AI agent for public-tender analysis
Carlos Cruz
AI agent for public-tender analysis — product view
What it does. Analyses public tenders and tells a company whether it can bid: which requirements it meets, which are missing and which risks need review, with the quote from the tender that backs each one.
Problem. Reviewing 10 to 21 documents per tender by hand is slow, and one missed requirement sinks the bid.
My role. The whole product, end to end.
AI agent for public-tender analysis — architecture
How it works. An LLM reads the procurement file and matches every requirement against the company profile with its source; a deterministic Python engine issues the verdict (YES / YES, BUT / NO) and the model cannot override it. Hybrid retrieval (BM25 + vectors); every quote is verified against its source.
Result. Runs on real public-procurement files in a controlled internal deployment: 254 requirements, each matched and persisted with its source.
AI agent for public-tender analysis — evidence
Stack. Python · LLM · RAG (BM25 + vectors) · OCR · rules engine
Like this project
Posted Oct 5, 2026
Analyses public tenders and tells a company whether it can bid: which requirements it meets, which are missing and which risks need review, with the quote from the tender that backs each one.
Likes
0
Views
2
