Monthly release notes for Staple AI
Stay up to date with everything new in Staple AI — new features, enhancements, and fixes shipped each month. Subscribe via RSS to get release notes as they go out.
Stay up to date with everything new in Staple AI — new features, enhancements, and fixes shipped each month. Subscribe via RSS to get release notes as they go out.
March’s release focuses on improving scanning performance, global usability, model configuration efficiency, integration flexibility, and data accessibility. These updates reduce manual effort, improve reliability, and make the platform more scalable for enterprise workflows.
Document scanning has been enhanced to improve both processing speed and reliability, especially for large and complex documents.
Creating similar templates previously required manually re-tagging fields, even when configurations were identical.
This release introduces copy and paste functionality for sidebar fields in Model Builder.
Localisation support has been expanded further across the platform to improve usability for international teams.
Localisation now covers:
Supported languages:
Webhook creation previously required selecting an event, which prevented manual or ad-hoc workflows.
This release introduces support for event-optional webhook configuration.
Email metadata was previously available in the system but not included in exported or downloaded data.
This release enables email metadata to be included in all exports and downloads.
Included metadata fields:
Available in:
Document actions were previously not controlled at a granular level based on user roles.
This release introduces role-based access control for export and document actions.
Multiple export (re-export) restricted based on user roles
Role-based control for:
Removing documents
Removing exported documents
The backend ORM layer has been upgraded to improve system stability and maintainability.
February’s release focuses on improving navigation, reconciliation flexibility, localisation, and document processing accuracy. These updates make it easier to manage large numbers of queues and models, add greater control to reconciliation workflows, and improve the platform experience for international teams.
Managing a large number of Queue Groups, Queues, and Models can make navigation difficult. Previously, users relied on browser search to locate items, which did not preserve queue hierarchy and made it harder to understand context.
This release introduces native search and sorting capabilities to improve discoverability and help users prioritize their work more efficiently.
Search
Users can now search for Queue Groups and Queues directly from the Queues page.
Users can also search Models by name from the Models page.
Sorting
Queues can now be sorted using multiple criteria:
Models can also be sorted alphabetically:
Reconciliation previously relied only on system-generated statuses such as Fully Reconciled, Partially Reconciled, Not Reconciled, and Error. These statuses were sometimes insufficient to represent business-specific reconciliation scenarios.
This release introduces custom labels for reconciliation, allowing users to classify reconciliation results based on configurable business rules.
Users can now create reconciliation rules using field comparisons: When the rule condition is satisfied, a user-defined label is automatically applied to the reconciliation.
Example:
Custom labels work alongside the existing reconciliation statuses without changing current workflows.
Previously, the system used file name as the primary identifier for documents in reconciliation views. File names are often auto-generated and not meaningful for users.
This release allows customers to configure a business-relevant field to display as the document identifier.
A new configuration allows users to define a display_name_field for each document type.
Examples of display fields:
When configured, the selected field appears as the primary document identifier in reconciliation lists and document links.
If no display field is configured, the system continues to display the file name.
Staple now supports localisation for Chinese and Japanese users across the main dashboards.
This enhancement allows users to interact with the platform in their preferred language, improving usability for international teams.
Chinese localisation is now available for:
The platform interface, including buttons, labels, notifications, and messages, is translated to support Chinese-speaking users.
Japanese localisation support has also been extended to improve accessibility for teams operating in Japan.
Some documents contain multiple tables of the same category on a single page. Previously, these tables were processed separately, which could create duplicate table structures.
This release introduces automatic consolidation of tables with the same category.
Previously, complex table extraction always triggered translation requests to the translation service, even when translation was disabled or unnecessary. This caused unnecessary processing overhead.
This release improves the translation workflow for complex table extraction.
To simplify the product interface and align with the current architecture, several legacy options have been removed from the UI.
This simplifies the configuration experience and removes outdated UI elements.
This release includes backend improvements to improve security and maintainability of the GraphQL service.
Express upgraded to a security-patched version
Legacy MySQL-related code removed
Critical dependencies upgraded:
cookie
jsonwebtoken
qs
New yer and powerful new releases.
January’s release focuses on improving reconciliation clarity,model flexibility,document navigation,export reliability, andglobal readiness. These updates reduce manual work, remove workflow friction, and give users more control over how documents are viewed, edited, reconciled, and exported.
