Tender / RFQ Finder + Bid Drafter

Scans Kenyan tender boards for the RFQs a business can actually win, then drafts the response.

The problem

Kenyan SMEs and contractors read tenders.go.ke and dozens of county portals by hand every week. Promising RFQs slip past unnoticed, deadlines get missed in the noise, and every bid restarts the same boilerplate from a blank page.

What the agent does

  1. 1Fetch the open tenders from Kenya's national procurement portal — 500+ live per run — with graceful degradation when a source is blocked or stale.
  2. 2Score every tender 0–5 on five axes: capability fit, deadline feasibility, buyer credibility, value attractiveness and competition. Deterministic rubric — same input, same scores, and every number traces back to a real source URL.
  3. 3Flag the top picks and write a first-draft bid response for each: company profile, capability mapping to the tender's requirements, compliance checklist, pricing placeholder.
  4. 4Ship one ranked report, deadline-aware and ready to work from.

Agent architecture

The big picture of how the agent is put together — the open-source models behind each mode, the kinds of services it talks to, and how work and memory flow. The specific tool wiring and internal API structure stay internal by design.

Models

  • Deterministic Python pipeline — no model in the scoring path

Tools

  • National tender-portal API client
  • Rate-limited HTTP with 24h fetch cache and retry/backoff

Memory

24h fetch cache + per-run report history

Routing

One command: fetch → parse → score → ranked report + bid drafts