10-year bid-document classification AI
Developed an automatic classifier for accumulated internal bid text, using PyTorch-based LLM fine-tuning for a document collection spanning ten years.
Review the contribution↗SERVICES / AI / AX
Applied AI & workflow transformation
From adding AI to defining work it can actually take on.
- Searching documents and answering work questions
- Bringing classification, analysis and multilingual conversation into products
- Automating recurring work with appropriate human involvement
01SCOPE
A scope that connects
judgment and implementation
01
Use-case strategy and AI user experience
02
RAG, agents, automation and model/API integration
03
Applications, data pipelines and evaluation
04
Fine-tuning, proprietary model training and foundation-model research
05
Deployment, monitoring and ongoing evaluation
Implementation varies by task. Research capability is distinct from project results; accuracy and outcomes depend on the agreed conditions.
02WHAT JUDGMENT CHANGES
The choices define
the conditions of the build
Define sufficient information, the actions AI may take and what a person must review. Choose models against these conditions and actual evaluation.
03RELEVANT EVIDENCE
Review the actual
contribution
YOUR NEXT PROJECT
Let’s make a complex
beginning concrete
A new product, connected systems, or work that needs AI.
Start with what you know today.