the entire history, which is not useful when getting started.

Now notes published before the user joined are treated as already read: the What's new panel starts with no backlog and only reappears when a note is published after the user joined.

No impact for existing users on notes released after they joined: those stay unread until opened.

In the Aggregate column filter on the Transactions page, the two options were both labeled "Aggregate movement" and therefore indistinguishable. They now show the specific names - POS collection and Card statement - so you can filter by the desired type.

Aggregate reconciliation now handles online payment network settlements, which arrive in the bank net of fees.

• Channel recognition: if the settlement comes from a configured channel (Settings > Fiscal > Collection Channels), the popup recognizes it from the description and shows it in the header.
• Gross proposal: knowing the channel's commission, the popup proposes invoices at gross (the settlement is net) and shows the commission = selected gross - net credited.
• Balancing and cost: on confirm, the selected invoices are closed at gross and the commission is recorded as a cost on the account configured for the channel, so the income statement correctly reflects the network's charge.

To use it, first configure the channel (name, keywords, commission %, cost account) under Settings > Fiscal > Collection Channels.

Work-order auto-assignment during invoice finalization has been made more robust.

• When many invoices are finalized at the same time (typical of a bulk import), the new work-order number is now computed and assigned atomically.
• This removes numbering conflicts between documents processed in parallel: every invoice without a work order gets one automatically, with none left behind.
• The fix only affects automatic work-order creation: the visible numbering and existing work orders are unchanged.

Background imports (bank statements, invoices) now stop on their own when they get stuck.

• If an import stops progressing for several consecutive attempts, the system halts it instead of retrying forever.
• The halted import is marked as failed and a notification is sent, so it is noticed even with the tab closed and can be retried.
• Previously, in rare cases, an import could stay stuck indefinitely without completing or reporting anything.

Nothing changes for imports that complete successfully: this only affects the cases that used to hang.

When entering a new organization, the first login used to show the initial setup wizard, the page's guided tour and the What's new panel all at the same time, overlapping and hard to follow.

The three are now coordinated and appear one at a time:

• on an organization that still needs setup you only see the setup wizard; the tour and What's new show up on later logins;
• once setup is done, when a tour is available the tour runs first and the What's new panel opens afterwards.

All of these remain available on demand from the help button at the bottom right.

The Code field for collection resources (bank accounts, cash, cards, online platforms) is now optional.

• When adding a bank or another resource the code can be left blank.
• If the field is empty, the system assigns a unique code automatically (e.g. BANK-1, CASH-1), so the resource stays identifiable in lists and searches.
• Anyone who prefers to keep using their own code can still type it: in that case an already used code is flagged as a duplicate, as before.

No existing resource changes: this only affects how the form is filled in.

On Balance sheets you can now drill down to the individual customer or supplier without leaving the matrix.

Every leaf account that has entries tied to a subject shows an extra small chevron: expand it and the account opens into one row per customer or supplier (each with its icon), plus a "No counterparty" row for entries without a subject. Each row carries the per-period amounts and a total, and the subject rows always sum to the account value: it's the customer/supplier subledger read straight inside the balance sheet.

Clicking a cell on a subject row opens the usual ledger drill-down, but filtered to that single name. The detail loads on first expand and the chevron only appears on accounts that actually have linked subjects. The feature is available on the base chart.

Small refinements to the transactions grid:

• the aggregate movement column is now simply called Aggregate and can be widened freely by dragging its border (its width used to be capped);
• the Attachments column is hidden by default: it stays available and can be turned back on when needed from the column selector (the grid's Columns menu), just like the competence dates.

No data changes: this only affects how the columns are displayed.

When importing transactions from a file, automatic recognition reads the counterpart's name from the description (e.g. a wire-transfer beneficiary, A fav: ROSSI SRL). Until now this name was discarded when it didn't match an existing registry subject: the Transactions grid showed an empty counterpart, and recovering it required selecting the rows and running "Process with AI" by hand.

The import now persists the name as a free-text counterpart, matching the behavior of "Process with AI":

• Rows linked to a registry subject keep showing the subject.
• Rows with no match now immediately show the AI-read name as free text, instead of staying empty.
• No registry entry is created automatically, and an existing name (e.g. from the file's counterpart column) is never overwritten.

Everything else is unchanged: collections from POS or payment channels still wait for reconciliation against invoices, and your own company is never set as the counterpart.

Automatic recognition of bank transfer charges relies on a list of typical descriptions. The variant "costo per bonifico" (with "per" in the middle) was not covered: although it is a genuine bank charge, the movement stayed "To classify" with no account, because the word "bonifico" made it look like a possible invoice payment to reconcile.

The wording is now among the recognized ones:

• An outgoing movement described as "costo per bonifico" (or "costi per bonifico") is classified as Bank Charges, not from invoice, and linked to the bank as subject, as already happened for "spese bonifico", "spese per bonifico" and "costo bonifico".
• Only the automatic suggestion at import time changes: the classification can always be edited by hand.
• Existing movements are not changed automatically: they remain editable as before.

On issued invoices, the item of each line is now recognized automatically, in a cascade:

1. first it looks among existing catalog items for a name that appears in the line text;
2. if that is not enough, a dedicated AI recognizes the service or item;
3. if the identified item is not in the catalog yet, it is created (once, with no duplicates) and linked to the line.

Once the item is linked, at finalization the line account (Registro column) is taken from the item's default revenue counterpart. Stamp duty lines stay excluded: they create no items and keep posting to the stamp-duty account.

AI inference of the competence period now first applies a deterministic reading of the description and falls back to the model only when the period cannot be derived with certainty.

Periods written plainly in the description are always recognised, the same way already used during import:

• month + year: "for the month of May 2026", "January 2026";
• quarter / half-year: "Q1 2026", "first quarter 2026", "second half 2025";
• numeric month/year: "05/2026", "2026-05";
• explicit ranges: "from 01/05/2026 to 31/05/2026";
• year only: "year 2026".

Previously this step relied entirely on the model's extraction: a line like "Services performed in the month of May 2026" could fail to be recognised as May. Now a certain period is resolved directly, with a stable and repeatable result, also saving model calls on entries that are already clear.

A new Collection Channels registry is available (Settings > Fiscal).

• For each channel (e.g. online payment networks) you configure the keywords that identify it in the movement description, the commission (percentage) withheld by the network, and the account where it is recorded as a cost.
• It is the first step toward automatic balancing of online payouts: network settlements arrive net of fees, and knowing the channel's commission will let the reconciliation popup propose the right invoices and record the commission as a cost. The popup integration ships in upcoming releases.

When looking for matches between bank transactions and schedules, a different counterpart now weighs against the candidate instead of being simply ignored.

