Skip to content

Draft an Enterprise AI Vendor RFP With Measurable Acceptance Criteria

Structure an AI request for proposal that forces vendors to disclose underlying foundation models, sub-processors, SLA latency, and real pricing tiers.

Copy the prompt and paste it into the AI assistant you already use.

Procurement preparation template only; have corporate legal and infosec teams review final RFP terms before issuance.

Essential sections for an AI-specific vendor RFP

  1. 01

    Demand full model and sub-processor lineage

    Require vendors to list every foundation model provider, hosting region, fallback model, and advance notice window for model swaps.

  2. 02

    Attach a anonymized 10-sample test packet

    Require bidders to return raw unedited outputs and processing times on ten sanitized company records alongside written answers.

  3. 03

    Specify p95 latency and throughput SLAs

    Ask for 95th-percentile response times at peak concurrency rather than best-case averages from lightweight demo environments.

  4. 04

    Separate platform fees from consumption credits

    Force itemized pricing tables covering base seats, sandbox environments, SSO surcharges, API rate limits, and unit overage rates.

Common questions

Should we let AI vendors know which foundation model we prefer in an RFP?

Keep requirements outcome-based first—specify accuracy, latency, and retention rules—while requiring transparency on which models power each workflow step.

How do we stop vendors from hand-editing sample RFP test outputs?

Reserve five unseen validation samples for a live 45-minute screen-share session where finalists run your files through their production interface.

What contractual right protects us if an AI vendor swaps to a cheaper model?

Request notification of material model changes plus a benchmark regression clause allowing remediation or exit if accuracy drops below baseline.

Learn more

Related guides and background articles for this task