Interactive challenges

Production AI judgment you can inspect

These labs are small by design: each one exposes a trace, cost model, recovery decision, or artifact boundary you can inspect quickly.

Guided trace diagnosis

Debug This Agent

Follow the span clues, choose a root cause, then compare your answer with the diagnosis and production fix.

Open challenge

Unit economics model

Cost Anatomy

Toggle designs and see how routing, retries, sandbox reuse, and judge coverage move cost units.

Open challenge

Workflow recovery simulator

DAG Execution Simulator

Step through dependency state, judge failure, recovery choices, and downstream readiness in a lightweight simulator.

Open challenge

Deck artifact inspector

Deck IR Previewer

Edit synthetic Deck IR, inspect validation errors, and switch between preview, outline, and speaker notes.

Open challenge