# AIVAX Documentation > Build, operate, and evaluate AI applications with AIVAX. Every documentation page is also available as Markdown by replacing .html with .md. Language: English. - Complete documentation: https://docs.aivax.net/llms-full.txt - Public API reference: https://inference.aivax.net/apidocs/llms.txt Embedded API references are represented by endpoint links in Markdown; their remote content is not copied into this static site. ## Introduction - [Overview](https://docs.aivax.net/docs/overview.md): Overview AIVAX is an AI orchestration platform for building, operating, and evaluating AI applications through one account, API surface, and billing wallet. It … - [Getting Started](https://docs.aivax.net/docs/getting-started.md): Getting Started This guide takes you from an AIVAX account to a verified OpenAI-compatible chat completion. The example uses Python and a private API key from a … - [Authentication](https://docs.aivax.net/docs/authentication.md): Authentication AIVAX authenticates API requests with account API keys. AIVAX accepts API keys via: Authorization: Bearer Authorization: Basic … - [Pricing](https://docs.aivax.net/docs/pricing.md): Pricing Service usage prices are listed below in USD. M means one million tokens; 1k means one thousand units. Approximate prices (~) vary with the model used … - [Plans and Limits](https://docs.aivax.net/docs/limits.md): Plans and Limits AIVAX has three account plans: Free, Pro, and Max. The current plan is stored on the account and controls model access, commissions, rate … - [Data Collecting](https://docs.aivax.net/docs/data-collecting.md): Data Collecting AIVAX offers an optional semantic data collection program for accounts that choose to contribute eligible RAG and reranking data to model … - [Changelogs](https://docs.aivax.net/docs/changelogs.md): Changelogs Technical changes that affect AIVAX products, services, or the public API. Dates identify when entries were added or updated, not confirmed … ## RAG and collections - [Collections and Documents](https://docs.aivax.net/docs/rag/collections.md): Collections and Documents AIVAX provides a RAG (Retrieval-Augmented Generation) service for storing documents and retrieving them later through semantic search. … - [Media Injector](https://docs.aivax.net/docs/rag/media-injector.md): Media Injector Media Injector turns a source file into focused, self-contained documents inside an AIVAX RAG collection. It examines the source, identifies … - [Semantic Search](https://docs.aivax.net/docs/rag/semantic-search.md): Semantic Search The semantic search API searches one or more collections and returns the most relevant indexed documents for the supplied search terms. If your … - [Text Segmentation](https://docs.aivax.net/docs/rag/text-segmentation.md): Text Segmentation Text segmentation divides source documents into semantically cohesive strings that can be embedded or indexed in a RAG collection. It returns … - [Text Classification](https://docs.aivax.net/docs/rag/classification.md): Text Classification Use text classification to rank a fixed set of labels for one or more documents without training a custom classifier. AIVAX embeds every … - [Rerankers](https://docs.aivax.net/docs/rag/reranking.md): Rerankers Rerankers reorder an existing set of candidate documents for a query. They do not search a collection or recover text that is absent from the input. … - [Reflex](https://docs.aivax.net/docs/rag/reflex.md): Reflex Reflex is AIVAX’s collection-less search for RAG. Send a query together with candidate document strings and receive the most relevant items in ranked … - [Best Practices for RAG](https://docs.aivax.net/docs/rag/best-practices.md): Best Practices for RAG Use this guide when preparing documents for AIVAX collections and semantic search. The search pipeline indexes document text as … ## Filters - [Document Filters](https://docs.aivax.net/docs/filters/document-filters.md): Document Filters A document filter restricts a RAG search to the documents that match a condition, such as a tag, a metadata value, or a date range. Only … ## Inference - [AI Gateway](https://docs.aivax.net/docs/inference/ai-gateway.md): AI Gateway An AI Gateway stores a reusable inference configuration. Pass the gateway’s ID or slug in the request’s model field, and AIVAX applies its model … - [Inference](https://docs.aivax.net/docs/inference/inference.md): Inference AIVAX exposes an OpenAI-compatible chat/completions API with additional AIVAX parameters. The additions are optional and are designed to support … - [Agentic Tests](https://docs.aivax.net/docs/inference/agentic-tests.md): Agentic Tests Agentic Tests evaluates how an AI Gateway behaves across a complete, goal-oriented conversation instead of grading one isolated response. AIVAX … - [Voice Session](https://docs.aivax.net/docs/inference/voice-session.md): Voice Session Voice Session is AIVAX’s low-latency, stateful voice API. It keeps the authenticated AIVAX WebSocket endpoint while connecting to a realtime … - [AI Pipelines](https://docs.aivax.net/docs/inference/pipelines.md): AI