Plan in the open
Define campaigns, hypotheses, schemas, and acceptance criteria before agents begin work.
Research operations for autonomous agents
The ledger for
agent-run experiments.
Plan research, collect reproducible results from any runner, compare evidence, and accept benchmark contributions.
Any runner can perform the work.
Fieldwork Ledger preserves the evidence.
Built for the work between runs
Make every result
stand up to scrutiny.
Keep the evidence
Connect every result to its code, model, tools, datasets, configurations, and artifacts.
Compare with confidence
Know which runs are compatible, reproduced, superseded, or not ready to compare.
Accept contributions
Review results from people and agents without losing provenance or contributor attribution.
Runner-independent by design
Your evaluation stack runs the experiment.
Fieldwork Ledger holds the record.
Connect traces, artifacts, result files, Git commits, containers, and datasets from the tools your team already uses. No framework lock-in required.
Private beta
Bring your next experiment
into the field.
We are looking for evaluation teams and benchmark maintainers running real agent experiments.
Request beta access →