Choose a workflow

Choose an Adapt-1 workflow when learning from observations and outcomes, or retaining evidence and corrections, can improve a result your application uses. For work centered on interpreting prose or generating it, use a language model for that part of the application. The use cases show where Adapt-1’s learning and retained state contribute to a workflow. The quickstart checks authentication. First learned result contains a complete observation-to-prediction program.

What should Adapt-1 return?

I need to…	Start with	The application supplies
Forecast an observable value	First learned result	Completed observations and a query with the target withheld
Learn which choice works in a context	Learn from an outcome	Decision-time context, the executed policy, and its measured outcome
Keep and correct evidence across calls	Continuity across calls	Useful records, constraints, and corrections
Discover predictive fields and relationships	Use Discovery	Eligible observations, the target, and a stable task boundary
Classify a structured observation	Categorical Discovery example	Pre-result attributes and completed examples with category labels
Discover causal relationships	Causal questions and evidence	Intervention records, measured outcomes, a causal result to evaluate, and a defined experimental protocol
Use later outcomes to revise earlier actions	Sequential Discovery	Ordered transitions, episode IDs, rewards, and terminal status
Acquire or refine numeric command sequences	Machina configuration	Numeric state, goal, action and timing conventions, plus an executor
Track recurring signal intervals	Adaptive policy contract	Ordered observations matching that policy’s contract
Request a spatial judgment or grounded point	Spatial Vision	Inputs accepted by the spatial route

Choose the scope

State gateway

Use prior evidence and constraints without defining a task grammar. The bearer token determines hosted identity.

Domain

Keep a task contract, its observations, outcomes, and learned state together under a domain_id.

Use State when evidence and corrections retained across requests improve subsequent work. Queries use that evidence in the context of the current request, and corrections update records as their content changes. State accepts these records without a Domain declaration. Use a Domain when the application has a defined prediction, decision, or control task to learn and evaluate. For independent results, use an applicable route with the documented non-retaining behavior. Control query writes with the route’s retention setting; allow_exploration controls exploration. See HTTP and contracts.

Define one useful interaction

Name the result

Write what code needs back: a numeric prediction, a selected action, a command sequence, a hypothesis, or an evidence packet. Use the result fields documented for the selected route.

Provide the public task contract

Both Discovery and authored setups use a Domain. Define observation and outcome meanings, result type, available choices when needed, and boundaries. Use Discovery when supported learner-facing structure should form from evidence. Author the structure when it must remain explicit and reviewed.

Complete the return path

Use the result and return the eligible observation or measured outcome. Prediction tasks learn from completed records; control feedback identifies the executed action.

Check the later result

Check learner admission before querying again. Compare the relevant learner diagnostics to determine whether the update was applied.

Configure scope and learning

Choice	Question	More detail
State scope	Which records and task meanings should stay together?	State continuity or Domain design
Domain structure	Which learner-facing structure should form from evidence, and which should be explicit?	Use Discovery or author a Domain
What should improve	Should observations improve predictions, action choices, or decisions with delayed outcomes?	Learning from evidence and outcomes
Learning schedule	Does state start empty or come from prior acquisition, and are writes allowed now?	Choose a learning setup

The learning schedule applies to both Discovery and authored Domains. Either can accumulate experience during use or reuse state acquired earlier with learning frozen.

Copy the Domain worksheet

Domain scope:
Public vocabulary:
Structure to author or discover:
Observed before the result:
Available choices:
Result Adapt-1 should return:
Observable consequence or feedback:
Learning relationship and return route:
Starting learner state:
State retained across items or episodes:
State reset at each episode:
Episode, reset, and evaluation boundary:

A Domain declares the task interface. Keep the correct action, hidden mapping, and withheld target out of the pre-result inputs. Use outcome feedback when the application can measure and attribute a consequence. Evidence-only workflows can use the State gateway.Complete the worksheet with one concrete item or action before implementing the full run. Write the query-time input, the result your code will consume, and the evidence returned afterward. For a structured prediction, specify the target shape and units. For an action, specify its parameters, duration or completion boundary, and the outcome attributed to it.Use multistep action setup or structured scene setup to turn these choices into a learning loop. Choose retention and reset boundaries through Learning schedules.

Use cases

Explore forecasting, causal discovery, adaptive workflows, diagnosis, spatial reasoning, and learned control, with concrete inputs and feedback.

Build with an AI agent

Turn the selected workflow into a bounded implementation and test plan.

Get started

Concepts

Guides

Planning

Discovery

Operations

API reference

What should Adapt-1 return?

Choose the scope

State gateway

Domain

Define one useful interaction

Configure scope and learning

Use cases

Build with an AI agent

​What should Adapt-1 return?

​Choose the scope

State gateway

Domain

​Define one useful interaction

​Configure scope and learning

Use cases

Build with an AI agent

What should Adapt-1 return?

Choose the scope

Define one useful interaction

Configure scope and learning