Labs & Modeling
Labs & Modeling tests saved work and keeps evidence for a choice. Laboratory runs studies. Models hold goals and releases. Sandboxes keep scenario data apart. Simulations test sealed inputs.
Boundary
Labs & Modeling owns study design, scores, comparisons, and model meaning. Each subject's service owns the Run outcome. A score or ranking does not deploy a Model, change an Agent, or write live Records. Agents & Workflows owns live work. Apps & Dashboards owns the view that presents results. Keep evidence and delivery as separate decisions.
Services
Labs & Modeling opens in the Console at /labs. Choose the service that owns the fact you need to inspect, then follow its links to the related evidence.
| Reader decision | Start with |
|---|---|
| Explain a study, score, or ranking. | Experiments and evaluations connects subjects, Cases, Comparisons, and evidence lineage. |
| Inspect a release or change scenario assumptions. | Models and sandboxes separates reviewed model use from bounded scenario edits. |
| Explain a sealed-input Run. | Simulations connects the frozen release and inputs to state and output. |
Start here
In Labs & Modeling, run and compare a study when a decision needs retained evidence. The Experiment create contract records its pinned revision. Operate simulations to start a Run, check its output, or stop it. Both paths keep failed and cancelled work visible. A later decision can thus account for work that did not succeed.
Next steps
- API reference owns exact HTTP requests, permissions, and responses for Laboratory, Models, Sandboxes, and Simulations. Consult the relevant native Surface guide for availability.
- The Console guide explains access before opening a study or Run.
- The API guide explains calls to the supported study and Run contracts.