Pipelines AI Gateway pipelines are the processing steps AIVAX applies before and during inference. They can add context, rewrite queries, expose tools, … - [Structured Responses](https://docs.aivax.net/docs/inference/structured-responses.md): Structured Responses AIVAX can produce structured JSON through two paths: response_schema: AIVAX validates the final model output against a JSON Schema and … - [AI Workers](https://docs.aivax.net/docs/inference/workers.md): AI Workers AI Gateway workers are HTTP hooks that let an external service control gateway execution at runtime. A worker can allow an event, stop it, rewrite … ## Web Foundation - [Web Search](https://docs.aivax.net/docs/web-foundation/web-search.md): Web Search Web Search retrieves current information from the internet for research, fact checking, and answers that need sources beyond the model’s training … - [Fetch and OCR](https://docs.aivax.net/docs/web-foundation/fetch-and-ocr.md): Fetch and OCR Fetch and OCR extracts readable text from web pages and supported documents so applications and agents can use their content for summaries, … ## Generations - [Semantic decisions](https://docs.aivax.net/docs/generations/decisions.md): Semantic decisions Semantic decisions evaluate named questions against a shared state and return structured answers rather than a generated explanation. Use … - [Speech Generation](https://docs.aivax.net/docs/generations/speech.md): Speech Generation Use Speech Generation when your application already has final text and needs playable audio without running a chat completion. Typical uses … - [Audio Transcriptions](https://docs.aivax.net/docs/generations/audio-transcriptions.md): Audio Transcriptions Use Audio Transcriptions to convert recorded speech into text for search, review, captions, or downstream automation. Typical uses include … - [Media Descriptions](https://docs.aivax.net/docs/generations/media-descriptions.md): Media Descriptions Use Media Descriptions when an application needs structured information from audio, images, video, or PDF content. Typical uses include … - [Teach Skill](https://docs.aivax.net/docs/generations/teach-skill.md): Teach Skill Use Teach Skill to turn recorded demonstrations into reusable, step-by-step skill instructions. Typical uses include capturing an expert’s screen … - [Image Generation](https://docs.aivax.net/docs/generations/images.md): Image Generation Use Image Generation to create images from a text prompt in an application workflow. Typical uses include draft illustrations for editorial … ## Resources and features - [Skills](https://docs.aivax.net/docs/features/skills.md): Skills Skills (also known as abilities) can be used to improve how your agent performs on specific tasks. Skills are special instructions that are retrieved on … - [Chat Clients](https://docs.aivax.net/docs/features/chat-clients.md): Chat Clients A chat client provides a user interface via an AI Gateway that allows the user to converse with their assistant. A chat client is integrated with … - [Batch](https://docs.aivax.net/docs/features/batch.md): Batch Batch is AIVAX’s feature for running the same AI workflow over many independent items. It turns a list of inputs into a background-processed queue with … ## Tools - [Built-in Tools](https://docs.aivax.net/docs/tools/builtin-tools.md): Built-in Tools AIVAX provides a list of built-in tools for you to enable in your model. These tools can be used together with the server-side functions. Some … - [Support for Model Context Protocol (MCP)](https://docs.aivax.net/docs/tools/mcp.md): Support for Model Context Protocol (MCP) You can bind external MCP protocol tools to your AI Gateway. The protocol defines tools that run on the server side and … - [Protocol Functions](https://docs.aivax.net/docs/tools/protocol-functions.md): Protocol Functions Protocol functions are AIVAX server-side tools. They let a model request a named action while AIVAX executes the action from the server by … - [Shell](https://docs.aivax.net/docs/tools/shell.md): Shell AIVAX offers a virtual shell environment that can be used by agent assistants to execute terminal commands during inference. This feature is especially … ## MCP Utilities - [Collections MCP](https://docs.aivax.net/docs/mcp-utilities/collections-mcp.md): Collections MCP The Collections MCP exposes one or more AIVAX RAG collections as tools for compatible MCP clients. Use it when an external model, agent, IDE, or … - [Documentation MCP](https://docs.aivax.net/docs/mcp-utilities/documentation-mcp.md): Documentation MCP The AIVAX documentation MCP exposes AIVAX documentation, API reference content, and model metadata to MCP-compatible clients. It is designed … - [Account management MCP](https://docs.aivax.net/docs/mcp-utilities/account-management-mcp.md): Account management MCP The account management MCP exposes selected AIVAX account operations to an MCP-compatible client, such as an IDE, desktop assistant, … - [Web utilities MCP](https://docs.aivax.net/docs/mcp-utilities/web-utilities-mcp.md): Web