Skip to content
— Before you are assessed —

Start collecting the evidence
while the work is live

Accreditation at the Institute of AI is assessed on practice. Almost everyone who struggles with that has done the work and simply cannot reconstruct the reasoning a year after the fact.

— Four kinds that count —

What an assessor is reading for

None of this requires a side project or a public portfolio. All four come out of the AI work you are already doing, if you write them down as you go.

Decisions

The calls you made, with the reasoning attached

A short note written at the time, saying what you chose, what you rejected and what you were worried about, is worth more than any artefact produced afterwards.

Failures

The approach that did not work

Assessors read recoveries closely. A clear account of an approach that failed, why you spotted it, and what you did next demonstrates practice that a clean success story cannot.

Measurement

How you knew whether it worked

The evaluation you designed, the outcome it was tied to, and what you did when the numbers disagreed with the expectation in the room.

Responsibility

What you did about risk while it was live

The bias check you ran, the data you refused to use, the escalation you raised. Governance evidence is strongest when it is specific and unflattering.

— Sorting it —

Keep the left column,
lose the right

  • A dated note on why you chose one approach over another
    A polished case study written six months later
  • The evaluation design and what it actually returned
    A dashboard screenshot with no stated baseline
  • Your own contribution, stated plainly
    “We” used across work you did not personally do
  • A problem you argued should not be solved with AI
    A list of the models and frameworks you have tried
— Find the target first —

Know the grade you are collecting for

The free self-assessment tells you which level your practice supports today, which is what turns a vague habit of note-keeping into a deliberate one. Joining the Institute of AI as a Member is free as well.