• Different counterpart = penalty: if a transaction and a schedule have different assigned subjects, or both carry a counterpart name but a clearly different one, the score drops below the noise floor. The candidate is no longer proposed as the best match. Previously a transaction with the same amount and date but the wrong party could outrank the correct one.
• Uncertainty stays neutral: if the counterpart is unknown (unidentified transaction, missing name), there is no penalty: it could still be the right one, so it stays among the candidates without being penalised.
• Name wins over subject: if the names match but the assigned subject differs, no penalty applies (the subject may have been misclassified).
• Explicit label: in the rare cases where the candidate stays visible (e.g. the invoice number is cited in the transaction) a "Different counterpart" chip is shown.

On organizations with many due dates in the selected period (typical of online takings settled periodically, which open a wide window), the aggregate reconciliation popup could get stuck on "Loading...": the automatic search for the combination of due dates that balances the movement could become too expensive.

That search is now bounded: the popup always opens and, when no exact combination exists (as for a settlement net of fees), it still proposes the best available coverage, to refine by hand or with "Re-propose".

The aggregate reconciliation popup now handles online takings better (electronic payments settled periodically that bundle several days of sales).

• Automatic trailing period: for these settlements the Online takings type is proposed automatically, with the due-date period set from the days before the settlement date (no longer the settlement day alone, which came up empty because sales accrue earlier). Type and period remain editable by hand.
• Re-propose the combination: after manually changing the selected due dates, the Re-propose button automatically re-selects the combination that balances the movement.
• More robust loading: if loading the due dates fails, the popup now shows a message with a Retry button instead of getting stuck on loading.

During reconciliation, in some cases (for example linking a due date to a movement with the opposite direction) an unreadable technical message with internal identifiers appeared.

• Reconciliation errors are now clear and localized in Italian and English.
• Direction mismatch, over-allocation, item not found and similar cases have a dedicated message.
• Any unexpected error now shows a generic notice ("Operation failed. Please retry"): raw technical text will never appear again.

In the Movements grid, cumulative takings (POS and card) are now easier to spot.

• The column badge now reads Aggregate movement, no longer "POS" / "Card".
• An aggregate movement cannot have a customer/supplier: the field is disabled both in the movement popup and in the grid, and any pre-existing subject is removed.
• Reconciliation is unchanged: the clickable POS / Card drill-down badge stays in the reconciler.

The aggregate reconciliation popup (POS takings, card statements, online takings) has been refined to identify the movement at a glance and to work on long lists.

• Description in the header: the header now shows the bank movement description, next to date and amount, so it is immediately clear which movement the reconciliation refers to.
• Grid-only scrolling: the filters (period and type) at the top and the totals with the buttons at the bottom stay fixed; only the list of due dates scrolls, even when there are many rows.

When you reopen an imported invoice to edit it, it now stays open until you finalize it again manually.

Previously, imported invoices were finalized automatically, and this applied to a just-reopened invoice too, which could therefore be closed again after a short while. The result was a confusing error: you edited a line and saw "finalized invoice" while the status still showed Open. Reopening is now recognized as a deliberate choice and is no longer undone automatically.

In addition, if an invoice gets finalized while you are viewing its detail (for example from another workstation), the screen now refreshes on its own to show the correct status and prompts you to reopen it to make changes, instead of returning an unclear error.

The help button at the bottom-right of every page has been redesigned: instead of the round ? icon there is now a single round button with a chat icon, coloured with your organization's colour.

Clicking it opens the menu with the available functions - My tickets (the assistance panel with your tickets), AI Assistant, Guides & FAQ, Tutorial and What's new - depending on the features enabled for the organization and personal preferences. The goal is a single, recognizable entry point for assistance and help.

The movement allocation - which let a single movement be replaced by multiple detail items - has been removed.

The Allocation column (Allocate button) in the Transactions grid, the matching button in the movement edit popup, and the dedicated item in the reconciler row menu are gone. Movements split in the past are left unchanged.

On the Documents page, the grid Date column now has its own header filter: documents can be shown by a date equal to, before, after, or within a range, without touching the other filters.

The column filter works inside the Da / A range set at the top: the calendars do not allow picking dates outside that range when it is set, so the selection always stays consistent with the time window already chosen.

On the Documents page, the Type column filter now distinguishes only between Invoice and Credit note.

It previously combined type and direction (Issued Invoice, Received Invoice, Issued Credit note, Received Credit note). Now that the Issued / Received direction has its own dedicated switch in the page header, that distinction in the filter was redundant: the two choices are now separate and clearer - set the direction at the top, pick the document type from the column.

The Transactions page opens with the date filter preset from the 1st of the current month. If there are no transactions in that period, the list appeared empty with no explanation, suggesting there was no data at all.

Now, in that case, the grid shows an explicit message: No transactions in the selected period. Widen the date range or clear the filters to display transactions. The Month/Quarter/Year presets or the ✕ on the date filter bring the transactions back into view right away. If instead the selected account has no transactions at all, the message says so clearly. The notice appears only once loading has finished, to avoid confusion while data is still loading.

A movement's classification has only four states: To classify, Without invoice, With invoice, Internal transfer.

• In the Movements grid a POS taking (or cumulative payment) not yet matched shows "To classify" again, no longer the "To reconcile" badge. Once you reconcile it against a schedule it becomes "With invoice".
• Reconciling cumulative POS/card takings stays where it was: the clickable drill-down badge is in the reconciler, not in the grid.

On wire-transfer descriptions the company's own name (the account holder) often appears next to the counterpart's name. Automatic subject recognition could therefore set the company itself as the transaction's customer or supplier, instead of the real counterpart.

Two improvements:

• Automatic recognition now receives the company name as the account holder and skips it: it extracts the real beneficiary or originator (e.g. from COMPANY LTD ... ROSSI MARIO V/ORDINE E CONTO ... it recognizes Rossi Mario).
• The company itself is never linked as a transaction's customer/supplier: the registry subject that corresponds to the organization - by VAT number or fiscal code, or by a similar business name - is excluded from automatic matching across every import channel (file, bank sync, AI processing).

Transactions already linked by mistake are not changed automatically: fix them by re-running "Process with AI" from the Transactions grid, which now removes the wrong link.

When you delete invoices and re-import the same archive (XML or file), the documents are now re-created correctly.

Previously the duplicate check kept recognizing the deleted invoices as already present: the import ended with a completion notice but without importing anything, because every row was counted as a duplicate. The check now ignores deleted invoices, so re-importing after a deletion succeeds as expected.

Finalized invoices stay protected: they cannot be deleted until they are reopened to draft, so this case does not affect them.

Some bank movements do not match a single invoice but bundle many: a POS collection (a day's sales credited together, to different customers) or a credit-card statement (a month's purchases debited together). For these, the name shown in the description is the payment channel provider or the merchant (often your own organization), not a customer.