utilities MCP The web utilities MCP exposes AIVAX web retrieval tools to any MCP-compatible client. Use it when an agent, IDE, desktop assistant, or … - [Media generation MCP](https://docs.aivax.net/docs/mcp-utilities/media-generation-mcp.md): Media generation MCP The media generation MCP exposes AIVAX image generation and speech generation (text-to-speech) to any MCP-compatible client. Use it when an … - [Inference MCP](https://docs.aivax.net/docs/mcp-utilities/inference-mcp.md): Inference MCP The Inference MCP exposes an AIVAX integrated model or AI Gateway as a tool for compatible MCP clients. Use it when another model, agent, IDE, or … ## Legal documents - [Privacy Policy](https://docs.aivax.net/docs/legal/privacy-policy.md): Privacy Policy Revision: 1.4 Effective date: August 1, 2026 Previous update: July 20, 2026 Last updated: September 26, 2026 Welcome to AIVAX. This Privacy … - [Terms of Use](https://docs.aivax.net/docs/legal/terms-of-service.md): Terms of Use Revision: 1.4 Effective date: August 1, 2026 Previous update: July 20, 2026 Last updated: September 26, 2026 Welcome to AIVAX. These Terms of Use … - [Data Processors](https://docs.aivax.net/docs/legal/third-party-processors.md): Data Processors AIVAX uses third-party services for specific operations such as infrastructure, object storage, email delivery, payment processing, web search, … ## Learn: Introduction - [What is Learn](https://docs.aivax.net/learn/introduction/what-is-learn.md): An orientation: who Learn is for, how the modules and learning paths are organised, what the visual blocks mean, and how to track your progress. - [Introduction to artificial intelligence](https://docs.aivax.net/learn/introduction/introduction-to-ai.md): What artificial intelligence is in plain language, how it differs from ordinary software, where language models fit, and why the word agent suddenly matters. ## Learn: Agents - [Introduction to AI agents](https://docs.aivax.net/learn/agents/introduction-to-ai-agents.md): Understand how an AI agent combines language, information and permitted actions to help complete a task. - [What is an LLM](https://docs.aivax.net/learn/agents/what-is-an-llm.md): Learn how a large language model generates text, where its apparent knowledge comes from and why a fluent answer can still be wrong. - [From LLMs to agents](https://docs.aivax.net/learn/agents/from-llms-to-agents.md): Build a mental model of the instructions, context, tools, knowledge, guardrails and memory that turn a language model into a useful agent. - [Adding context](https://docs.aivax.net/learn/agents/adding-context.md): Give an agent the relevant instructions, conversation and evidence it needs without treating its context window as an unlimited filing cabinet. - [Adding tools](https://docs.aivax.net/learn/agents/adding-tools.md): Understand how an agent requests an action, how software executes it and why a confirmed tool result matters more than a confident promise. - [Adding skills](https://docs.aivax.net/learn/agents/adding-skills.md): Package a repeatable way of working into a reusable skill, distinguish it from tools and knowledge, and keep it useful through review and testing. - [Adding guardrails](https://docs.aivax.net/learn/agents/adding-guardrails.md): Set clear boundaries for an agent's requests, answers and actions, combine independent checks, and recognise when a person must take over. - [Adding knowledge](https://docs.aivax.net/learn/agents/adding-knowledge.md): Give an agent reliable company documents, choose useful starting material, and keep its answers connected to current sources. - [Connecting to the world](https://docs.aivax.net/learn/agents/connecting-to-the-world.md): Choose when an agent needs current external information and how to use web pages, files, and scanned documents without treating them as trusted instructions. - [Connecting to existing systems](https://docs.aivax.net/learn/agents/connecting-to-existing-systems.md): Understand how an agent can work with business software through carefully scoped interfaces, permissions, and integration rules. - [Agent-to-agent communication](https://docs.aivax.net/learn/agents/agent-to-agent-communication.md): Understand when specialised agents should share work, how responsibility moves between them, and how to avoid costly or confusing collaboration. - [Workflows as skills](https://docs.aivax.net/learn/agents/workflows-as-skills.md): Turn a recurring business procedure into reusable guidance with explicit steps, checkpoints, approvals, and a clear boundary between model judgement and … - [Writing good prompts and instructions](https://docs.aivax.net/learn/agents/writing-good-prompts-and-instructions.md): Write an agent's job description with clear goals, evidence rules, boundaries, tone, and examples, then improve it using real conversations. ## Learn: Teaching agents - [Introduction to knowledge creation](https://docs.aivax.net/learn/teaching-agents/introduction-to-knowledge-creation.md): Understand which knowledge an agent needs, how it becomes usable evidence, and