During reconciliation, users had limited visibility into the original source documents and their related matches. Verifying values often required navigating away or manually locating documents, slowing down reconciliation and increasing error risk. To eliminate this, we’ve launched scanned document view in Reconciliation.
How it works
A new “eye” iconis available underActions in:
Matches view
Matches list
Reconciled view
Clicking the icon opens the scanned document view, showing:
The anchor document (for example, Invoice)
All related matching documents (for example, Purchase Orders, Delivery Orders)
Supports:
Zoom
Edit
Redirection to the specific document
Multiple matching types are supported:
Type 1: doc1, doc2, doc3 (thumbnail view)
Type 2: doc1, doc2, doc3 (thumbnail view)
Type 3 and beyond — up to 10 document types
Example: In a three-way match scenario (Invoice, Purchase Order, Delivery Order), users can view the invoice alongside all its matching POs and delivery orders in one place.
Benefits
Users previously had to rely on numbering field names to maintain sidebar order, making models harder to manage as they evolved. A new reorder iconis available in the sidebar field list forcustom models
Users can:
Benefits
Applicable Use Cases
In Germany and parts of Europe, invoices cannot be reliably ingested via email forwarding due to regulatory and technical constraints. Hence, Germany launched e-invoicing via PEPPOL. Staple AI quickly integrated with Peppol to help transfer the documents and stay compliant.
How it works
Benefits
Users working with complex tables had to repeatedly close or switch panels to reach their preferred view, slowing down document editing. A new queue-level preference allows users to define how the Edit Document page opens:
Options:
Configured via: Queue Settings → Preferences → Edit Document Page Layout
Benefits
Global teams were forced to use the platform in English, limiting adoption and usability in non-English regions. Henceforth, they can use Staple AI in their native language.
How it works
Full frontend migration to an i18n framework
Supported languages:
English (US/UK)
Simplified Chinese (zh-CN)
Japanese (ja-JP)
All UI text replaced with translation keys
Locale-specific date and time formatting
Benefits
Applicable Use Cases
Document intake states were rigid, limiting flexibility when users needed to correct where documents sat in the workflow.
How it works
New option to set default intake state to “Received”
Users can:
Move documents from Received → Uploaded
Move documents from Uploaded → Received
Documents in the Uploaded tab can now be deleted
Benefits
Automatic export previously:
This caused operational friction and manual exports.
How it works now:
Small or irregularly sized QR codes were not reliably scanned. QR codes are now scanned regardless of size
December’s release focuses on speed, accuracy, and control across the document lifecycle. This update makes it easier to find documents instantly, gives teams more flexibility when working with complex data and mappings, and ensures that user edits are consistently respected across scanning, reconciliation, search, and downstream systems. Together, these enhancements reduce manual effort, improve traceability, and help teams move through reconciliation with greater confidence.
Finding a specific document just got faster.
You can now search using Document ID(a system-generated unique identifier) in both theScanning pageand theAdvanced Search page. This is especially useful when you already have a document ID from an API, email, support ticket, or audit trail.
What’s included
Why it matters
Note: For now, Document ID is not added as a visible column on the scanning table, and search behavior remains within the selected tab (Received, Completed, etc.).
We’ve introduced a robust way to track and prioritize user edits made on the scanning screen, especially for documents processed using the data mapper.
When users edit extracted or mapped values, those edits are now treated as the single source of truth across the product.
Order of precedence When displaying or exporting data, the system now follows this order:
What’s included
User edits are clearly identified and tracked for data‑mapper‑related fields. Therefore,
Changes made during data extraction automatically reflect in the matching and reconciliation screens.
Regardless of whether the extracted data is in English or in the document’s original language, user edits are reflected consistently across both versions.
Display, exports and APIs all reflect user‑edited values consistently.
In the queue search as well as advanced search, users can search using the edited value.
Assuming, a field during scanning has multiple matches from the master data, the user can always pick the right match from the dropdown for each field.
Edited values always override extracted, matched, or system‑generated values.
Edits are saved only when the user clicks Complete.
Unsaved edits are discarded on refresh, rescan, logout, or tab close.
Why it matters
Reconciliation now handles all complex table types. Previously, only select table structures were supported; all table variants are now fully supported in Recon.