From this release, aggregate movements carry a dedicated flag (POS collection / Card statement):

• it is recognized automatically during processing (file import and bank synchronization) and can be set or cleared by hand from the movement popup;
• a flagged movement cannot have a customer/supplier: the field stays empty and no record is created from the name in the description;
• the flag is visible in the movements grid, with a filter to isolate them;
• in the reconciler, these movements show a more prominent button to open the aggregate reconciliation (one movement → several invoices), and the other column no longer proposes one-to-one matches.

Existing aggregate movements have been updated: where a POS collection had mistakenly taken the merchant name as a customer, the field has been cleared.

Reconciling aggregate movements (POS takings, card statements, online takings) is now fully driven from the popup header.

• Takings date: the date the movement refers to, proposed automatically and editable.
• Type: POS, Credit card or Online takings. The type sets the due-date period - the given day for POS, the whole previous month for card and online takings - and the two period endpoints remain editable by hand.
• Open invoice from the grid: each due date in the list opens on the fly, without leaving reconciliation.
• Optimal preselection: on opening, the due dates that cover the movement amount are already ticked.
• Cash day-close (POS): the day's due dates not covered by the credit can be marked as cash and recorded in the cash register. It is not automatic: if there is a second POS, exclude the relevant due dates before confirming.

The Transactions page now opens with the date filter already set from the 1st of the current month (on the transaction date), while the Documents page opens from January 1st of the current year. In both cases no end date is set, so the range stays open up to today.

The goal is to show the most relevant period right away, without loading the entire history on every visit. The filter remains fully editable: on Transactions you can use the Month/Quarter/Year presets or the ✕ to clear it, and on Documents you can change or empty the date fields.

An assistance panel is now available directly inside Numbers.

• An always-reachable button lets you open new support requests and review your own tickets and those of your company.
• Access is automatic: anyone already signed in to Numbers reaches the assistance area without a second login.

In the reconciler (opened from Transactions and Schedule) the Flip button moves to the center of the header, above the divider between the two columns (schedules and movements) that it swaps. Hovering shows a tooltip explaining its effect: it swaps the columns while keeping the selected item on the left.

Reconciliation aligned with the real movement model (internal transfers, not from invoice, linked to invoice):

• Reconciler: the category filter now starts from "To classify", i.e. invoice payments not yet matched (POS takings, transfers) - exactly the things to reconcile. It previously started from "Not from invoice" (fees, taxes), which hid the very POS takings you want to reconcile.
• Movements grid: a POS taking (or cumulative payment) not yet matched now shows "To reconcile" in blue, instead of the generic "To classify": it is a movement generated from invoices, linked to an invoice but still to be reconciled. Once reconciled it becomes "With invoice".
• Clearer classification names: Invoice settlement is now "With invoice" and Internal transfer keeps its name (the three types are: Without invoice, With invoice, Internal transfer).

The Process with AI bulk action on the Transactions grid now adapts to the size of the selection:

• Up to 20 transactions: processing stays immediate and inline, with direct feedback on completion, as before.
• More than 20 transactions (including "Select all"): processing is queued in the background. You can keep working or change page; progress and the completion notice show up in the activity indicator at the bottom of the sidebar. The grid refreshes on its own once the job finishes.

Previously, large selections ran entirely in the foreground: the screen stayed blocked for a long time and, on very large selections, the operation could time out. Moving it to the background reuses the same flow already used by movement import and by the bulk Process with AI on invoices. The outcome (counterparty, account, competence period, classification) is identical to the immediate path.

The Process with AI bulk action on the Transactions grid now treats the customer like the other fields: it re-determines it on every selected transaction, instead of only filling the ones without a customer.

• If processing finds a better-matching subject (from history, a rule or an exact match in your contacts), it replaces the current customer - even a wrong one.
• If it finds nothing reliable, it clears the transaction's customer.
• Bank charges still receive the bank's subject.

Previously, transactions that already had a customer were skipped: a customer picked by mistake was neither corrected nor removed. Now "Process with AI" can correct or remove it.

Worth knowing: if a transaction has a correct customer but processing has no way to recognise it (unique description, no history or rule confirming it), the customer is cleared. Use "Process with AI" on transactions when you want customers cleaned up or recomputed, not to preserve isolated manual assignments.

The direction switch on the Documents page has been moved next to the page title, on the same row, to free up space and make it more immediate.

It now offers two options - Issued and Received - instead of the previous three: the All option has been removed. On open, the page shows issued documents; one click switches to received ones. The switch stays visible at all times and combines with the other filters (type, status, dates, payment status).

The reconciler (opened from Transactions and Schedule) becomes more movement-centric:

• Per-column filters: filters sit above the column they act on and follow the swap (the Flip button swaps the columns and their filters move with them). On movements: Category (default Not from invoice, i.e. not yet linked to an invoice; multi-select to include the others), Link (Linked / Unlinked, default Unlinked) and More (direction, counterpart, period, excluded). On documents (schedules): Hide paid and period.
• Cumulative takings: movements recognized as POS takings, card statements or a payment processor payout show a clickable badge in the reconciler (not in the Movements grid). It opens a near-full-screen drill-down that proposes the previous day's invoices, balances them and, for POS, records any cash difference in the cash account. On card statements you can also close without invoice (the purchase may not have one), reversibly.
• Movements on the left: the board opens with movements in the left column, the natural starting point of the work.

Two refinements to selection in the full-screen reconciler:

• Deselect with one click: clicking the already-selected schedule (or movement) again deselects it, with no need to look for another control.
• Clean reference switch: when switching to a different schedule/movement as the reference, the not-yet-confirmed matches are cleared, so confirmation does not accidentally include links from the previous selection.

Improvements to aggregate movement reconciliation (POS takings and card statements):

• Card statements: the purchase invoices of the previous month are now proposed correctly in the common cases too (previously, in some situations, the search landed on the wrong month and found no invoices).
• POS cash: the cash movement proposal for the cash difference appears only when the movement carries the sale day date; without that date the invoices are selected manually, without forcing a cash movement.
• Cash date: the cash movement created for the difference is recorded on the sale day, not on the credit date.
• Clearer messages: when automatic matching finds no exact combination, or there are too many invoices in the period, or credit notes (refunds) are present, the situation is explained with a dedicated notice.
• Consistent selection: the checked invoices match exactly those that will be reconciled on confirmation.
• More robust confirmation: the cash movement and the links are recorded in a single operation, avoiding incomplete states.

The custom print layout editor (Configuration - Print layouts) gains two features in advanced mode.

Multiple, free logos. The logo is no longer tied to the header: you can add up to five logos, each picked from the image library or set to the organization logo, and position and resize them freely by dragging them in the preview.

Text formatting. Every document text - title, header, footer, organization name and fiscal data, and each section - has a formatting bar with font, size, bold, italic, underline, color and alignment. Controls left empty inherit the layout's general typography.

In normal mode layouts are unchanged.

