Not every transformation comes with fanfare. Some are built quietly — line by line, model by model — until they change what’s possible.
Like teaching machines to read documents like humans, for example.
That’s what our Head of AI, Andrew Bird, and our Product and Engineering teams have achieved. Under Bird’s guidance, Affinda’s new agentic AI platform lets organizations quickly configure and deploy AI agents that can process any document accurately and at scale. It’s fast to set up, remarkably precise and works for any document type. Not only that, it also improves with every interaction – making it possible for organizations to automate workflows end to end.
Bird has been recognized for his work with a nomination for AI Software Engineer of the Year at the Australian AI Awards 2025, with the winners to be announced in Sydney on November 5, 2025. This nomination recognizes both the technical brilliance and quiet leadership of Bird in building the next generation of the Affinda platform.
So, how did Bird redefine applied AI? And what does that mean for the customers using it?
Rethinking what applied AI can do
The application of AI needed a new mix of deep technical design and practical execution to meet teams who wanted to do more with less – less time spent on manual document processing and more on improving operational workflows with more accurate data.
Traditional machine learning approaches rely on lengthy training cycles, static prompts and constant retraining. They hit ceilings on performance, scalability and cost. So, Affinda took a different path – wrapping an end-to-end platform around LLMs that integrates retrieval-augmented generation (RAG), model memory, agentic workflows, proprietary reading order algorithms, optical character recognition (OCR) and more.
Bird led the shift to a large language model (LLM)-first architecture, combining engineering and data science to create a modular, self-serve design that enables even non-technical users to build and deploy document extraction models on their own. This “model memory” approach brings together three attributes rarely seen in one system:
- Document agnostic – it works for any document type, no matter the format
- Superior accuracy (99%+) – enabling straight-through processing
- Fast configuration – users can set up data extraction models in minutes, not months
The Platform also learns instantly from every interaction, removing the need for massive datasets and lengthy training cycles.
A new AI of possibilities
By designing a platform that’s self-learning, self-serve and endlessly adaptable, Andrew and the team have opened new possibilities for organizations of every size to automate document workflows – without losing control or context.
It’s technology that works with people, not instead of them – and that balance has always been central to Affinda’s philosophy.
For our customers, that means less manual data checking, faster onboarding and automation that scales across every workflow. Whether it’s verifying identity documents in lending, processing invoices in accounts payable or managing complex asset valuations in asset finance, the outcome is the same: time back, fewer errors and teams free to focus on more valuable work.
Transform how your team operates with AI
Bird’s work reflects what makes Affinda different – an AI platform that learns instantly, adapts effortlessly and delivers superior accuracy for every document. If you’d like to automate your document-heavy workflows, talk to our team or give our AI platform a try for free.
-min.png)
