> For the complete documentation index, see [llms.txt](https://docs.docbits.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.docbits.com/overview-and-basics/release-notes/incremental-updates-4-14-july-2026.md). # Hotfixes 4–14 July 2026 *A rundown of what changed for you in this DocBits release. Every service below lists the version now rolling out, followed by what's new or fixed in plain language — no ticket numbers, no engineering jargon. Services not listed had no customer-facing changes in this window.* *** ## Highlights * **Multi-organisation login.** Users who belong to several organisations now get a proper organisation picker at login, an organisation switcher in the header, and a default-organisation setting. Sessions are securely bound to one organisation at a time, and the app automatically follows the region of the active organisation. Logging in against the wrong region now retries the correct one automatically instead of failing. * **Release channels (frozen / latest).** Organisations can now be pinned to a stable ("frozen") release while others receive the latest updates — enabling controlled rollouts. The Service Versions dialog shows a new *Release* column, and administrators manage the channel from Company Information. Several services show larger version jumps in this window purely because of the new channel version numbering — those jumps carry no functional change. * **Configurable rules engines.** Three new rule systems arrive in the API (each off by default, enabled per organisation): **validation rules** that check extracted values and flag failures directly on the document, **transformation rules** that clean up or rewrite extracted field and table values automatically, and **rule-based layout selection** that picks the right document layout by rules instead of by where the document came from. * **E-mail import transparency.** The e-mail import log now shows one expandable row per attachment, tells you which documents were created (with buttons that jump straight to them on the dashboard), flags skipped and split items, and lets you download the original e-mail as an `.eml` file. * **AI table extraction.** A new structured AI extraction mode for tables, with a "Use AI" checkbox per table and per column in Document Type settings. * **Web App stability.** Fixed an infinite reload loop after an expired session, fixed the broken Layout Builder, and extraction tables now have a draggable height resizer. * **New: Auth Bridge Service.** A new service keeps login data continuously in sync between the EU and US regions, with self-healing and monitoring built in. *** ## API Service — live: `12.57.8` * **Validation rules (new, per organisation):** an administrator-configurable rules engine checks extracted values (totals, required fields, and more) and marks failures directly on the document, including which rule fired. Rules can be tested in a dry run before enabling, can be switched on per document type, and ship with a starter catalogue of default rules (all disabled until you opt in). * **Transformation rules (new, per organisation):** automatically clean up or rewrite extracted field and table values during processing — configurable per document type or for the whole organisation. * **Rule-based layout selection (new):** document layouts can now be chosen by configurable rules instead of being tied to the document's origin. Existing origin-based behaviour is migrated automatically, layout templates can be renamed, and duplicate layout titles are prevented. * **Faster dashboard exports:** exports triggered from the dashboard are now dispatched to a dedicated worker instead of waiting for a polling cycle, so they start promptly. * **Duplicate Detection export block fixed:** the export block for suspected duplicates works again. * **Settings that wouldn't stick:** fixed saved preferences occasionally not persisting when an older deleted copy of the same setting existed. * **Documents with unusual characters:** fixed save errors caused by invisible NUL characters in extracted data. * **Correct "Updated by":** documents uploaded automatically as e-documents no longer show a system user as the last editor — the field stays empty until a person actually edits. * **Scanned PDFs with a good text layer:** a new option lets DocBits trust the text already embedded in a scanned page instead of re-running OCR — faster and often more accurate. * **E-invoices:** more robust detection of embedded XML when the original file needs to be re-checked. * **Tasks:** new organisation toggle that lets non-administrators use the "All" filter in the task list. * **Line-item matching:** fuzzy-matching behaviour is now configurable per line. * **Stability:** WebSocket connections close cleanly on errors instead of raising server exceptions; permission-cache synchronisation verifies and repairs itself; the service version is now visible on the health endpoint. ## Auth Service — live: `1.71.1` * **Multi-organisation login:** login now asks which organisation to enter when a user belongs to several, sessions are bound to that organisation, and new endpoints support selecting, switching, and setting a default organisation. Duplicate or conflicting organisation memberships were cleaned up and are now prevented at the database level, with faster membership lookups. * **Default organisation fixes:** logging in auto-selects your default organisation (not an arbitrary one), and changing the default takes effect immediately instead of showing stale profile data. * **Logout fixed:** resolved a server error (HTTP 500) on logout and restored the token-revocation endpoint. * **Token safety:** token verification and caching now respect the organisation a token was issued for, and token revocation is centralised. * **Release channels:** the organisation's release channel is stored here, manageable by org administrators, and exposed to the app and routing layer. ## Auth Bridge Service — live: `0.2.4.2` *(new service)* * **What it is:** a new service that continuously replicates authentication data between the EU and US regions, so accounts and logins stay consistent across regions. * **Self-healing:** it detects and repairs data drift between regions — including making sure deletions propagate — and recovers automatically from connection loss instead of dropping data. * **Safety and monitoring:** an earlier bidirectional replication loop was stopped and is now actively detected and gated; error tracking and alerting are wired in; and the service reports its version in the Service Versions dialog. ## Docflow Service — live: `2.6.1` * **Workflow cards accept empty values:** checkbox and partner cards no longer fail when a field is legitimately empty; card type checks are stricter and more predictable. * **Workflows re-run on real changes:** the workflow lock again respects the document status from the trigger, and now also tracks the document version — so a document whose data genuinely changed can go through the workflow again even with the same status, while true duplicates stay blocked. * **Bigger advanced workflows:** the limit on workflow nodes was raised and is now configurable per environment. * **Alternate export:** workflow-triggered alternate exports are now labelled as such so downstream systems can distinguish them. * **Resilience:** the service reconnects automatically when a database connection is dropped mid-use, tolerates a slower message broker instead of failing, and failed API requests are now logged with full context and traceable execution IDs. ## Email Service — live: `1.38.4` * **Import log, rebuilt for traceability:** every imported e-mail now records which documents were created from it, with per-attachment detail rows. * **Original e-mail download:** the original message can be downloaded as an `.eml` file straight from the import log. * **Attachment recovery:** the corruption-recovery path now also handles plain-text messages, so more damaged inbound e-mails are recovered instead of skipped. ## Extraction Service — live: `1.51.6` * **Tax/net no longer swapped:** fixed a case on US documents where the tax amount could be assigned as larger than the net amount when several candidate pairs were found. * **Multiple tax rates per supplier:** extraction now handles suppliers whose invoices carry different tax rates on one document. * **AI table extraction (new, opt-in):** structured AI extraction endpoints for tables, activated per organisation via feature flag. * **Faster AI calls:** tuned the AI model configuration used during extraction to avoid unnecessary processing time. * **Crash fix:** resolved an error on documents that produced an empty candidate list during extraction. ## Fulltext Service — live: `1.37.2` * **Search index migrations repaired:** restored migration definitions that had drifted, keeping search index upgrades reliable. * Internal routing work for the new release-channel infrastructure. ## PO Match Service — live: `1.58.2` * **More tolerant matching:** PO matching no longer fails on unusual data — non-text item numbers, missing quantities, and non-text amount values are now handled gracefully instead of erroring. ## Web App — live: `10.41.8` * **Multi-organisation experience:** new organisation picker page at login, a dedicated organisation-switcher icon in the header, default-organisation settings, and the app follows the region of your active organisation. Logging in against the wrong region silently retries the correct region and routes you to the organisation picker when needed. * **No more endless reloads:** fixed an infinite reload loop that could occur when the server rejected a stored session token — the app now forces a real token refresh instead of reloading forever. * **Layout Builder fixed:** the Layout Builder works again, and layout selection is decoupled from the document's origin (matching the new rule-based selection in the API). * **Extraction tables:** line-item tables now have a draggable resize handle so you can give the table more room while validating. * **E-mail import log:** new skipped status and split badges, expandable per-attachment rows, original-e-mail download, and document-ID buttons that jump straight to the dashboard filtered to that document. * **Dashboard search:** the query-value dropdown now shows the localized label for list-of-values fields, and the search help examples were reworked. * **Settings reliability:** user preferences now load reliably when signing in via SSO, and the save confirmation is only shown when the save actually succeeded. * **Tasks:** the "All" filter can be restored for non-administrators via a new organisation toggle. * **Watchdog logs:** no longer capped at 10,000 entries, plus general usability improvements. * **Support tickets:** the support form pre-fills your e-mail address from your profile. * **Document Type settings:** new "Use AI" checkbox on tables and columns to control AI-assisted table extraction. * **Service Versions dialog:** new *Release* column showing each service's channel (frozen/latest), routed so it stays fast for pinned organisations. * **Field Validation:** fixed an error when returning to Field Validation from another screen, and the "Scripts" button no longer routes to a 404 page. *** ## Version renumbering only (no functional changes) **Auto Accounting** (`1.20.1`), **Barcode Service** (`1.17.1`), **OCR Service** (`1.9.1`), **FTP Service** (`1.31.1`), **Operator Service** (`1.40.2`), and **Ideas Service** (`0.3.1`) were re-versioned as part of the new release-channel infrastructure. Their larger-looking version jumps carry no feature or behaviour changes in this window. **Docnet Service** (`1.54.6`) is unchanged since 19 June. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.docbits.com/overview-and-basics/release-notes/incremental-updates-4-14-july-2026.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.