POS takings and credit-card statements group several invoices into a single bank movement: they can now be reconciled in a guided way.

• Automatic detection: POS movements (incoming) and card statements (outgoing) are flagged with a badge in the movements grid, both for card-type resources and, for bank accounts, based on the description.
• Assisted reconciliation: the badge opens aggregate reconciliation, which lists the invoices in the period and shows in real time whether the selection balances the movement amount.
• POS takings: the issued invoices of the day are proposed. The sale date is taken from the movement competence when present, otherwise from the days before the credit.
• Cash difference: if the day's invoices exceed the POS credit, the difference was collected in cash: the system proposes the matching cash movement and records it only on confirmation, on the chosen cash register.
• Card statements: the purchase invoices of the month before the charge are proposed.
• Missing invoices: if the registered invoices do not cover the amount, the situation is flagged and nothing is forced: the invoices may not have been registered yet.

In print layout management (Configuration - Print layouts) every layout, including the built-in ones, now has a View button: it opens a preview of the layout on sample data without entering the editor. Handy for comparing layouts before choosing a default.

In addition, print PDF generation (invoices and movements) has been optimized and is now faster, especially on repeated prints.

In production, PDF generation could end with an error, both for document printing (invoices and movements) and for the PDF export of reports (balance sheet, cash flow). The issue was in the server-side runtime, not in the data.

PDF creation has now been restored: printing and exports produce the file again as expected.

The KPI page is a matrix with one indicator per row and the twelve months in columns: great on a large screen, but on phones it forced you to scroll the table sideways to read the values.

Now, on mobile only, the page looks like the Balance sheet section: a month selector at the top and the indicators as stacked cards, with the name on the left and the value for the selected month on the right, grouped by category. Everything reads without sideways scrolling.

On desktop the page stays exactly as before, with the full twelve-month table.

On the Documents page the menu previously called "Reconcile" becomes "Process with AI" and groups every AI operation on the selected documents. The operations work on the lines of the invoices that are not finalized yet (finalized ones are skipped) and write suggestions to review, without changing anything permanently.

Available items:

• Reconcile counterparts: links invoices to subjects by tax code / VAT number (deterministic, does not use AI).
• Articles (AI): suggests the catalog article for the lines of issued invoices.
• Competence (AI): infers the economic competence period of the lines.
• Account (AI): proposes the ledger account for the lines.
• Fill all with AI: runs account, competence and articles together - the same processing finalization performs, but without finalizing the documents.

Processing runs in the background: you can select many invoices at once, progress and the completion notice appear in the bell at the top right, and the list refreshes on its own when done. Finalization stays where it was, in the "Change status" menu.

Seeing issued and received invoices mixed in the same grid could be confusing. The Documents page now has a prominent switch at the top with three options - All, Issued, Received - to separate documents in a single click.

The switch stays visible at all times (including on mobile, where the other filters remain grouped under Filters) and combines with the other active filters - type, status, dates and payment status. It replaces the previous direction dropdown, which did the same thing in a less immediate way.

In the Change log (configuration section), changes to the chart of accounts - creating, renaming or deleting rows and their links - were not associated with the correct organization, so they did not show up in the log for those allowed to consult it (the org's User management and administrators).

Each chart of accounts change is now recorded with its own organization and shows up correctly in the Change log, in line with the other configuration entries.

Historical entries recorded before the fix remain without an organization and are not backfilled.

A POS or Stripe collection is the cumulative settlement of amounts that come from invoices or receipts: the counterparty shown in the description is the payment channel provider (bank acquirer, PSP), not the individual customer. For these transactions, during automatic processing (file import, bank sync, AI fill):

• no subject is linked and no record is created from the channel provider's name;
• the transaction stays pending for reconciliation (it is not booked as standalone revenue): it must be matched to the corresponding invoice(s) in the reconciler.

The rule recognizes descriptions with the standalone term "POS" (for example "Accredito POS", "Incasso POS") and those with "Stripe", on incoming transactions only. You can always assign a subject manually.

The automatic account suggestion on transactions and invoice lines relies, beyond the account name, on its description. The descriptions of the standard chart accounts have been rewritten and completed: each account now lists its typical wording and the terms it usually appears with, and notes how to tell it apart from similar accounts.

In practice:

• full description coverage for commercial and service businesses too (many accounts previously had no description);
• fewer mix-ups between look-alike accounts. For example, doctors' fees paid to professionals are routed to the dedicated account instead of the generic collaborators account; an accountant's consulting fee goes to Professional Consulting rather than collaborators;
• wording that only indicates the payment method (for example "electronic payment orders") is left without an account instead of getting an arbitrary one.

The updated descriptions are applied to existing charts without overwriting any that were already customized, and can be edited any time from the Chart of Accounts page.

The Stamp duty report previously showed a subject x month matrix, of little use for stamps: the subject column was almost always empty and the grouping did not help understand how much to pay. The page now serves its real purpose, settling the stamp tax on issued electronic invoices.

For the selected year you see three figures at the top - stamp duty due, paid and to settle - and a table with one row per quarter:

• Stamp duty due: number of stamps and amount (2 € per issued invoice with a stamp), computed automatically from the invoices.
• Paid: the amount you paid via the F24 form, entered manually. The value is saved and restored when you reopen the page.
• Residual: due minus paid (red = still to pay, blue = credit).
• Deadline for the quarter's payment.

A note at the bottom recalls that stamp duty on electronic invoices is paid quarterly (31 May, 30 Sep, 30 Nov, 28 Feb) and the 5,000 € threshold for deferring the early quarters. Always verify the figures with your accountant.

The automatic reconciler computes the safe matches between open schedules and bank transactions. It now shows a preview before applying them.

• Preview before applying: each proposed pair shows the schedule (invoice, date, amount), the transaction and the reason for the match. Proposals can be deselected individually, and only the confirmed ones are applied.
• No silent linking: automatic reconciliation no longer runs on its own during transaction import or when a bank is linked. It is started explicitly from the reconciliation screen, so no link between a schedule and a transaction is created without confirmation.

Matches are still the deterministic, safe ones (invoice number in the description, consistent amount): uncertain cases keep being proposed as candidates to confirm manually.

You can now invite someone to create their own organization, even when that organization does not exist yet.

A new button - in the user menu for everyone, and in the Organizations area for platform administrators - lets you enter an email address and send the invitation. The invited person receives an email with a link valid for 7 days: on first sign-in they set a password and are guided through a short wizard that asks for the new organization's details (name, type, currency), creates it, and makes them its administrator. The standard setup wizard starts right after.

• The link stays valid for 7 days, like the one used to set a password.
• The organization is created only when the invitee completes the wizard.
• Whoever creates the organization becomes its administrator, managing users and accounting.

In some cases - typically bulk imports on new organizations - many invoice and transaction lines were left without an account, even when the description clearly named the nature of the service or product. Those entries ended up among the ones to classify by hand.

