LogicalDOC 9.3: AI-Powered Document Processing, Object Detection and More

A document management system should do more than keep information organized. The real value begins when the information contained in documents can be identified, extracted, routed, and used with as little manual intervention as possible.

LogicalDOC 9.3 takes that direction further.

This release introduces new capabilities for AI-assisted document processing, extends communication and security through WhatsApp, and improves how information entering the repository is classified and processed.

Rather than adding isolated features, version 9.3 focuses on a common objective: reducing the number of manual steps between receiving a document and turning its content into usable business information.

Object Detection: AI that understands where information is

One of the most significant additions in LogicalDOC 9.3 is Object Detection, a computer vision capability that identifies and locates specific elements within documents.

Traditional image classification answers a relatively simple question: what does this image represent?

Object Detection goes further. It can identify multiple elements inside the same document and determine their exact position.

In a business document, these elements could include an invoice number, a date, customer or supplier information, totals, logos, signatures, or payment details.

LogicalDOC uses this capability to identify relevant sections of a document before text extraction occurs. OCR can then be applied specifically to the detected regions rather than treating the entire page in the same way.

The result is a more targeted approach to extracting structured information from documents.

LogicalDOC supports Object Detection models based on the YOLO (You Only Look Once) architecture. Organizations can train a YOLO model using their own annotated document datasets and import the resulting model into LogicalDOC. Once configured, it can be used to identify relevant regions in new scanned or digital documents.

This makes the technology particularly relevant for documents with recurring visual structures, including:

•  invoices;
• purchase orders;
• forms and applications;
• contracts;
• shipping documents;
• other structured business records.

Detected regions can subsequently be used by OCR engines, AI classifiers, or automation workflows, creating a bridge between visual document recognition and downstream business processes.

From extracted information to document metadata

Object Detection becomes even more useful when the information it identifies can directly contribute to the document's classification.

LogicalDOC 9.3 introduces automatic filling of extended attributes, further reducing the need to manually enter metadata after a document has been captured. The release extends LogicalDOC’s AI capabilities from analyzing document content to transforming it into structured metadata that can be searched, filtered, and used by other processes.

This fits within LogicalDOC’s broader Filler architecture, where document content can be analyzed to automatically populate properties and metadata. Fillers can also be combined into processing chains and applied manually or via folder policies, making document enrichment part of the normal capture process.

For organizations processing large volumes of repetitive business documents, this is an important shift: the document does not simply enter the repository — it can arrive already prepared for the processes that follow.

WhatsApp becomes a LogicalDOC communication channel

Not every important notification is best delivered by email.

With version 9.3, WhatsApp can be integrated with LogicalDOC as an additional communication channel, allowing organizations to reach users through an application that is already part of their everyday workflow.

LogicalDOC can now send notifications through a corporate WhatsApp account, including alerts generated by workflows. The integration uses predefined WhatsApp message templates configured through the Meta Business platform.

This can be particularly useful when users are away from their desks, work across different locations, or need to react quickly to an event generated inside the document management system.

A workflow, for example, can trigger a notification that reaches the user directly on WhatsApp rather than relying exclusively on email or in-system messages.

Two-factor authentication through WhatsApp

The WhatsApp integration also extends to security.

LogicalDOC 9.3 can use WhatsApp to deliver the security code required during two-factor authentication (2FA).

This gives administrators another option when configuring authentication procedures and gives users a familiar channel for receiving their second authentication factor.

The change is relatively simple from the user’s perspective. Still, it reflects an important aspect of enterprise document management: security controls are most effective when they can be integrated naturally into the way people actually work.

By combining notifications and authentication into a single integration, WhatsApp becomes more than just another messaging option. It becomes part of the connection between LogicalDOC and its users.

More control over imported emails

Email remains one of the main entry points for business documents, and LogicalDOC 9.3 also improves how those messages are captured.

Administrators can now assign a custom document template to downloaded emails for individual email import inboxes.

This is especially useful when multiple mailboxes serve different business processes.

Incoming messages from a purchasing mailbox, for example, may require different metadata and protocol rules than those for emails received by a customer service or administrative account. Even incoming and outgoing messages handled through the same IMAP account may need to be classified differently.

Instead of applying the same structure to every imported email, organizations can therefore align email capture more closely with the process to which each mailbox belongs.

It is a relatively focused improvement, but one that can make a considerable difference in environments where email is automatically archived or used as an input for document workflows.

A more consistent approach to document capture

Version 9.3 also continues the consolidation of document-processing operations around Fillers.

The previous scheduled tasks dedicated specifically to Zonal OCR and barcode processing are now deprecated in favor of Fillers.

The change is important beyond the interface itself.

Rather than maintaining separate mechanisms for different types of extraction, LogicalDOC is moving toward a more consistent processing model in which different techniques — including OCR, barcode recognition, and AI-assisted extraction — can participate in configurable document capture processes.

That provides a cleaner foundation for building increasingly automated document flows.

LogicalDOC 9.3: less manual work between capture and action

The most interesting aspect of LogicalDOC 9.3 is not any single feature.

It is how the different improvements connect.

A document can enter the repository, have relevant areas identified through Object Detection, have information extracted and used to populate metadata, and then participate in workflows capable of reaching users through channels such as WhatsApp.

At the same time, security mechanisms such as two-factor authentication become easier to integrate into users’ everyday working environment.

This is the direction behind LogicalDOC 9.3: documents should not simply be stored and found. They should become active sources of information for the processes around them.

For organizations dealing with growing document volumes, the objective is straightforward: spend less time interpreting, classifying, and routing information manually, and make the document management system an increasingly active part of the business workflow.