01
Plan
Frame the question. Record the hypothesis, the inputs, and how the result will be judged.
Collaborative research management for the computational sciences
Know what's working.
Track what's working and what isn't across the whole team — every result traced to the data, code, and analysis that produced it, with the artifacts kept alongside.
Plan
Run
Evaluate
Publish
Trusted by ETH Zurich
The problem
The result is easy to find. Everything that produced it is not.
Generic experiment trackers log metrics. Methodic records the whole decision behind a result — the hypothesis, the inputs, the code, the artifacts, and the attempts that failed — and keeps it searchable for everyone on the project.
Institutional memory that outlasts any one researcher.
Metric ● Status quo ● Methodic
Provenance
Every result traces back to the exact inputs, code, and configuration that produced it.
Institutional memory
Every experiment's design, lineage, and outcome is recorded — including the attempts that didn't work.
Onboarding
New collaborators start from everything the group has already established.
How it works
Go from idea to shared knowledge.
02
Run
Launch on your machine, your cluster, or managed compute. Code, config, and environment are captured as it runs.
03
Evaluate
Compare against the baseline and the work it descends from. Regressions are recorded as plainly as gains.
04
Publish
Commit the result to the shared graph, artifacts and lineage attached. The next question starts from it.
Experiment graph
Every experiment becomes a node that informs future directions.
→ Immutable
Once an experiment is committed, its design is frozen. Once it completes, its outputs (inputs, hyperparameters, metrics, and artifacts) are locked.
→ Graph lineage
Each experiment records the prior work it descends from. Improvements over a baseline are explicit, not narrated after the fact.
→ Searchable
Find every prior result that touched the same dataset, method, or metric — across teams, across years.
→ Auditable
Trace any figure or claim in a write-up back through the analysis and the data that produced it.
Run lineage · turbulence-closure
r-08c1
Reference closure, uncalibrated
eval/rollout-skill: 0.612 · parents: ∅
r-12a4
Calibrated on the flume corpus (v3)
eval/rollout-skill: 0.741 (+0.129) · referenced by 6 downstream studies
r-19f7
Subgrid stress term removed
eval/rollout-skill: 0.689 (−0.052) — regression vs parent.
r-1bd2
Recalibrated across Reynolds regimes
eval/rollout-skill: 0.778 (+0.037) · promoted to group reference.
Example workflow — illustrative until a real pilot is published.
Security & compliance
Built for secure collaboration.
ACCESS
Role-based access control
Org-wide RBAC, IdP-mapped granular roles.
ENCRYPTION
Encrypted at rest and in transit
AES-256 / TLS 1.3, per-tenant keys.
AUDIT
Fully audited actions
Append-only audit log of every action and configuration change.
PROVENANCE
Lineage and auditing
Know the provenance of every artifact and who took each action — agent or human.
FAQ
Frequently asked.
Trackers log metrics from a run. Methodic records the research around it — the hypothesis, the inputs, the code, the artifacts, and the reasoning — so a result can be traced end to end and reused by someone who was not there.