PProposalBuddy

Documentation

ProposalBuddy

Proposal intelligence for RFPs, grants, and tenders.

01

What it is

ProposalBuddy is a proposal-intelligence workspace. A bid team pastes an RFP or grant document and immediately gets the compliance requirements pulled out as a checklist, AI-drafted responses for each one, and a readiness score that flags missed mandatories.

The problem it solves is small and brutal: RFPs are long and dense, and a single missed shall clause disqualifies the whole bid. ProposalBuddy turns "did we cover everything?" from a manual reading exercise into a structured, scored list.

02

Who it's for

  • Proposal and bid managers at small-to-mid businesses chasing government and enterprise RFPs without a full bid team.
  • Nonprofits and grant writers responding to funding opportunities on tight cycles.
  • Procurement consultants and agencies managing multiple bids on behalf of clients.
03

How it works

The product is one tight loop:

  1. Paste or load an RFP. Drop the full document text into a new RFP, or one-click the sample to see the loop end-to-end.
  2. AI extracts requirements. Gemini reads the document and returns a structured list — every compliance line, deadline, and evaluation criterion, each tagged with a category and a mandatory flag for must / shall / required language.
  3. AI drafts responses. Click Draft response on any requirement (or Draft all) and Gemini writes a concise, professional proposal answer you can edit inline.
  4. Readiness score. A deterministic score tracks the share of requirements drafted, with a separate Gaps list that surfaces mandatory items still uncovered. The goal is simple: close every mandatory gap before submission.
04

The stack

React + Vite on the front end, shadcn/ui and Tailwind for the editorial UI, and Supabase (Postgres + Edge Functions) for storage and the AI boundary. Gemini powers extraction and drafting.

One design decision shapes the AI layer: a single edge function, ai-gateway, routes every task through one model. Routing across multiple models at this scale adds failure surface without buying meaningful quality. The function validates the model's JSON before any database write, retries once on malformed output, and keeps API keys strictly server-side. RFP text is treated as data, never instructions — embedded prompts inside a document cannot rewrite the system behavior.

05

How to use it

  1. Open the Dashboard and click Load Sample RFP, or click New RFP to paste your own.
  2. On the intake screen, fill in the title and paste the document text. Click Analyze RFP. You'll land on the requirements view in a few seconds.
  3. Review the extracted checklist on Requirements. Mandatory items are flagged in clay.
  4. Click Draft response on individual items, or Draft all at the top to fill in every gap. Edit drafts inline and save.
  5. Open Drafts for the assembled proposal — every requirement and its response, ready to copy out.
  6. Watch the readiness score on the Dashboard. Aim for 100% drafted and zero mandatory gaps before submission.

Built for bid teams who don't have time to re-read 60 pages of procurement boilerplate to find the three lines that disqualify them.