What we shipped, in plain language.

Earnings, allowances, overtime, terminal benefits, deductions, statutory and technical wage types are seeded into every tenant, so compensation plans, income-tax records and payslips can be configured immediately instead of starting from an empty list.

Seeding is keyed on the wage type code, so a re-run changes nothing and an administrator’s edits to an existing wage type are never reverted.

The Inventory Dashboard grouped receipt history by material and stamped that one date on every batch of it. Batch-managed stock therefore reported a single age for every batch:

• On the goods-receipt basis, the material’s earliest ever receipt was applied to everything, so stock received this morning was reported as months old and fell into the wrong ageing bucket
• On the last-receipt basis, one new receipt reset the reported age of all existing stock of that material to zero — hiding genuinely old stock from the slow-moving and write-off reports, which is the dangerous direction

Each batch is now aged from its own history, with the material rollup used only for stock that is not batch-managed. A back-dated report also no longer picks up receipts that had not happened yet.

The report date used to change only the age arithmetic — the quantities were always today’s, so an “as at 30 June” report showed current stock against June ages and could never tie to the stock balance on that date.

• Quantities are now reconstructed for the chosen date by rolling back the movements since, so the report ties to the balance as at that date
• A NEW stock class separates recently-arrived stock from the ageing buckets it would otherwise distort
• Ageing policy — bucket boundaries and the ageing basis — is set per company

Foreign-key lookups now resolve on keys that are genuinely unique, a lookup field is only ever upgraded to a better one, and a date typed into a date-and-time column is accepted instead of rejected.

Uploading a corrected version of a file created a second copy of every row rather than matching and updating the existing ones. Matching is now reliable, and soft-deleted rows no longer claim a key they should have released.

A lookup can be constrained to a subset of candidate records, so an ambiguous name resolves within the right scope. A separate utility identifies and merges tasks that earlier uploads duplicated.

The ship-to address lived only on the order header, so a single customer purchase order covering several destinations had to be entered as several sales orders — splitting the commercial document to describe a logistics fact.

• A sales order line can now name its own delivery address; a line that names none continues to use the header address, exactly as before
• Nothing is backfilled and existing orders are unchanged
• Material planning now distinguishes identical lines bound for different destinations. Previously four identical lines to four warehouses collapsed into one — a 75% short plan, announced only in a log line

Combinations that would record something untrue are refused rather than saved: a line addressed to a different customer, a delivery note line bound somewhere other than the note, and consolidating notes with different destinations onto one invoice.

Upload a logo against the tenant and it is shown on the sign-in page for that tenant’s address, before anyone has signed in. A tenant with no logo uploaded sees the page exactly as before.

The worksheet no longer opens empty until a paging control is touched, and a second attempt to save the same set of edits no longer reports “Nothing to save”.

The task list can be narrowed to a due-date window, and a change to a task’s planned end date is noted in its chat, so the reason travels with the date.

A task that has not been finished is now its own status. It renders in the ordinary colour and only turns red once its planned end date has passed, so “not done yet” and “overdue” stop looking the same.

The task worksheet gains a Pivot Analysis tab, and both the worksheet and the pivot remember your layout — column order, widths, grouping and filters — as named variants saved per user.

Debit and credit notes offered the full tax code list regardless of the company on the note or whether it was a customer or a vendor note. The list is now scoped to both.

Simulating a posting shows who created the document, drafts included, so a preview passed to a reviewer carries its own authorship.

This is deliberately not the same list as the Outstanding Payables tab beside it. That one shows invoices and debit notes inside the selected fiscal year; Open Items shows the sub-ledger’s own uncleared items in both directions across every year — so an unadjusted advance appears here and nowhere else on the page.

Totals come from the server over the same filters, never from the rows on screen, and an item carrying no exchange rate is excluded by name on the banner rather than quietly folded in.

Presentation and filtering improvements across the Vendor Ledger and the Bank Book.

The Payments Received form derives the currency and exchange rate from the invoices selected, and drives the amount from them, so a foreign-currency receipt cannot be captured against a mismatched rate. A manual General Ledger entry added to an export receipt now carries its own foreign amount.

The journal entry screen now opens on the fiscal year that contains today, and vendor invoices open as a worklist of what needs attention rather than the entire history. Both grids also stopped issuing a query per row, so they load faster on top of loading less.

Saving a receipt without picking any invoice did not hold it unapplied — it settled the customer’s open items oldest-first at posting time. An unidentified remittance therefore cleared invoices on a guess, on a posted document, which then had to be unpicked.

A receipt can now be marked on account, so “not chosen yet” and “deliberately unapplied” are different states. Existing receipts are unaffected.

Down payment requests, advance payments and on-account payments all put money on the debit side of a supplier account before an invoice is settled in cash. One page now shows everything an invoice can be cleared with and applies it.

• Setting an advance against an invoice is a reclassification, not a payment — no bank account moves on this screen
• Fixes a live defect: an invoice relieved by a down payment adjustment still read as fully open, so the next payment run proposed it at full value and the supplier could be paid twice for the part already advanced

Provisioning ran the full set of roughly 1,270 schema migrations against an empty database — upwards of eight hours for a recent onboarding, and growing with every release.

A new tenant is now created as a file-level copy of a pre-migrated template: no schema statement is re-executed, and the clone starts fully up to date. The cost is the size of the template, not the number of migrations.

• The template is kept level with the live schema as part of the normal upgrade run
• The old path remains the last rung — no template, a stale template or a failed clone all fall back to the full migration run, correct and merely slow, with the reason logged

The application carries hundreds of maintenance and reporting jobs — seeders, reconciliations, backfills, health checks — and every one of them was reachable only over a server session, with several different spellings of “report” versus “apply” and two opposite defaults.

• A published catalogue of what each job does, with one uniform way to say “report only” and “apply”
• The catalogue is an allow-list: a job that has not been reviewed and registered is not runnable from the console, full stop
• Apply mode requires the job’s own domain permission, not merely access to the console
• Every run is recorded — who, when, against which tenant, with what result — and a run that was queued but never picked up is reported as stalled rather than failed