Action
An action is a choice or operation selected or performed by an actor that can affect a system, environment, or process.
Grokking Simplicity
An action is a computation whose observable result depends on when or how it is performed and may involve effects beyond producing a value.
- How it is used
- Uses actions to locate time-dependent work and decide where functional techniques should concentrate attention.
- How deep it goes
- Defines actions, identifies common forms, traces their spread, and stops short of detailed safety techniques for managing them.
- Examples
- Uses UI events, database operations, email delivery, coupon campaigns, affiliate payouts, shopping, and JavaScript expressions.
- What you can do after
- Diagnose actions in new and existing code, separate calculations from them, and identify where timing and repetition require care.
Applied Reinforcement Learning
An action is a choice made by an agent that can affect the subsequent state of its environment.
- How it is used
- Frames actions as controllable choices in sequential business decisions and language generation.
- How deep it goes
- Defines actions and their effects on rewards and transitions, then works through business and language-model mappings.
- Examples
- Illustrates actions through daily pricing, staffing, ad selection, and next-token generation.
- What you can do after
- Builds a Python ad-campaign simulation from scratch to compare action-selection strategies.
Optimization Algorithms: AI techniques for design, planning, and control problems
An action is a choice made by a reinforcement-learning agent in a given state that affects subsequent states and rewards.
- How it is used
- Frames sequential decisions for reinforcement learning and combinatorial optimization.
- How deep it goes
- Defines action sets and policies, contrasts state transitions with bandits, and illustrates action-selection strategies in simulation.
- Examples
- Illustrates choices with uncertain train and car commutes and a three-machine slot simulation.
- What you can do after
- Evaluate action choices under uncertain outcomes and compare exploration and exploitation in a bandit setting.
Authorization in Action
An action is an operation that a principal attempts to perform on a resource.
- How it is used
- Uses actions to identify the operation being authorized within a principal, action, resource, and conditions decision.
- How deep it goes
- Defines actions, compares policy representations, and traces how an attempted operation becomes an evaluation request.
- Examples
- Illustrates read versus write permissions on a shared folder and a document-sharing request.
- What you can do after
- Apply it via Cedar by mapping a protected operation to an action in an authorization request.
AI Agents in Action
An agent action is an operation an agent selects or performs to advance a task.
- How it is used
- Extends agents with actions for completing tasks, exploring, and communicating.
- How deep it goes
- Defines action targets, effects, and generation methods, then demonstrates configuration and execution through practical workflows.
- Examples
- Uses a daily-task service, an image-inspection skill, and a file-saving function to illustrate actions.
- What you can do after
- Apply it via FastAPI and AutoGen Studio to connect a service or configure an image-inspection skill.
Build a Robo-Advisor with Python (From Scratch)
An action is a decision selected by an agent that can change the state of its environment.
- How it is used
- Frames portfolio allocation as an agent decision for maximizing the probability of reaching a financial goal.
- How deep it goes
- Defines actions, then implements discrete action search through transition probabilities and backward dynamic programming.
- Examples
- Uses stock-fund versus risk-free-bond allocations and tests 21 discrete allocation choices at each wealth-time state.
- What you can do after
- Apply it via dynamic programming to select state-dependent asset allocations that maximize expected goal attainment.
Read how Build a Robo-Advisor with Python (From Scratch) explains action →
Grokking AI Algorithms, Second Edition
An action is a choice or operation that an agent performs in an environment to change its situation or obtain an outcome.
- How it is used
- Uses actions as the choices an agent learns to make toward a known goal when the best sequence is unknown.
- How deep it goes
- Traces actions through a simulated environment, including allowed moves, state updates, feedback, and performance measures.
- Examples
- Illustrates actions with a dog learning to sit and a self-driving car moving through a parking lot.
- What you can do after
- Evaluate action sequences using cumulative rewards, penalties, and average reward per action.
Read how Grokking AI Algorithms, Second Edition explains action →
Designing AI Agents
An agent capability for executing external operations or producing artifacts in response to reasoning.
- How it is used
- Turns reasoning proposals into external interactions or structured outputs within an agent’s perception-reasoning-action loop.
