A little while back, I posted about an experiment: running the Microsoft Agent Framework (MAF) entirely offline on an Amazon Web Services (AWS) EC2 instance, paired with Ollama and a Docling MCP server. The goal was simple - test whether MAF could operate in a fully contained environment with no external API calls. It worked.
That experiment became the foundation for something bigger: an RFP Agent Factory.
Over the past while, I’ve built (with help from Claude) two working proof-of-concept agents - one for RFP response generation and one for RFP response assessment.
Both share a common architecture:
● Offline MAF agents powered by Ollama (Qwen3 8B in my case)
● Flexible ingestion via Docling MCP (PDF/Docx conversion)
● Lightweight RAG using ChromaDB
● Automated outputs written directly to Word (.docx) - including reasoning
● No human intervention required - the agents can wake up, process whatever files are in a folder, produce the draft output, and go back to sleep (i.e. cron job wrapper).
These PoCs aren’t polished (yet), but they work - and more importantly, they can scale.
Why bother with an “RFP Agent factory”?
Because two of the most expensive, time-consuming (and frankly, painful) processes in many organisations are (I’ve been on both sides of these):
● RFP preparation (sales and pursuit teams burn hours aligning capabilities to requirements)
● RFP assessment (procurement teams and downstream evaluators spend days scoring responses)
(Repo link https://github.com/steveh250/MAF-RFP-Factory)
So many opportunities for future development.