> ## Documentation Index
> Fetch the complete documentation index at: https://docs.reilabs.org/llms.txt
> Use this file to discover all available pages before exploring further.

# Learning and planning

> Understand how learning, search, and execution work together, and choose the guide for your planning task.

Learning changes what Adapt-1 can predict and which outcomes it expects. Planning compares possible continuations before selecting an action, using learned predictions or values where the task calls for them. The selected action then determines what the system experiences next.

When learning remains enabled, these processes form a continuing loop. Actual outcomes revise the learner, and later plans use the revised state. During frozen use, planning continues while the learned state stays fixed.

## How learning and planning work together

The application presents the current observation and available actions. The planner compares their possible continuations, and the executor carries out the selected action or sequence. The application then observes what actually happened.

Eligible observations and outcomes update the relevant learner. Transition learning improves predictions of action effects; value learning revises estimates of useful outcomes. Procedural acquisition and refinement can also produce reusable action sequences for later selection.

Planning therefore influences learning through the actions selected, while learning influences later planning through retained predictions, values, or skills. Keep imagined continuations distinct from executed experience. If training uses search-derived targets, explicitly define how a planned estimate and an observed outcome form a target at an actually observed state. Search does not automatically turn hypothetical transitions into experience.

## What a plan compares

A planning task needs a current state, available actions, a way to obtain successors, and a way to evaluate them. These parts can be combined in several ways:

* **Exact transitions and learned values:** task rules determine what happens next, while a learned evaluator estimates unfinished states.
* **Learned transitions and a declared goal:** a transition model predicts successive states, and the goal supplies their utility.
* **Learned transitions and learned values:** one model predicts action effects and another scores the resulting states.
* **Retained action sequences:** each candidate is a complete acquired or refined skill, evaluated according to its outcome.

A transition model predicts effects; a value model estimates utility. Exact terminal scores can evaluate completed outcomes without a learned estimate. Record which source actually determined the choice.

Define the execution boundary before choosing a horizon. Some applications execute the first action, observe again, and replan. Others commit to a complete retained sequence. A search step can represent either a primitive action or a whole skill, so depth only makes sense alongside that definition.

## Choose the next step

* [Planning integration](/docs/planning/integration.md): connect actions, successors, and evaluators through HTTP or streaming search, then inspect the returned diagnostics.
* [Planning over learned transitions](/docs/neuroadapt/design-transition-domains.md#planning-over-learned-transitions): configure Domain-query planning, state bindings, goals, and horizons.
* [Combine predictions and values](/docs/neuroadapt/design-transition-domains.md#combine-predictions-and-values): connect separate transition and value learners while tracking their identities and update behavior.
* [Decisions beyond the next move](/docs/neuroadapt/learning-patterns-for-domains.md#decisions-beyond-the-next-move): distinguish planning horizon, learning credit, and the execution boundary.
* [Select from retained sequences](/docs/machina/retained-use.md#select-from-retained-sequences): bind candidates to executable commands and preserve complete-sequence commitment.
* [Interactions during frozen use](/docs/neuroadapt/choose-a-learning-setup.md#interactions-during-frozen-use): keep every participating learner fixed when evaluating retained behavior.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.