We reworked the automatic account assignment:

• More stable matching: the account choice is now anchored to the chart-of-accounts code, removing a source of errors that left a line without an account even when a suitable one existed.
• Issued invoices: when the description directly names the service or product, the matching revenue account is assigned even if the customer is a private individual.
• It applies to both invoices (issued and received) and bank transactions.

Lines that still have no account are classified as usual from the ledger; the manually chosen account is then reused on similar entries.

Bank stamp duty transactions on the current account (e.g. "Imposta di bollo al 31/12", "Bollo c/c") are again recognised and booked automatically as an expense on the Stamp Duty account, instead of waiting to be reconciled.

Previously these transactions ended up among those to classify manually. The rule has also been made more precise: it now matches only actual stamp duty and no longer triggers on wording like "bollo auto" (vehicle road tax, which is a different thing).

No impact on invoice lines, where stamp duty was and is already handled correctly.

The Resend invite section used to disappear once the user had completed their first sign-in. It is now always available in the user's detail panel, under the Profile tab.

• For a user who has not signed in yet, it sends the invitation link to set a password and sign in.
• For a user who has already signed in, sending it acts as a password reset: the user receives an email to set a new one.
• The short wait between two consecutive sends for the same user still applies.

In progressively loaded lists (transactions, invoices, ledger, registries and similar), every data refresh sent the list back to the top: after saving a change, deleting a row or finishing an import, you lost the page and the point you were at, and had to scroll back down to where you were.

The list now reloads while keeping the current position: the page and scroll point are restored as soon as the updated data is available. Jumping back to the top only happens where it makes sense, that is when filters change (the result is a different set of rows).

When you invite someone and the invitation email does not arrive, or expires before it is used, you can now resend it with one click.

In the user's detail panel, as long as the person has not completed their first sign-in, a Resend invite section is shown. Pressing the button sends the user a fresh email with a link to set their password and sign in.

• Works with no extra email-service configuration.
• Available only until the user has completed their first sign-in; after that, the standard password recovery is used.
• To prevent repeated sends, there is a short wait between one resend and the next.

When you change the account on an invoice's ledger, Numbers asks whether to apply the choice to the other documents of the same customer or supplier. Until now this popup did not appear if the chosen account already matched the subject's default account: the correction stayed on the single document.

A typical case is a document classified on the wrong account that you set back to the right one (= the subject's default): the popup now appears and lets you propagate the account to the subject's other documents. The "Save as default" option is hidden when the account is already the default (it would have no effect): only Only this document and Save and propagate remain.

No change on bank transactions, where propagation does not apply.

When starting a bank account connection via Open Banking, in some environments the flow stopped immediately with a service-unavailable error, without reaching the bank's consent page.

The cause was an invalid return address sent to the linking service. The address is now generated correctly and compactly, and the redirect to the bank to authorize account access works as expected.

The reconciler's two columns (Transactions and Schedules) were capped at the first rows loaded: beyond a certain amount the list stopped and later entries could not be reached by scrolling. On archives with many schedules or transactions this hid everything after the first few hundred rows.

Now:

• Newest-first ordering: both columns show the most recent entries at the top (schedules by due date, transactions by transaction date), in descending order.
• Full scrolling: the list no longer stops. Scrolling down automatically loads the following blocks, until all entries in the filtered range are covered.

To narrow the list to a specific range or counterpart, the Advanced filters remain available.

When a transaction cites an invoice in its description and settles one instalment, auto-reconciliation could fail to confirm it on its own and leave it as a suggestion to approve by hand, even in the most straightforward cases.

These cases are now confirmed automatically, and the match is always applied to the correct schedule entry. Genuinely ambiguous cases (the same amount tied to several invoices) stay as a suggestion to confirm by hand, and every automatic match remains reversible.

Auto-reconciliation confirmed a transaction "by invoice number" only when it settled the invoice in full (or all its instalments together). If a transaction cited an invoice but paid a single instalment, it was left to match by hand.

Now, when a transaction cites an invoice number and its amount exactly matches one open instalment of that invoice, auto-reconciliation settles that instalment. It only fires when the amount identifies a single invoice (no ambiguity): ambiguous cases stay as suggestions to confirm manually, and every automatic link can always be undone.

In parallel, recognising the invoice number in the transaction description is more precise:

• the year is no longer matched (a short numbering like "20/26" reduces to "2026", which would appear in nearly every transaction);
• the number is no longer recognised when its digits fall inside a longer code (POS id, IBAN, amount, date): it must align to the number written in the description.

This removes spurious matches that previously polluted the suggestions and prevented correct matches from being auto-confirmed.

Under Collection Resources, saving a bank account now requires the directory subject that represents the bank or the resource issuer. If the subject is missing on save, a prompt appears with a search field to link a subject already in the directory or to create a new one on the spot. Saving proceeds only once a subject is linked.

Always linking a subject to the bank makes the features that use it as the default counterpart more reliable, for example the automatic attribution of bank fees and charges.

Some Excel bank statements failed to import: processing started but stayed stuck, and a new attempt reported that the import was already running.

It affected files where:

• the movement date is spelled out (e.g. "29 maggio 2026") instead of a numeric format;
• a long top section with account details and balances pushed the column-header row further down the sheet.

These formats are now recognized and imported correctly. In addition, when a statement reports both the accounting balance (as of the export date) and the period closing balance, the balance check uses the closing balance, avoiding false "balance does not match" warnings. A run interrupted midway now resumes on its own instead of staying stuck.

Imported invoices are finalized automatically, in batches, by a background process. In some cases - typically after re-importing documents already present and partly already finalized - that process could stall: already-finalized invoices wrongly stayed in the queue, monopolized processing and prevented later drafts from being finalized, leaving them in Open state indefinitely.

What changes:

• The finalization process now self-heals: it detects already-finalized invoices left in the queue by mistake and removes them, immediately resuming work on the pending drafts.
• A re-import no longer re-queues already-finalized invoices, so the stall cannot recur.

No action is needed: imported invoices left pending are finalized normally.

Bank transaction import (Transactions → Import / Export → Bank statement) no longer accepts PDF files. The supported formats are now Excel (.xlsx, .xls) and CSV.

To load a statement that is only available as a PDF:

• export it from your home banking as Excel or CSV (most portals offer a direct export), or
• enable automatic synchronization via Open Banking, where available.

Transactions previously imported from PDF are untouched: they stay in your archive as before.

Moving money between two accounts of the same organization - from a current account to petty cash, or between two banks - takes two transactions: an outflow from the source account and an inflow into the destination. Until now you had to enter them by hand, one at a time.

The New transaction popup now has a Transfer between accounts toggle:

• Turn it on at the top of the form (next to Balance row), on creation only.
• Pick From (source account) and To (destination account) - they must be two different accounts - plus amount and date.
• The description is optional: if left blank it's auto-generated (Transfer from <source> to <destination>).
• On save Numbers creates the two linked transactions in a single operation, both classified as Internal transfer.

