Command Palette

Search for a command to run...

RFP proposal system

Segment
Procurement
Role
Sole developer at Skillion
Period
07 – 10.2026

An internal system Skillion uses to find, review and bid on public-sector and nonprofit RFPs. An AI-assisted pipeline collects RFPs each month, has AI agents review every one against the company's capabilities, and turns the chosen ones into finished Word and PDF proposals. A private dashboard shows every opportunity, its requirements and deadlines, and how each submitted proposal is doing.

I built both parts and ran the process. Over three months it handled 114 RFPs and produced 7 submitted proposals.

Key features

  1. Monthly RFP intake. Leads from newsletters and lead sheets are collected into monthly batches, with the original documents stored, including scanned PDFs.
  2. AI review of every RFP. AI agents read each RFP in full, check its requirements (certifications, insurance, references, submission rules) against the company's profile, and grade it so the team knows which ones are worth bidding on.
  3. Requirements evidence. Open RFPs are checked for the proof they need, such as case studies or past performance, so the team can prepare it early.
  4. Proposal workspaces. Each RFP being pursued gets its own workspace for requirements, decisions, research, sections, pricing and reviews. An AI agent drafts sections with me, and I make the decisions.
  5. One-command document builds. A command-line tool assembles the sections into a branded Word document and PDF, then checks every page before submission.
  6. Opportunity dashboard. A private dashboard showing each RFP's summary, requirements, deadlines and grade, plus a deadline calendar.
  7. Submission tracking. Each submitted proposal's dates, follow-ups and milestones are tracked, and bad data fails the build.
  8. Secure by default. Access goes through Cloudflare Access, and every deploy is checked so no client names leak into what the browser downloads.

Tech stack

AI
Claude Code and Codex agents for RFP review and proposal drafting
Frontend
Astro 7, React 19, Tailwind v4, shadcn/ui
Backend
Cloudflare Workers, a Node.js CLI for building documents
Data
Git-based records (JSON, Markdown, source PDFs)
Integrations
Pandoc, LibreOffice, Microsoft Word
Infra
Cloudflare Workers, Cloudflare Access, Wrangler, GitHub Actions

← All projects

built with nextjs and cloudflare mindfuk
© 1969 MIT Licensed to fkn no one
view source