How it works
One engine. Many sports.
Every gym in AIGym is a different sport — fiduciary judgment, interview craft, negotiation, crisis command — but they share one underlying engine.
Scenarios, not quizzes
Every gym is built on genuinely hard situations — the kind where reasonable, well-informed professionals can land in different places. There is no hidden 'right' answer to pattern-match to.
Your reasoning is the artefact
You pick a move, but more importantly, you write the reasoning. That is what is evaluated. The choice itself is barely scored — the discipline is in the thinking that led to it.
K2 grades on named dimensions
Each gym defines five discipline-specific dimensions. K2 Think V2 grades your reasoning against each, returns a score, and writes a precise comment on where it landed and why.
Repeat, with the gap visible
Over time the same blind spots resurface in your reasoning across scenarios. The gyms make that visible — so you can attack the pattern, not the individual scenario.
Why K2 Think V2
AIGym runs on K2 Think V2, the reasoning model from MBZUAI in Abu Dhabi. K2 is purpose-built for long-form chain-of-thought evaluation — the right tool for grading the structure of an argument, not just whether it produced the expected word. The gyms are built around the model's strengths.