Choose the work.
Then inspect the output.
Every task preserves its brief, conditions, artifacts, failures, personas, versions, time, and cost.
Synthetic evaluation using demonstration outputs. Not real customer demand, adoption, or revenue.
Build a conversion-ready ceramics studio site
Create a responsive local-business site with classes, membership, pricing, and a book-now journey.
02 / Lifecycle emailRecover a stalled free-trial user
Write and render a concise lifecycle email that explains the missing setup step without pressure.
03 / App builderPrototype a calm household budget app
Build onboarding, transaction review, category editing, and a weekly decision summary.
04 / Video + animationCompare motion where continuity breaks
Review source-supplied animation clips for subject identity, camera logic, cut quality, and professional usefulness.
05 / Playable gameWatch whether the first minute earns a replay
Review sustained game footage for control readability, first reward, risk, friction, and visible interaction evidence.
06 / Research agentAssess a new Toronto retail market
Produce a source-led decision brief with claims, contradictions, limitations, and a recommendation.
07 / Structured extractionExtract a 500-page product catalog
Return normalized inventory, variants, price status, source freshness, and failed-page evidence.
08 / Teaching answerExplain why the harmonic series diverges
Give a correct proof, anticipate one misconception, and adapt the explanation for a first-year student.