who keeps it trustworthy. - [What is a RAG](https://docs.aivax.net/learn/teaching-agents/what-is-a-rag.md): Learn how Retrieval-Augmented Generation finds relevant evidence before a model writes an answer, and why evidence still needs checking. - [How to find and prepare knowledge](https://docs.aivax.net/learn/teaching-agents/finding-and-preparing-knowledge.md): Build an approved knowledge inventory by finding useful sources, prioritising real questions and resolving gaps before indexing. - [How to write good documents](https://docs.aivax.net/learn/teaching-agents/writing-good-documents.md): Write knowledge documents that remain clear when searched in pieces, with explicit facts, conditions and useful examples. - [Measuring adherence and hallucination](https://docs.aivax.net/learn/teaching-agents/measuring-adherence-and-hallucination.md): Evaluate whether an agent's answers follow the evidence, cover the intended questions and acknowledge what the sources do not establish. - [Retrieval strategies](https://docs.aivax.net/learn/teaching-agents/retrieval-strategies.md): Choose search and ranking techniques that bring the right evidence into an agent's context without overwhelming it. - [Preparing documents for knowledge](https://docs.aivax.net/learn/teaching-agents/preparing-documents-for-knowledge.md): Turn everyday files into clear, traceable knowledge that an agent can retrieve without losing important conditions. - [Dynamic context](https://docs.aivax.net/learn/teaching-agents/dynamic-context.md): Give an agent current, authorised facts about a conversation without confusing them with durable knowledge or permanent memory. ## Learn: Prompt engineering and context - [Anatomy of a prompt](https://docs.aivax.net/learn/prompt-engineering/anatomy-of-a-prompt.md): Understand the roles inside a model request and how an application rebuilds the conversation each time someone sends a message. - [Prompting techniques](https://docs.aivax.net/learn/prompt-engineering/prompting-techniques.md): Choose practical prompting techniques that clarify a task, demonstrate the expected answer and make outputs easier to use. - [Context window, tokens and cost management](https://docs.aivax.net/learn/prompt-engineering/context-window-tokens-and-cost.md): Understand how tokens share a limited context window and how to manage useful input, answer space and operating cost. - [Memory: short-term and long-term](https://docs.aivax.net/learn/prompt-engineering/memory.md): Distinguish conversation history from stored memory and design useful recall with permission, privacy and expiry in mind. - [Common prompt mistakes and fixes](https://docs.aivax.net/learn/prompt-engineering/common-prompt-mistakes.md): Recognise eight recurring prompt problems and replace them with clear goals, consistent evidence, usable outputs and safe escalation. ## Learn: Models and parameters - [Model families and how to choose](https://docs.aivax.net/learn/models/model-families-and-choosing.md): Choose a model by matching its capabilities, response time, operating cost, and quality to the work you need done. - [Parameters: temperature, top-p, max tokens](https://docs.aivax.net/learn/models/parameters.md): Adjust model settings to balance variation, response length, and consistency without confusing those settings with factual accuracy. - [Multimodality: images, audio, documents, video](https://docs.aivax.net/learn/models/multimodality.md): Choose how an assistant receives and produces media, and understand what can be lost when pictures, recordings, and documents become text. - [Embeddings and semantic search](https://docs.aivax.net/learn/models/embeddings-and-semantic-search.md): Understand how numerical representations help find related ideas, where similarity can mislead, and how search supplies evidence to an assistant. - [Reasoning vs standard models](https://docs.aivax.net/learn/models/reasoning-vs-standard-models.md): Decide when additional model reasoning is worth the waiting time and cost, and how to present useful explanations without relying on hidden thought. ## Learn: Tools and integrations - [Function calling step by step](https://docs.aivax.net/learn/tools-and-integrations/function-calling.md): Follow a tool request from a model's decision through execution and back to an answer the user can trust. - [MCP and tool standards](https://docs.aivax.net/learn/tools-and-integrations/model-context-protocol.md): Understand how a shared tool protocol reduces integration work without removing the need for permissions, trust and careful testing. - [Authentication and permissions for tools](https://docs.aivax.net/learn/tools-and-integrations/authentication-and-permissions.md): Give an agent enough access to help a user without handing it the unrestricted authority of the whole organisation. - [Integrating channels: WhatsApp, web chat, e-mail, voice](https://docs.aivax.net/learn/tools-and-integrations/integrating-channels.md): Adapt one agent's knowledge and rules to different communication channels while preserving identity, conversation continuity