A proposal team for firms that don't have one

The problem

Federal RFPs are long, dense, and written for firms with proposal teams. A small contractor reads the same documents with a fraction of the people, so they bid on fewer contracts and lose more of the ones they bid on. The opportunity is real; the reading is the bottleneck.

What it does

A bidder uploads the solicitation. GovHub splits the PDF into chunks, runs Claude over each one to pull out structured answers, and assembles an eight-tab Intelligence Brief: the requirements, the risks, and what a winning response needs to cover, in a form a small team can act on.

Decisions

• Chunk before you ask. Long solicitations do not fit in one pass, and asking a model about a whole RFP at once invites vague answers. Splitting the PDF first and extracting from each piece keeps every answer tied to the text it came from.
• Structure over prose. Extraction fills a fixed schema rather than writing a summary, so the brief looks the same for every RFP and a bid team learns where to look.
• Edge first. Cloudflare Workers keep the processing close to the user and cheap to run, with Supabase for accounts and saved work.
• Three paid tiers. Pricing is part of the product test, not an afterthought.

Built by one person

I built GovHub alone in about four to five months Reported, with AI tools doing much of the typing and me doing the deciding. It is the same loop I write about in the Enterprise AI Delivery Loop: clear intent, human accountability, and evidence at each step, at the scale of one.