Adapt-1 API

Build with a learning system that returns structured results and carries experience into later calls.

Adapt-1 is a non-LLM learning and reasoning system. Send observations, request structured results, and return measured outcomes so later calls can use what it has learned.

Your application can consume predictions, decisions, hypotheses, evidence, and numeric action sequences directly. The API connects that result to a task contract and retained state.

Quickstart

Authenticate and make your first request with cURL, Python, or JavaScript.

First learned result

Create a Domain, send observations, and query the state they form.

Explore application ideas

Support request routing, tool selection, troubleshooting, process monitoring, personalized software, and interactive control are possible directions. The application ideas page explains what Adapt-1 could contribute, what your application needs to supply, and which guide to start with.

Match a result to an interface

Your application needs	Start here
A prediction from measurements and prior observations	Predict an observable output
Decisions that learn from their consequences	Learn from an outcome
Evidence and constraints that carry across calls	Maintain continuity
Useful fields and relationships discovered from experience	Use Discovery
Acquired and revised numeric action sequences	Configure Machina

Combine these workflows when their inputs, outcomes, and state scopes are compatible. Choose a workflow compares their contracts.

How the loop works

A predictor learns from completed observations. An action policy learns from attributable outcomes. Your application supplies the available inputs, validates and uses the returned result, and records what followed.

Your application supplies	Adapt-1 provides
Task meanings, observations, and available actions	Contract-defined predictions, selections, or evidence
An executor and measured consequences	Learning from eligible events and outcomes
State boundaries and result checks	Retained state and supporting diagnostics for later calls

Choose the right API surface

Surface	Use it for
Domains	Task-scoped events, predictions, policies, hypotheses, feedback, and explanations
State	Continuing evidence, constraints, retrieval, reasoning, and correction
Numeric trajectories	Sequence acquisition, structural revision, contextual correction, and retained execution
Interval policies	The documented ordered interval/scalar observation and prediction contract
Spatial Vision	Spatial judgments and grounded pointing

These are access paths into Adapt-1. Each has its own request and response contract. Start with Core concepts for Domains, events, learning, and state scope, or HTTP and contracts for authentication and route behavior.

Build with an AI agent

Give your coding agent the relevant contracts and a complete implementation brief.

Read Core concepts for terminology and implementation boundaries.