Being internal transfers, the two transactions produce no economic ledger entry (they don't enter Balance sheets by category) and aren't proposed for invoice reconciliation. They stay visible in Transactions and in each account's balance.

When a bank movement matches no known rule or pattern, the classification is suggested by the AI. For outgoing transfers this suggestion was too eager: a generic transfer to a supplier (e.g. "BONIFICO IN EURO VERSO UE/SEPA"), even for a sizable amount, could end up as "Not from invoice" - recognized as an autonomous cost and kept out of reconciliation.

But an outgoing transfer is usually the payment of an invoice to reconcile. A rule symmetric to the one already active on incoming movements ("revenue comes from invoices") now applies:

• An outgoing transfer with no cost signal stays "To classify", no longer "Not from invoice". It should then be reconciled with the matching invoice (auto-reconciliation included).
• Genuine recurring costs stay "Not from invoice": rent, salary, taxes, withholdings, bank fees and charges, interest expense, insurance premiums are still recognized automatically when the description says so, even when paid by transfer.
• The amount is not a hint: a large transfer is not treated differently from a small one.
• Only the automatic suggestion changes: the classification can always be edited by hand, and direct debits (utilities, mandates) are not affected.

The Documents list now refreshes itself while a background job is working on your invoices, with no page reload needed:

• when an import finishes, the freshly imported invoices appear in the list;
• as automatic finalization progresses, imported rows move from Pending to Finalized on their own (or surface a finalization error).

The refresh is unobtrusive: your scroll position and selection stay where they are. The same behavior was already in place on the Movements list once imports completed.

In the subject form, the Default accounting account field now appears only when the subject is a supplier.

• It was misleading for customers: the field also showed up for customer-only subjects, but the chosen value was discarded on save and never used in the books. It no longer appears in that case.
• Unchanged for suppliers: the default account is a cost counterpart, applied to the supplier's received invoices and outgoing bank movements.
• For customers: the default revenue account is set on the item, not the subject.

The initial setup wizard is now actionable even when an area is already configured, instead of showing just a count.

• Configured items visible: in the Sequences, VAT codes, Payment types, Collection resources and Causes steps, when entries already exist the step lists them.
• Inline edit and add: each entry can be edited with the pencil and new ones added through the same edit dialog used in Settings, opened on top of the wizard without changing page.
• Reordered buttons: the Next action is now on the right, Back and Skip on the left.

When importing invoices, the automatic classification of the line accounting account and competence period used to be computed twice: once at import and once at finalization. It is now computed once at import and reused at finalization.

What changes for users:

• Faster finalization, especially on imports with many invoices.
• Same result: the proposed account and competence are the same as before, and any manual corrections made from the Ledger are still honored on re-finalization.

If a line's classification is not available at import time, finalization computes it on the fly as before: no line is left without an account because of this.

The initial setup wizard has been revised to remove a few inconsistencies between steps.

• Fiscal year: when a year is already configured a summary is shown and you move on, instead of re-showing the creation form prefilled with the current year (which would have raised an overlap error).
• Numbering sequences: the step now directly follows the Fiscal year it depends on, so the "create a fiscal year first" hint points to the correct previous step.
• Aligned the step labels in the final summary.

Until now, importing electronic invoices created the customer or supplier registry entry without an address: the information, although present in the XML, was lost and had to be re-entered by hand.

The import now reads the invoice address block (street, ZIP code, city, province, country) and copies it onto the registry entry:

• New entry: the invoice address becomes the primary address.
• Existing entry without an address: the address is added automatically.
• Existing entry that already has an address: no change - data entered or corrected by hand is never overwritten.

This applies to electronic invoice imports in XML format (manual upload and tax-drawer acquisition).

While importing electronic invoices, the final step that turns the preview into imported invoices could fail because of a momentary outage of the data service. When that happened the import stopped and the notification showed a long, effectively unreadable technical error page.

Two improvements:

• Automatic retry: a temporary service outage no longer fails the whole import. The operation is retried a few times automatically and, if the outage clears within a few seconds, the import continues on its own.
• Readable message: if the error persists, the notification shows a short, understandable description (e.g. "service temporarily unavailable") instead of the full technical text.

Genuine errors (for example an invalid invoice) are still reported immediately, with no pointless retries.

Invoices and transactions can now be printed as PDF with configurable layouts.

How to print. The new Print button is available in the list (with one row selected), in the invoice detail and in the transaction edit popup; on mobile, via the print icon on the transaction card. Clicking prints immediately with the default layout; the arrow next to it opens the menu to print with any of the other active layouts.

Predefined layouts. A set of ready-to-use layouts is available to all organizations: from Configuration → Print layouts they can be enabled, disabled, and one can be chosen as the default for each document type. Printing always works, even with no configuration at all.

Custom layouts. For a tailored result, custom layouts are created from a starting template and edited in a live-preview editor on sample data, where every document element can be turned on or off (logo, header, document data, counterpart, line items, totals, withholdings, schedule, footer). The simple mode shows just the on/off switches; the advanced mode adds full control: free positioning - drag and resize every element right in the preview to place them side by side as you like, section reordering - header and footer included, a per-section color, customization of every label (titles, columns, totals, wording), plus margins, paper size, orientation and typography.

Transaction printing. The transaction document carries the accounting information that matters - collection resource, ledger account and movement type - alongside date, amount, direction, subject, reference, accrual period, attachments and the linked document. Every field can be turned on or off and, in advanced mode, reordered.

Logo library. The organization's print images are uploaded once and reused across all layouts; when no dedicated logo is set, the organization logo is used.

In the Stamp duties and Bank costs views, on phones the period selector stayed on a single row and the last buttons were clipped at the right edge, effectively impossible to tap.

The period buttons - last quarter, year to date, last 12 months, all and custom - now wrap onto multiple rows on narrow screens: every option stays visible and tappable. On desktop the bar stays on a single row, unchanged.

The interface has been refined for use on phones.

• System safe areas respected. The page header, side menu and pop-over panels now account for the areas taken by the device (status bar and camera at the top, navigation indicator at the bottom): titles and controls are no longer covered.
• More tappable controls. Checkboxes, small buttons and action icons (filters, reconciliation, rename) have a larger touch area.
• Collapsible Ledger filters. On a narrow screen the filters fold behind a Filters button, with an indicator when any are active, like the other lists.
• Full-height panels. Some dialogs and overlays now use the actually visible screen height, preventing buttons from ending up behind the browser bar.

What changes

On phones, when a page had several buttons at the top (for example a primary action plus links), the header could clip the buttons on the right edge or squeeze the title until it was unreadable.

