# AI consultant hiring checklist Use this worksheet to assess fit, inspect evidence, and record open questions. It is useful whether you choose John Rice, another provider, or decide to handle the next step internally. It does not award a score or certify a provider. Business / team: ____________________ Provider: ____________________ Meeting date: ____________________ Person recording the discussion: ____________________ ## 1. State the help we need Current situation: ____________________ Task or business decision to discuss: ____________________ People who do and review the work: ____________________ Result we want from the engagement: ____________________ - [ ] Identify priorities and a first task - [ ] Train the team around current work - [ ] Test or implement one workflow - [ ] Improve adoption or maintain an existing workflow - [ ] Other: ____________________ What we can discuss without sharing confidential information: ____________________ Questions we need answered before granting access: ____________________ ## 2. Inspect relevant evidence Ask the provider to show permitted work and explain what it establishes. A clearly labeled prototype can be useful evidence of design or technical judgment; it does not prove production adoption or business impact. | Example | Provider's actual role | Live product, prototype, walkthrough, or proposal? | What was tested? | Limits and remaining work | | --- | --- | --- | --- | --- | | | | | | | | | | | | | Questions to ask: - [ ] Which parts did you build, co-build, design, or contribute to? - [ ] What inputs were used, and who reviewed the output? - [ ] Which reported results were observed, and which are forecasts? - [ ] Can you show an incomplete or difficult example as well as a successful one? - [ ] If client work is confidential, is there a permitted public artifact or labeled fictional demonstration? - [ ] How does this example relate to the task we want to buy? Evidence we can inspect again: ____________________ Claims needing clarification: ____________________ ## 3. Ask for a plain description of the work Inputs and permitted sources: ____________________ Expected output: ____________________ Reviewer and approval step: ____________________ Next action after approval: ____________________ Missing or conflicting information behavior: ____________________ Who can pause the workflow: ____________________ Fallback for the underlying task: ____________________ Dependencies that still need checking: ____________________ - [ ] Preparing a draft and taking an action have explicit responsibilities. - [ ] The provider explains our team's required work and decisions. - [ ] The initial boundary and exclusions are understandable. ## 4. Agree on evaluation Representative ordinary example: ____________________ Incomplete or contradictory example: ____________________ Acceptance conditions and reviewer: ____________________ Source checks or required fields: ____________________ Preparation, review, and correction effort to observe: ____________________ How findings change the delivered version: ____________________ Limits that will remain: ____________________ For training, also ask: Audience's starting point: ____________________ Task participants will practice: ____________________ Materials and instructions they receive: ____________________ How they will get help afterward: ____________________ ## 5. Confirm access, accounts, and ownership - [ ] Required information and systems are identified before access is requested. - [ ] An internal owner authorizes and removes temporary access. - [ ] Account administrators and continuing charges are assigned. - [ ] Relevant tool settings and terms are checked for the selected task and account. - [ ] The agreement identifies ownership or licenses for delivered code, instructions, and materials. - [ ] Third-party dependencies and subscriptions are documented. - [ ] The people and suppliers doing the work are identified. - [ ] There is a clear contact for access requests and decisions. Written confirmations we still need: ____________________ Our access owner: ____________________ ## 6. Read the commercial proposal Deliverables: ____________________ Exclusions: ____________________ Our responsibilities and dependencies: ____________________ Fees and payment arrangements: ____________________ Third-party charges paid separately: ____________________ How changes are approved: ____________________ What happens if a dependency is unavailable: ____________________ Acceptance and handoff process: ____________________ ## 7. Plan support and continuation Support includes: ____________________ Support excludes: ____________________ How to report a problem: ____________________ Who maintains instructions, access, and integrations: ____________________ When the work is reviewed again: ____________________ What an internal maintainer or another provider receives: ____________________ How to stop or change tools without losing the underlying process: ____________________ ## 8. Make a decision with the unresolved questions visible Why this provider fits the task: ____________________ Evidence supporting that decision: ____________________ Open question / person responsible / next action: ____________________ What needs to be resolved before approval: ____________________ Possible next step: continue the discussion / request a revised scope / run an agreed test / select another provider / handle internally / stop. Further reading: [NIST AI Risk Management Framework](https://www.nist.gov/itl/ai-risk-management-framework) and [NIST AI RMF Playbook](https://www.nist.gov/itl/ai-risk-management-framework/nist-ai-rmf-playbook). These are voluntary guidance; this checklist is an original buyer worksheet and is not a NIST certification or endorsement. John Rice's introductory AI consultation is free and 30 minutes. Booking it does not commit you to paid work. Bring your questions and a general task description without confidential records or credentials.