- How deep it goes
- Defines action contracts and summarizes dispatch, chaining, planning, and guardrail controls, including acceptance evidence and recovery concerns.
- Examples
- Uses map-layer publication, duplicate payroll transfers, and a structured code-review output to illustrate action and its failure modes.
- What you can do after
- Evaluate action boundaries and completion evidence, and choose controls for failures such as duplicate calls, stale state, and missing receipts.
Practical Automation with PowerShell: Effective scripting from the console to the cloud
An automation action is an individual operation or step performed during an automated process to achieve its goal.
- How it is used
- Frames automated work as goal-directed steps initiated by triggers and considered alongside maintainability.
- How deep it goes
- Defines actions and classifies them into logic, tasks, and logging with concrete control-flow and operational steps.
- Examples
- Uses a web-server log cleanup that archives old files, copies them to Azure Blob Storage, verifies transfer, and records results.
- What you can do after
- Evaluate/understand it by classifying workflow steps as logic, endpoint tasks, or logging.
The Design of Web APIs, Second Edition
An action is the intended effect that an operation applies to a resource, expressed by the main verb in the operation description.
- How it is used
- Identifies what an operation does to its resource before that operation is represented with HTTP.
- How deep it goes
- Defines actions by main verbs, works through catalog operations, and defers HTTP-method selection and detailed data modeling.
- Examples
- Uses product-catalog operations, a GET /products search, a POST-versus-PUT creation choice, and a course-platform identification exercise.
- What you can do after
- Practice/perform action identification from operation verbs, then use the result to choose an appropriate HTTP method.
Read how The Design of Web APIs, Second Edition explains action →
AI Agents in Action, Second Edition
An action is an operation an AI agent performs that can affect its environment or internal state.
- How it is used
- Deploys actions as the mechanism connecting agent goals, planning, tools, state, evaluation, and task completion.
- How deep it goes
- Explains action categories, tool boundaries, deployment choices, and loop execution, but stops short of model internals.
- Examples
- Uses context retrieval, get_research_sources(), MCP servers, and the repeated sense-plan-act-learn loop as concrete examples.
- What you can do after
- Apply it via tool calls and agent loops while choosing boundaries, validating results, and defining termination conditions.
Read how AI Agents in Action, Second Edition explains action →
Experimentation for Engineers
A candidate choice made by a contextual bandit, such as displaying one particular post.
- How it is used
- Frames post selection as the decision made by a contextual bandit in a production app.
- How deep it goes
- Defines action through post-display examples and logging, but stops short of detailed action-selection policies.
- Examples
- Uses displaying one post among many and logging context, action, and reward as concrete examples.
Timeless Algorithms, The Foundational Ideas
A choice available to a decision-maker in a decision problem.
- How it is used
- Represents the choices a decision-maker can control when evaluating consequences under uncertainty.
- How deep it goes
- Defines feasible acts and their relation to states and experiments, then connects action evaluation to expected utility.
- Examples
- Uses treatment choices under diagnostic uncertainty and ordering inventory amid uncertain demand.
- What you can do after
- Apply expected utility to compare actions whose consequences depend on uncertain states.
Read how Timeless Algorithms, The Foundational Ideas explains action →
Refactoring to Rust
An action is an event or command describing a state transition that a component should process.
- How it is used
- Coordinates user events, asynchronous results, and state changes inside a Yew component.
- How deep it goes
- Defines actions as enums and traces their handling through update, stopping short of broader action-system design.
- Examples
- Uses pagination, feed-fetch states, button callbacks, and asynchronous search results as concrete action examples.
- What you can do after
- Build it from scratch by defining message variants and implementing update logic for component events and asynchronous results.
Causal AI
An action is a decision or operation selected by an agent or performed by an individual in a given situation.
- How it is used
- Frames decisions as causal interventions or agent operations within sequential environments.
- How deep it goes
- Explains actions through causal hierarchy, state changes, availability constraints, delayed feedback, and Markov decision processes.
- Examples
- Uses investment choices, chronic-pain medication, and grid-world movement to illustrate causal and sequential action.
- What you can do after
- Apply action modeling to distinguish observed choices from interventions and represent agent behavior within an MDP.