We’ve increased the data mapper rule limit per queue from 7 to 15.
This change is driven by e‑invoicing and complex document scenarios where more fields need to be mapped reliably.
What’s included
Why it matters
As always, these updates are live for all customers. If you have feedback or want to see related enhancements next, please eor your Staple AI contact.
This month’s updates focus on giving you more control, accuracy, and security across your reconciliation workflows. From multi-key document matching to force-unmatch capabilities and expanded reporting categories, each enhancement is designed to make Staple AI more reliable in real-world, high-volume operations. We’ve also added OAuth2.0 support for enterprise webhook delivery, helping teams meet stricter security requirements without changing existing workflows.
When fuzzy matching fails, or when no unique match is found based on the business rules due to insufficient data, we may need to use Force Unmatch.
You can now manually remove incorrect matchesduring reconciliation. If the system auto-matches two lines that don’t actually belong together, you can override it with a single action. The newForce Unmatch option lets you break a match created by the reconciliation engine.
This is useful when:
Once removed, the match is blocked from appearing againunless you reconfigure it. Once a match is removed, the system immediatelyre-runs reconciliation for that document, ensuring your view stays consistent. Reach out to us at support@staple.io to setup it up for you.
You can now configure document matching using multiple fields instead of a single key, allowing far more accurate reconciliation across documents such as invoices, POs, GRNs, and delivery notes.
This update is especially useful when a single identifier isn’t enough to guarantee a correct match. Reach out to us at support@staple.io to setup it up for you.
You can specify whether the system should match documents only when allconditions meet (AND) orany of the selected fields match (OR).
This gives you more control over how strict or flexible the matching rules should be.
Eg, in three way matching (Invoice vs multiple purchase orders vs Delivery note)
You can now combine fields such as:
This reduces mismatches and improves accuracy in reconciliation-heavy workflows. Reach out to us at support@staple.io to setup it up for you.
You can now choose how many documents you want to see at once in the Scanning page list view. Your selection is saved automatically and applied across all your scanning queues.
Default Setting: The list view loads 50 rows by default.
To Change the Row Count:
Tips
Performance: Higher row counts (100–200) may take slightly longer to load, but give a better overview of your queue.
Consistency: Your choice instantly applies to all scanning queues for a uniform experience.
Confirmation: A success message appears each time you update your setting.
We’ve extended the Document Counter inside Staple Chat to support three additional document statuses: Failed to Export, Rejected, and Removed. These new categories help teams generate complete status-level reporting without disrupting the existing flow.
New Status Categories:
You’ll now see three new options when you use the Document Counter:
These appear alongside your existing status options and follow the same selection flow (Queues → Time Range → Report Type).
This works across all queue selections and date ranges, including predefined options like “Last 7 Days” or “This Month.”
We’ve introduced support for OAuth2.0 Client Credentials authentication for inbound webhook requests, enabling secure delivery of events to enterprise-grade endpoints.
Organizations with strict security requirements can now ensure that all webhook calls from Staple:
Once OAuth2.0 is set up for your company:
Authorization: Bearer {access_token}Share these details with the Staple AI team:
Reach out to us at support@staple.io to setup it up for you.
This month, we’ve focused on making document handling more visual and intuitive, simplifying complex table workflows, and strengthening audit data security. Below are the new features and enhancements rolled out in October.
Ever opened a single file and found ten different documents inside?
Two invoices, eight purchase orders, a couple of delivery orders… all bundled together like a mystery box.
Every large enterprise faces this challenge: messy, multi-type documents that need to be split, rotated, renamed, and sent to the correct queues before processing can even begin. And, every industry faces them with documents from claims to contracts, from financial documents to KYC docs.
That’s exactly what the Document Editor in Staple AI fixes.
With Document Editor, you can:
It’s like giving your document workflow a pair of hands and a brain. No more manual sorting. No more re-uploading. Just upload once, edit visually, and let AI handle the rest. In short, you control how your documents look before they even reach processing.
Multiple varieties of tables in the same dataset? No problem.
What it does
Why it matters: Saves time when you have mixed table formats and speeds up bulk processing. Less manual mapping, fewer missed rows.
Values in a complex table sometimes change because of feedback from another document. Users saw the change and assumed it was an accuracy bug, but the change was intentional and came from a specific source document. There was no simple way to see which document caused the update.
