Three decisions every AI project should pass
Before you invest
A named business owner wants this problem solved.
The problem is stated in business terms, and today's cost or pain is known.
AI is the right tool. Simpler fixes such as process change, rules or standard software were considered.
The value is estimated and large enough to justify the effort.
There are early signs that the data exists and can be reached.
Ethical, legal and reputational risks are named, and none is a deal-breaker.
Verdict
Answer all six
Before you commit a team
The scope is written down: what is in and what is out.
Success measures have a baseline, a target and an owner.
The data has passed a readiness check.
The team and partners are named, with the skills the work needs.
Privacy, security and access rules for the data are agreed.
The budget covers the work after the pilot, not just the pilot.
Verdict
Answer all six
Before go-live
In testing on realistic data, the solution meets the agreed success measures.
Failure cases, bias and misuse were tested, and the results are recorded.
Users are trained and know when not to trust the output.
Someone owns it in operation, with monitoring for performance and drift.
There is a rollback plan and a clear route for incidents.
Decisions and approvals so far are documented.
Verdict
Answer all six
Start with a conversation
Tell us the decision you are facing. We will tell you plainly whether we can help, and what a useful first step would look like.