Natural Language Processing in Action, Second Edition
A bot action is an executable response or operation that a conversational system performs after processing user input.
- How it is used
- Uses executable responses and operations to implement multistep conversational flows in Rasa.
- How deep it goes
- Explains predefined and Python actions, external integrations, and YAML story wiring.
- Examples
- Demonstrates greetings, permission prompts, confirmations, and an onboarding story for a math tutor chatbot.
- What you can do after
- Apply it via Rasa by configuring response actions and sequencing them with intents in stories.
Read how Natural Language Processing in Action, Second Edition explains action →
Build a DeepSeek Model (From Scratch)
An action is a choice made by an agent at a decision point that can change the state of an environment.
- How it is used
- Frames action as the selectable unit in agent-environment interaction and token-by-token reinforcement-learning rollouts.
- How deep it goes
- Explains action through state transitions, stochastic sampling, trajectories, and a basic policy-gradient equation, while deferring implementation details.
- Examples
- Uses a cleaning robot, sampled language-model tokens, and a tennis-ball arithmetic response as concrete action examples.
- What you can do after
- Apply action sampling to construct language-model rollouts and relate sampled-token probabilities to policy-gradient updates.
Read how Build a DeepSeek Model (From Scratch) explains action →
Also covered in
- Robotics for ProgrammersIn early access (MEAP)
- Effective Behavior-Driven DevelopmentIn early access (MEAP)
- Reinforcement Learning from Human Feedback
- AI Agents for Offensive SecurityIn early access (MEAP)
- Transformers in Action
- How Large Language Models Work
- Build AI-Enhanced Web Apps
- Architecture Modernization: Socio-technical alignment of software, strategy, and structure
- Evolutionary Deep Learning: Genetic algorithms and neural networks
- Architecting for AutonomyIn early access (MEAP)
- Bayesian Optimization in Action
What to know first
- language modelA language model is a computational model that represents patterns in linguistic sequences to assign probabilities, predict, generate, or evaluate language.
- reinforcement learningReinforcement learning is a machine-learning paradigm in which an agent learns behavior or a decision policy through interaction with an environment and evaluative feedback, such as rewards.
- stateState is information describing the condition or configuration of a system or computation at a particular time that may influence its behavior or subsequent transitions.
- environmentAn environment is the external context, system, or conditions within which an agent or application operates and interacts.
Related concepts
- reasoningused by · Reasoning is the process of deriving conclusions, solutions, or actions from available information through inferential steps.
- examplepart of · An example is a concrete instance that illustrates expected behavior, a possible outcome, or an input-output relationship.
- interventioncontrasts with · An intervention is a deliberate, externally imposed change or control applied to a variable or system to alter or assess its behavior or outcomes.
- multi-armed banditproduced by · A multi-armed bandit is a sequential decision problem in which an agent repeatedly chooses among alternatives with uncertain rewards, balancing reward seeking with information gathering.
- PARC modelpart of · The PARC model is an authorization framework that describes an access decision using a principal, action, resource, and conditions or context.
- recommendationincludes · A recommendation is a suggested item, choice, or course of action offered to guide a decision.
Questions about action
What is action?
An action is a choice or operation selected or performed by an actor that can affect a system, environment, or process.
Which Manning books explain action?
Grokking Simplicity, Applied Reinforcement Learning, Optimization Algorithms: AI techniques for design, planning, and control problems, Authorization in Action, AI Agents in Action, Build a Robo-Advisor with Python (From Scratch), Grokking AI Algorithms, Second Edition, Designing AI Agents and 9 more.
What should I know before learning action?
Manning books that teach action usually introduce language model, reinforcement learning, state and environment first.
What is the difference between action and intervention?
An action is a choice or operation selected or performed by an actor that can affect a system, environment, or process. An intervention is a deliberate, externally imposed change or control applied to a variable or system to alter or assess its behavior or outcomes.
What kinds of action are there?
Kinds of action include recommendation.
What is action part of?
Action is part of example. An example is a concrete instance that illustrates expected behavior, a possible outcome, or an input-output relationship.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.
















