Personal framework

Questions to ask before starting an AI project

I wrote a set of questions to help decide whether an AI idea is useful, feasible, and ready for a small trial.

Personal planning framework · Not a delivered transformation program

Start with the business question

I start with the problem, who has it, and how we would know a solution helps. The requirements and test criteria come before choosing an AI model or platform.

Check implementation risk

I ask what the idea depends on, what data it can use, what needs testing, and how to recover if it fails. A useful demo is not proof that a system is ready for production.

Decide who can approve the work

I ask who owns the work, which actions need human approval, what information can be used, and what a reviewer needs to see.

Choose the next practical step

The next step might be a requirements review, a small prototype, a pilot proposal, or a decision to stop. The team would need to agree on the criteria, timing, and definition of success.

Scope and context

• Business value: named problem, intended users, measurable acceptance criteria, and alternatives.
• Feasibility: data access, integration effort, dependencies, operating cost assumptions, and testing.
• Governance: ownership, permissions, information handling, review points, and recovery.
• Pilot scope: boundaries, evidence needed, exit criteria, and a human decision before wider use.

These are planning questions, not employer decisions or a tested scoring method.

See examples of my work

My case studies describe my employer work. My separate personal prototypes let you try checks on synthetic examples.