What we changed
Why it matters
Success signals we’re tracking
Audit download links were valid for 48 hours. That’s more time than we want audit data to remain accessible, and it increases security and compliance risk.
What we changed
Why it matters
How it behaves
Availability: Live
What we changed
Why it matters
Need help or want a walkthrough?
If you’d like these features enabled or want a quick demo for your team, contact hello@staple.ai. We’ll help you get things set up and walk through any changes using your real documents.
We’ve rolled out several exciting updates this month to help teams work more efficiently across reconciliation, master data management, and document processing rules.
Automate data reconciliation across multiple documents and instantly spot mismatches. Staple AI now supports full-cycle reconciliation for enterprises dealing with complex document workflows. From invoices and purchase orders to delivery orders and financial statements, reconcile any document type.
Say goodbye to manual checks. Automate reconciliations 24/7 and ensure compliance with minimal effort. Would you like to automate reconciliation? reach out to us at hello@staple.io.
A next-gen Master Data Management Engine. Map your master data to extracted document fields and export seamlessly to your business systems, all without manual effort.
Simplify vendor, SKU, and cost centre mapping while eliminating redundant uploads. Faster data alignment, fewer mismatches.
New ‘Null’ Condition Check: You can now use a ‘null’ option under label rule conditions to handle blank fields.
For example:If the PO number field is empty, automatically send the document to “non PO invoice” queue for review.
This enables smarter routing logic and ensures incomplete data is always flagged before processing.
Staple AI now supports .bmp (bitmap) file formats for scanning and processing. Upload bitmap images just like PDFs or JPGs and extract data with the same AI accuracy and speed.
We’re continuously improving Staple AI to make automation smarter and simpler. If you have feedback or feature requests, reach out at hello@staple.io.
This month, we’ve rolled out a set of updates to make Staple AI more flexible, secure, and easier to use. From new email integration options to faster screens and improved multilingual support, these enhancements are designed to help your team work more efficiently.
This update improves both flexibilityandsecurity when using Staple AI with email.
You can now disable/enable root account check for the customer. (i.e) you can decide whether to receive documents sent from email ids of Staple AI users only or any email id configured using custom inboxes.
Trigger email notifications automatically to respective users upon success document upload
How to enable this feature?
Enable “Mail document to the uploader” checkbox
We’ve implemented comprehensive multilingual font support for document received via APIs with two predefined font stacks: “default” (Noto Sans) for broad language support including European languages, and “cjk” (Noto Sans CJK) for Chinese, Japanese, and Korean content. This enhancement resolves rendering issues with European character sets that previously displayed missing glyphs
We’ve increased character limits for model field descriptions from 512 to 1024 characters in custom model tagging pages, enabling more descriptive and accurate prompts for better model performance
We have extended our Access Point (AP) capabilities to support Advanced Ordering and SG BIS Order Balance documents, bringing our service in line with IMDA’s latest InvoiceNow requirements. This enhancement enables APs to send and receive five key document types—Order, Order Change, Order Cancellation, Order Response Advanced, and Order Balance while ensuring real-time XML validation, network interoperability across Peppol, and compliance with IMDA’s monthly reporting standards.
These updates strengthen Staple’s position as a trusted AP provider by ensuring readiness for IMDA’s September 30, 2025 compliance deadline and accreditation renewal on April 1, 2026. They also expand workflows beyond invoicing to cover purchase orders and balance tracking, improving B2G onboarding and ecosystem interoperability within Singapore’s nationwide digital invoicing initiative
Minor optimisations were done so that Staple AI screens load faster, improving your team’s efficiency
The digits inside brackets (x) indicate the number of documents in each tab. The number will be immediately updated when the document movement actions are triggered such as moving to different status, moving to another queue, and removing documents.
The system highlights the record that you are about to select in purple. Normally when you click anywhere on the record, the system will display the selected document in the current browser tab.
To open the document in a new browser tab, you can hover your cursor and click the arrow icon “*Open in a new tab”.*Once clicked, a new tab of the select document will appear in your browser.
Alternatively, you can hover your cursor over the file name and press Ctrl + Left click. This approach also opens a new tab of the selected document in your browser.
Also, You can use Ctrl + Left click with several parts on the interface to open a new tab such as
