AI Proficiency Assessment
How good is your team at this, actually?
Most AI training measures attendance. This measures demonstrated ability on real business work: whether someone verifies a plausible answer, knows when not to reach for a model, and can redesign a process rather than decorate it.
What it measures
Three sections, none of them a questionnaire.
- 01
Six judgment scenarios
Realistic situations where the right answer is often to not use AI, or to verify before acting. Confidence is not what is being measured.
- 02
A live work simulation
Not a chat box. The participant directs an agent with a sandboxed tool over supplied source files, so they can compute an answer instead of asking a model to guess one. That is what makes "did you check that?" a question the assessment can actually observe.
- 03
A workflow redesign challenge
Given a broken process, decide what stays human, what becomes deterministic, and what justifies AI. The same judgment a diagnostic engagement asks for.
Not open yet
It is built. It is not open.
Running it costs real money per participant, and the benchmark is meaningless until enough people have completed it. Both of those get resolved before anyone is invited, rather than after. Leave your details and you get one email when it opens.
Questions
Reasonable things to ask first.
- Is this a quiz about prompts?
- No. Prompt trivia measures whether someone read a blog post. This measures what they do when a model gives them a plausible answer that happens to be wrong.
- Who is it for?
- Teams that have rolled out AI tools and cannot tell whether anything changed, and leaders who need a real baseline before deciding where to invest in training.
- How is it scored?
- A model proposes per dimension scores with cited evidence, and the server computes the score, the level, and the benchmark. The scoring is deterministic and the model never decides the number.
- When does it open?
- When it is ready and its costs are understood, not before. Joining the list gets you one email at that point and nothing else.
Start with one workflow.
Thirty minutes, no deck. Bring a process that costs more than it should and we will tell you honestly whether there is something here worth paying for.