Affinda launches conversation tool for document workflows

Affinda has launched Affinda Agent, a conversational tool for configuring document automation workflows. The system is designed to set up an entire workflow through a single natural-language exchange.

The Melbourne-based document processing company says teams can describe their documents, business rules and data destinations in plain language, and the tool will propose settings for each stage of the process. These stages include ingestion, splitting, classification, extraction fields, validation rules, exception handling and integration with more than 2,800 business systems.

Affinda is targeting organisations that handle large volumes of paperwork but have struggled to automate processes because setup has been slow and fragmented. It highlighted sectors including insurance claims, lending, logistics and customer onboarding.

Document processing software has been available for years, but putting it into production has often required technical scoping, developer resources and separate configuration across different parts of the workflow. Affinda argues that this implementation burden, rather than the underlying AI models, has limited wider adoption.

Workflow control

The new tool sits within the Affinda platform as a conversation-based interface. New users are given industry-specific prompts to describe their use case, while existing customers can use the same interface to add document types, set up new fields or extend workflows.

Users can review, alter or reject each proposed configuration before it goes live. Affinda says this allows teams to test whether a workflow meets their needs before committing to a broader rollout.

"For too long, the barrier to document automation hasn't been the AI - it's been the setup. Every other approach still requires technical teams to stitch together the workflow: ingestion, splitting, classification, extraction, validation, integration - each piece configured separately. We've collapsed that entire process into one conversation. That's not an incremental improvement; it's a different category of product," said Andrew Bird, Head of AI, Affinda.

The product is available to all users of the Affinda platform. Affinda says the platform is used by more than 800 customers across more than 80 countries and has processed more than one billion pages.

Customer example

One early user cited by Affinda is Cookie Man, an Australian food brand that used the system to build a workflow for handling purchase orders. According to the company, the workflow extracts order data, checks it against internal business rules, sends it to Cookie Man's ERP system and triggers automated emails to partners.

"Setting up workflows in a conversation with the Affinda Agent has been a real turning point. In an afternoon, I've built a workflow that extracts data from customer purchase orders, validates it against our business rules, pushes it into our ERP and then also sends automated emails notifications to our partners. The accuracy has been excellent, and we're already planning more workflows with the Agent," said Dan Robinson, Supply Chain Manager, Cookie Man.

Affinda says such examples reflect a broader shift in industries where document-heavy workflows have often slowed operations. It argues that customer onboarding, claims processing, lending and logistics have been particularly affected because the work combines varied document formats with strict internal rules and downstream system requirements.

Faster rollout

This complexity has often meant separate tooling and long implementation cycles. Affinda says projects that might appear straightforward have regularly taken six to nine months because teams had to configure ingestion, document separation, classification, extraction, validation and integrations one by one.

"For a lending firm or an insurance team, full document automation has often felt out of reach - too much setup, too much developer time. We're now seeing workflows that once took six to nine months of implementation work configured as a working proof of concept in about 10 minutes. Teams can see it working before they've committed to anything or paid a cent. That's a fundamentally different conversation," said Charlie Bellingham, General Manager - Platform, Affinda.

The launch builds on Affinda's broader platform work around AI agents and persistent model memory, introduced last year as part of its push into custom workflow design without lengthy training cycles. The latest release extends that approach from document analysis to the configuration layer itself.

Most software in the intelligent document processing market still relies on menu-driven interfaces and multi-step setup tools. Affinda is betting that replacing those steps with a guided conversation will shorten deployment times for companies that need to process documents but have limited technical resources.