OpenRFP

Evaluate RFP responses intelligently. Free, open source, transparent.

OpenRFP helps facilities and procurement teams get from a stack of vendor proposals to a defensible ranking. Upload your RFP and OpenRFP builds a weighted evaluation rubric from it. Upload each vendor's response and every proposal is scored against that rubric, criterion by criterion, with a rationale and a quoted passage behind each score. Then compare the field side by side, see where it is too close to call, and export the result for your committee or board.

No account is needed to run a full evaluation. Add an email afterwards only if you want to keep it.

Built by Copeland Building Envelope Consulting as an open-source contribution to the industry. The rubrics it writes are tuned for building envelope and facilities management work: roofing, façade, waterproofing, envelope assessment and repair, and the consulting and contracting proposals that come with them.

How it works

1. Upload your RFP. OpenRFP reads it end to end, works out what you are actually buying, and drafts a rubric: the criteria that matter for this project, a weight for each one, and a written description of every score level so scoring is not a gut feel. Edit any criterion, rewrite a description, or change the weights before you accept it.
2. Upload the proposals. Each response is read against every criterion and scored on the rubric's scale. Every score comes with the reasoning behind it and a direct quote from the proposal, so you can check the evidence yourself. A summary notes where each proposal is strong and where it is thin. Scoring holds the line: 3 out of 5 means adequate, not good.
3. Compare and decide. The comparison screen ranks the proposals by their weighted totals and shows the whole field in one grid. It flags close calls, criteria where two or more vendors scored nearly the same, with a note on what separates them, and drafts the questions worth asking each vendor at interview. Disagree with a score? Override it, and re-rank when you are ready.

What you get

• A rubric you can defend. Weights total 100 percent. Every criterion has a name, a description, and a scoring scale with each level spelled out.
• Scores with receipts. No score appears without a rationale and a quoted passage from the proposal it came from.
• A ranking that knows when it is stale. Change a score after ranking and the comparison says so, tells you whether the order still holds, and offers a re-rank.
• Exports for the people who were not in the room. Download the ranking as a CSV or JSON file, or as a self-contained report you can open in any browser, print, or attach to a recommendation.
• Your work, when you want it. Run everything as a guest. Add an email afterwards only if you want to keep the evaluation, come back to it later, or add proposals to one you already ran.

Transparency

Every instruction used to write a rubric, score a proposal, or rank the field is in this repository, in plain text, in src/lib/prompts/. Each evaluation records which version of those instructions produced it, so a score from last quarter stays interpretable after the instructions change. You can audit any score against the quoted evidence, and you can read exactly how OpenRFP was told to judge.

Your documents

• Formats. Upload PDF or Word (.docx) files up to 25 MB each. Older .doc files should be saved as .docx or exported to PDF first.
• Only the text is read. Drawings, photos and charts do not contribute to a score. Word tables are kept one row per line so pricing and schedules stay legible.
• Scanned PDFs are flagged, not silently scored. If a file has no readable text layer, OpenRFP tells you it may need OCR rather than scoring an empty page.
• Where the text goes. On the hosted app, extracted text is sent to Fireworks AI, a US company, which does not store prompts or outputs and does not use them for training. The evaluation model is an open-weight model served there, and its developer never receives your documents.
• Confidential by default. Your uploads and results are visible only to your session or account. Guest work that is never saved is temporary and may be cleared after a period of inactivity.

What it does not do yet

OpenRFP is in active development. The full flow works, but it has not yet been exercised end to end against a very large real RFP, so treat it as a strong first read rather than a final word.

• No PDF export. Exports are CSV, JSON, and an HTML report.
• Scoring is text-only. Diagrams, photographs and charts in a proposal are not evaluated.
• No rubric templates or saved criteria libraries yet.
• No committee workflows, vendor portals, or contract management. OpenRFP evaluates responses. It is not a procurement platform.

Run your own copy

OpenRFP is free to use. If you would rather host it yourself, everything you need is in docs/DEPLOYING.md: the stack, the accounts to create, the database and email setup, guest limits, and how to swap the evaluation model.

Contributing

Contributions are welcome. CopelandBEC maintains this project. See CONTRIBUTING.md for guidelines.

License

MIT — see LICENSE