The header now adapts to narrow screens:

• the title takes the full first row and is shown in full;
• the page buttons move to a full-width second row and are no longer clipped;
• if there are many buttons, the row scrolls horizontally instead of truncating.

On desktop the header stays on a single row, exactly as before. The improvement applies to every page in the app.

On mobile devices the Trash did not show cards for deleted items: it fell back to the desktop grid, awkward to read on a small screen.

Now each of the four sections - movements, documents, ledger and subjects - lists items as compact cards with the key data (amount, date, counterpart, days remaining before permanent deletion). Every card carries the Restore and Delete permanently buttons, previously reachable only through grid selection on desktop.

Period selection has been unified and simplified on two pages.

Dimensional balance. Instead of the From and To fields there are now two controls: the granularity (Month, Quarter, Year) and the period combo (e.g. June 2026, Q2 2026, 2026). Picking a period shows the balance by Business Unit and Workplace for that period only. In the By period view the columns are the sub-buckets of the selected period (a month breaks down into weeks, a quarter or a year into months).

KPIs. Instead of the year arrows there is now a period combo (e.g. June 2026). The matrix shows a single column with each KPI's value for the selected period, read according to the KPI's frequency (monthly, quarterly or annual). The year-over-year comparison under each cell is still available.

Bulk finalization of invoices (multi-select in the list, then Change status > Finalize) has been rewritten for large volumes.

Previously each invoice was finalized one at a time from the browser: on hundreds or thousands of documents the operation was slow, kept the page busy, and stopped if you closed the tab. Now the selection is queued as a background job:

• the page is freed immediately and you can keep working;
• progress and the completion notice appear in the bell at the top right;
• the operation keeps running even with the tab closed and resumes on its own if interrupted;
• it uses the same engine as import (shared, parallel processing), so it is much faster.

The ledger, dashboard and balance sheet update once processing completes.

What changes

In the per-role cost views (Stamp duty and Bank costs), on desktop the monthly subject-by-month grid was squeezed into a small scroll area - with few rows it stayed only a few pixels tall and was effectively unusable. Now:

• the grid shows all its rows at their natural height and the whole page scrolls as one (the sidebar stays fixed) to reach the KPIs, charts and summaries below (Breakdown, Forecast, Bank comparison) - no more nested scrollbar inside the table;
• the page header stays hidden on wide screens to give the table room (the active sidebar item already signals the page).

A visual glitch was also fixed: while scrolling the table, row colours (bank subtotal, Total column) bled over the header and totals rows. Header and totals now always sit on a solid background.

On mobile the view is unchanged (per-subject cards, one month at a time).

The Log page has been aligned with the Balance sheet for filter handling and made more readable on the Changes tab.

• Collapsible filters - On both tabs (Access and Changes) the filter bar is now a collapsible group with a FILTERS heading, like in the Balance sheet. When closed, a summary of the active filters (or "No active filters") appears on the right; the open/closed state is remembered.
• Fully readable changed fields - In the Changes tab the Changed fields column shows the first fields on a single line with a +N indicator; hovering the cell opens a summary with the full list of fields.
• Changed fields on mobile - On phones each card now shows a labelled Changed fields section with the complete list of changed fields (previously the fields were truncated and unlabelled).

The Fixed assets page now organizes content into two views you can switch at the top, like the Users page.

• List: the register itself, with the status tabs (Active, Suggested, Excluded, Disposed) and all actions.
• Charts: the amortization charge per year and the active value breakdown by category.

The indicators on top stay visible in both views. The List is shown on open, so the register is immediately at hand.

The tutorial of work orders (episodes in the medical vertical) has been completed.

The "New work order" step only highlighted the button, without showing the creation popup. Now, moving forward, the tutorial opens the popup and illustrates each section with a dedicated step: Main details (number and description) and Client. Attachments, documents and ledger entries are linked later, when editing.

The tutorial of the Organization users page has been completed.

Previously the "Invite a user" step stopped at the button and the Statistics view was not illustrated. Now the tutorial:

• switches to the Statistics view and highlights the charts one by one: key indicators, daily logins and most-active users (on an organization without data the charts show a sample);
• opens the invite popup and walks its sections with a dedicated step each: Identity (first name, last name, email), initial Roles and Sign-in method (invite email or initial password);
• returns to the List view at the end.

The in-page step-by-step guide was named in different ways - "tour", "guided tour" or "tutorial" depending on the spot. Now there is a single term, tutorial, consistent with the Skip tutorial button and the per-user state.

The alignment covers the button that restarts the guide, the notices shown when a page without data uses a sample set, and the text of the individual steps.

The tutorial of the subject creation popup has been fixed and completed.

The step titled "Addresses and contacts" actually highlighted only the addresses section, leaving contacts out; the Tags section was not illustrated by any step. Now each form section has its own dedicated step: Addresses, Contacts and Tags, each with its area correctly highlighted.

The tutorial of the ledger entry creation popup has been completed.

The walkthrough stopped at the Classification section, leaving out the allocation sections (Business Unit and Workplace) that close the form. Now a dedicated step highlights the allocation area and explains how to split the entry by percentage across multiple dimensions.

The tutorial of the fixed assets register has been completed.

The "New asset" step only highlighted the button, without showing the creation popup. Now, moving forward, the tutorial opens the popup and illustrates each form section with a dedicated step: Asset origin, Details (including the code used for numbering), Amortization and, for manual assets, Account and supplier.

Tutorials now stay useful even on a freshly created organization that has no real data yet.

Previously, starting a tutorial on an empty page meant the steps illustrating grids, charts or indicators landed on sections with no content and lost their meaning. Now, when the page has no real data, the tutorial shows a set of sample data: each step highlights a populated section, just as it would in normal use.

A notice at the top of the page flags that the data shown is fictitious and for demonstration only. Sample data appears only during the tutorial and only when there is nothing real to show: as soon as the page has its own data, the tutorial uses that.

Small interface fixes.

• Invoice import: on confirm, the wait now shows a simple loading indicator. A progress bar that stayed stuck at zero and a Cancel button that did not stop anything were removed (the import runs in the background and the result arrives via notification).
• Medical dashboard: the Recalculate button stays disabled until there are changes to apply, avoiding pointless recalculations.

The Log page tutorial has been extended.

• In addition to the Access events tab, the walkthrough now also covers the Data changes tab, showing its filters and the list of writes.
• The step order follows the page top to bottom: header, organization filter, tabs, then each tab's filters and list.

The in-app Guides and FAQs list updates correctly again.

• Some recent sections could fail to appear: guides and FAQs are now synced in full.

The Automatic rules page is now easier to read.

• On phones the list no longer scrolls sideways: each rule is shown as a card with its conditions, output and actions.
• The When and Then condition text now wraps and is shown in full, instead of being cut off when it contains many values.

