What you’ll need before you start
- The RFP Request (requirements + instructions) as a file.
- The Vendor Response as a file.
- The Grading Criteria / Rubric (weights + definitions) (Optional).
- The Required Questions list (Optional).
How to run it (5–10 minutes):
- Open your approved AI workspace.
- Start a new chat, title it with the RFP name, and upload the two documents (RFP + Vendor Response).
- Paste the rubric into the correct spot in the prompt and add the required questions into the correct section of the prompt.
- Now paste the prompt (verbatim) from the next section.
- Send.
- When you receive results, skim the Scoring Matrix, Gap Analysis, Risk & Strengths, and Reviewer Q&A.
- If anything looks off, ask the model to “re‑score using stricter evidence standards” or “show only citations containing SLAs/time‑bound commitments.”
What you’ll get back:
- A Scoring Matrix with scores, justifications, and citations to the vendor’s own words.
- A Traceability table mapping each requirement to the response.
- A Gap Analysis (what’s missing/weak).
- A Risk & Strengths summary.
- A Reviewer Q&A set you can use in meetings or vendor follow‑ups.
Copy‑and‑Paste Prompt (verbatim)
Paste the text below into your AI workspace. Do not edit inside the box unless your team has agreed to a standard variant.
How the prompt “thinks” (in plain English)
- It parses first, judges second. That keeps us honest.
- It maps requirement‑to‑response so we can trace every claim to vendor text.
- It scores with receipts. Justifications must point to quotes, metrics, or artifacts.
- It flags gaps and risks (e.g., “after‑hours support via escalation only”).
- It drafts reviewer questions that close gaps before we sign anything.
A tiny example (from the prompt)
Requirement: “Provide 24/7 customer support.” Response: “Business hours, with emergency escalation after hours.” Evaluation: Partial (2/5). Risk: Limited after‑hours coverage. Reviewer Q: “What is the SLA for after‑hours support?”
Guardrails & good habits
- Use approved workspaces and follow data‑handling policies. Redact sensitive details when needed.
- Name your files clearly (
RFP-<project>-vFinal.pdf,Vendor-Response-<company>-v1.docx). - Ask for stricter evidence if the scoring feels generous. “Only cite lines with numbers, SLAs, or certifications.”
- Save what works in a shared team folder so our approach becomes a paved road—not a goat trail.
For leaders: make it the paved road
Publish three decisions and you’ll boost quality overnight:
- When to use the prompt (e.g., any RFP over $X or any regulated domain).
- How to store outputs (Scoring Matrix, Gap Analysis, Q&A) for auditability.
- Who signs off before awards go out and keeps a human in the loop.
This isn’t about replacing human judgment; it’s about aiming it. We free people to do the human parts—listening, deciding, caring—by giving them consistent scaffolding for the rest.
