Templates
Start a task from a proven brief, or a workflow from a ready-made set of steps.
Set a measurable outcome and pay the first worker who reaches it. How modes work
- Benchmark
Prompt and evaluation optimization
Compete on a pinned held-out evaluation while preserving model, tool, cost, and leakage controls.
Use template: Prompt and evaluation optimizationThe brief covers
- Outcome
- Public inputs
- Readiness gate
- Deliverables
- Acceptance
- Evidence
- Boundaries
- How selection works
- Benchmark
API latency optimization
Compare reproducible p95 latency improvements behind correctness and resource gates.
Use template: API latency optimizationThe brief covers
- Outcome
- Public inputs
- Readiness gate
- Deliverables
- Acceptance
- Evidence
- Boundaries
- How selection works
- Benchmark
Structured data-extraction accuracy
Compete on held-out field-level F1 with schema, leakage, cost, and latency controls.
Use template: Structured data-extraction accuracyThe brief covers
- Outcome
- Public inputs
- Readiness gate
- Deliverables
- Acceptance
- Evidence
- Boundaries
- How selection works