Some fixes to the Fixed assets section:

• Disposed assets now correctly show a fully accrued value and zero residual from the disposal date, in line with the reversal of later amounts: grid, statistics and chart stay consistent.
• In the invoice line editor the competence dates now stay locked for disposed assets too, not only confirmed ones, preventing edits that would be rejected on save anyway.
• Disposal is now blocked with a clear message when the source cost is no longer live in the ledger (document not finalized or in the trash).
• General robustness improvements for bulk actions on assets (confirm, exclude, undo disposal) and for disposal under concurrent operations.

The Fixed assets page gains statistics and charts, computed with the same criteria as the accrual balance.

• Amortization charge per year: how much amortization weighs (and will weigh) on the income statement, year by year, with disposal adjustments already included. Distant years are grouped at the chart edges, so the bars always add up to the total value.
• Active value by category: the composition of the assets being amortized.
• A new indicator shows the current-year amortization charge.

On the Fixed assets page an asset can now be disposed before the end of its amortization, choosing date and reason (scrapping or sale).

The accounting effect is previewed and handled automatically: accrued amounts stay unchanged, the residual value becomes a cost at the disposal date and later amounts are reversed, so from the following day the asset no longer weighs on the income statement. For sales, the proceeds are recorded as a regular invoice or collection and the gain/loss emerges in the income statement.

A disposal can be undone from the Disposed tab: the adjustments are removed and the asset goes back to amortizing.

Reconciliation between bank transactions and invoice schedules becomes more capable, while staying deterministic and transparent:

• Multiple transactions per schedule: when a down payment and a balance (or several installments) reference the same invoice and their sum covers the schedule, automatic reconciliation links them all, each for its own share. If a single transaction alone would cover the whole schedule, Numbers suspects a duplicate payment and leaves the decision to the user.
• Cumulative wires: a transaction referencing several invoices and covering the sum of their remaining shares gets linked to all the referenced schedules at once.
• Remaining-based matching: scores, combinations and auto-links use the still-open share of schedules and transactions, not the full amount. A partially-paid schedule is proposed and closed by the transaction covering the rest.
• More reliable counterpart matching: when transaction and invoice share the same assigned subject, the match is treated as certain, stronger than name similarity.
• Late payments: transactions up to 180 days after the due date now receive a reduced date score instead of zero, improving candidate ranking.

The new Fixed assets page keeps the register of the organization's depreciable assets: machinery, equipment, vehicles, furniture, hardware, buildings, software.

How assets are created. Numbers analyzes purchase invoice lines above a configurable threshold and suggests likely assets (deterministic rules plus AI, which excludes rentals, recurring fees and maintenance). Suggestions are confirmed or excluded with one click, from the register or directly from the line editor. Alternatively an asset can be created manually, including assets bought before adopting Numbers, or from an already recorded bank movement, with no double counting.

How amortization works. It is managerial, with a user-chosen duration and per-category defaults derived from fiscal coefficients: the cost is spread on an accrual basis over the whole window, and the accrual income statement apportions it pro-rata by day. For each asset the register shows the amount amortized to date and the residual, always consistent with the balance.

Each asset can have an inventory code, category, cost account, supplier, notes and a photo. The "Analyze existing invoices" button also covers the archive of the last 24 months.

Opening the browser developer console (the tool behind the F12 key) now shows a security warning.

It defends against a common scam: an attacker convinces the victim to open the console and paste code into it, promising to "unlock a feature" or "access another person's account". That code actually hands a third party access to your own account.

The warning is a reminder never to paste code whose exact effect you do not understand. It appears in the configured language or, for signed-out visitors, in the browser language.

In the clinical dashboard, pressing Recalculate statistics could collapse the Configuration tab to a single activity: the data was saved correctly and reloading the page restored the full list, but it looked emptied until the reload. The same could happen when renaming an activity, using AI re-classification, or refreshing the referring-doctor and therapist tables in place.

The list now stays complete immediately after the operation, with no page reload needed.

In the clinical dashboard, opening a referring doctor's detail, each patient now shows - alongside the number of physiotherapy sessions - two new columns with the amounts generated:

• Physio amount - total amount of the patient's physiotherapy sessions.
• Non-physio amount - total amount of the other services (first visits, specialist visits, follow-ups).

Amounts are based on the prices in the uploaded file, span the patient's whole history and exclude cancelled appointments. When the file has no price for a service type the column stays empty (-): the dash means missing data, not a zero amount.

Exporting a Ledger row, duplicating it in Excel and re-importing it used to drop the account and the Business Unit / Workplace allocation: the import ignored them. The round-trip is now complete.

AI automatically recognizes the Account, Subject, Business Unit and Workplace columns when present in the file. The account is matched by code, the subject by name, the allocations by entity name (format "Name 50%, Name 50%"). A value that doesn't match an existing record does not block the import: the row is created with that field left empty, and at the end of the import a notice summarizes how many rows had an unmatched account, subject or allocation.

Work order management (Episodes in medical organizations) now assigns the work order on its own, cutting manual effort.

• On invoice finalization, when it has no work order yet:
  • Issued: the work order is matched by the code carried in the XML (CodiceCommessaConvenzione) and, if missing, created with that number.
  • Issued and received: otherwise the same subject's history (issued and received documents) within a window of days is checked. If exactly one recently used active work order exists, it is assigned; if none, one is created; if two or more, the invoice is left without a work order (manual assignment).
• The window in days (default 120) is a global parameter, editable by the superadmin under Security.
• The work order can now also be set from the finalized document detail: pick it from the dropdown or create a new one on the fly. The ledger rows, including the VAT row, stay aligned.

Work orders management is now available from the sidebar. A work order collects the documents and ledger entries of a job, so you can analyze its profitability.

• Create a work order with number, description, client and attachments.
• Assign the work order to an invoice from the document creation form.
• On invoice finalization the link is automatically propagated to the economic ledger rows.
• Archive completed work orders without losing their links.

In medical center organizations the section is called Episodes.

Electronic invoices downloaded from SDI or your fiscal intermediary almost always arrive digitally signed, with a .xml.p7m extension (CAdES signature). Until now the importer rejected them and you had to extract the XML by hand.

You can now upload them directly:

• Single .xml.p7m files or files inside a .zip archive.
• The PKCS#7 cryptographic envelope is unwrapped automatically and the inner FatturaPA XML is imported like a plain .xml.
• Works in preview, bulk confirm and background import.

The signature is not verified: unwrapping only reads the document; formal validity is already guaranteed upstream by SDI.

The section that gathers every economic entry, until now labelled Partitario in the Italian UI, is now called Registro.

The word "partitario" was often read as tied to double-entry bookkeeping, which Numbers does not use. "Registro" better describes what the page actually is: the complete, authoritative list of the company's economic entries, distinct from bank Movimenti (transactions).

Only labels change (menu item, page title, guide and tour copy). Data, filters, exports and links are unchanged. The English label remains Ledger.