Model-guided experiment cycle

Use observations to propose a constrained next experiment batch while preserving each iteration.

collect observations→update model→propose experiments→apply safety constraints→run selected experiments→record observations

Model-guided experiment cycle

Activity responsibilities

Activity	Responsibility
collect-observations	Execute the collect observations stage and publish its declared outputs for downstream activities.
update-model	Execute the update model stage and publish its declared outputs for downstream activities.
propose-experiments	Execute the propose experiments stage and publish its declared outputs for downstream activities.
apply-safety-constraints	Execute the apply safety constraints stage and publish its declared outputs for downstream activities.
run-selected-experiments	Execute the run selected experiments stage and publish its declared outputs for downstream activities.
record-observations	Execute the record observations stage and publish its declared outputs for downstream activities.

Inputs

• Observation history
• Search space
• Safety and budget

Outputs

• Updated model
• Selected experiments
• New observations

Provenance to preserve

• Iteration parent
• Rejected proposals
• Experiment measurements

Runnable example

This is the complete checked-in bundle for this pattern. Download the environment, scope, topology, workflow, input, container recipe, runner, and validation contract from this page before executing it.

Read setup instructions ↓Environment YAML ↓Execution scope YAML ↓Network topology YAML ↓Workflow YAML ↓Input data ↓Container recipe ↓Run the workflow ↓Validate outputs ↓Expected output manifest ↓

Verified local execution

These captures and the output manifest were produced by the fixture's local Docker run and validator. They are published with the same bundle as the runnable files.

Workflow evidence

Execution evidence

Output evidence

Open verified output manifest ↓

Execution considerations

Represent each cycle as a durable iteration with explicit pause, approval, termination, and re-entry policies.

AkôFlow boundary: the engine schedules, deploys, executes, transfers data, and records evidence. The ML or agent framework remains an implementation choice inside each activity.