and human support. - [Webhooks, events and automations](https://docs.aivax.net/learn/tools-and-integrations/webhooks-events-and-automations.md): Build a clear mental model of event-driven work, scheduled tasks and the safeguards that keep repeated notifications from causing repeated actions. ## Learn: Advanced agents and workflows - [Planning and reasoning loops](https://docs.aivax.net/learn/advanced-agents/planning-and-reasoning-loops.md): Learn how an agent plans, acts, checks results and revises its approach without losing control of time, cost or scope. - [Multi-agent architectures and orchestration](https://docs.aivax.net/learn/advanced-agents/multi-agent-architectures.md): Choose when to divide work among agents and how to coordinate their responsibilities, evidence and limits. - [Human-in-the-loop and approvals](https://docs.aivax.net/learn/advanced-agents/human-in-the-loop.md): Design meaningful approvals and human handovers so an agent can help without taking consequential actions beyond its authority. - [Errors, retries and fallbacks](https://docs.aivax.net/learn/advanced-agents/errors-retries-and-fallbacks.md): Handle failed tools and model responses with bounded retries, safe alternatives and honest messages about what remains unknown. - [Long-running and asynchronous agents](https://docs.aivax.net/learn/advanced-agents/long-running-and-asynchronous-agents.md): Plan background agent work with durable progress, safe restarts, honest notifications and limits on time and spending. ## Learn: Quality, evaluation and observability - [Testing and evaluating agents (evals)](https://docs.aivax.net/learn/quality/testing-and-evaluating-agents.md): Build repeatable tests that show whether an agent reaches the right outcome, follows its rules and keeps working after changes. - [Metrics: accuracy, latency, cost, satisfaction](https://docs.aivax.net/learn/quality/metrics.md): Measure whether an agent solves the right problem at an acceptable speed and cost while giving people a useful experience. - [Logs, traces and monitoring](https://docs.aivax.net/learn/quality/logs-traces-and-monitoring.md): Follow an agent request from question to outcome, record useful evidence and investigate problems without collecting unnecessary private data. - [Continuous improvement from real conversations](https://docs.aivax.net/learn/quality/continuous-improvement.md): Use real conversations to find root causes, make focused corrections and verify that each release improves the agent without breaking other work. - [A/B testing prompts and models](https://docs.aivax.net/learn/quality/ab-testing.md): Compare two agent variants fairly, interpret uncertain results and roll out a change without mistaking chance or biased traffic for improvement. ## Learn: Safety, ethics and compliance - [Prompt injection and jailbreaks](https://docs.aivax.net/learn/safety/prompt-injection-and-jailbreaks.md): Recognise attempts to redirect an agent and build practical layers that limit their impact. - [Privacy, LGPD/GDPR and sensitive data](https://docs.aivax.net/learn/safety/privacy-lgpd-gdpr.md): Plan how an agent collects, uses, stores and deletes personal information without sending more than the task needs. - [Bias, fairness and responsible AI](https://docs.aivax.net/learn/safety/bias-fairness-responsible-ai.md): Recognise unequal treatment, test for it and make human accountability part of an agent's design. - [Content moderation and usage policies](https://docs.aivax.net/learn/safety/content-moderation-and-policies.md): Define an agent's boundaries and respond to unsafe or out-of-scope requests without abandoning legitimate user needs. - [Transparency and human escalation](https://docs.aivax.net/learn/safety/transparency-and-human-escalation.md): Make the agent's identity and limits clear, and design a handover that gives people useful context rather than more work. ## Learn: Production and scale - [Cost optimization and caching](https://docs.aivax.net/learn/production/cost-optimization-and-caching.md): Understand what makes an agent expensive and reduce unnecessary work without weakening answer quality or safety. - [Performance and latency](https://docs.aivax.net/learn/production/performance-and-latency.md): Find where an agent spends time and make waiting shorter, clearer and more predictable without sacrificing correctness. - [Versioning prompts, agents and knowledge](https://docs.aivax.net/learn/production/versioning.md): Keep a reliable record of agent changes so you can test releases, explain behaviour and restore a safer configuration when needed. - [Deployment checklist](https://docs.aivax.net/learn/production/deployment-checklist.md): Turn a promising agent into a controlled launch with clear owners, evidence, operating limits and a gradual rollout. ## Learn: Practical guides and case studies - [Customer support agent, step by step](https://docs.aivax.net/learn/guides/customer-support-agent.md): Build a bounded support assistant for a fictional online store, from instructions and order tools to a measured pilot. - [Sales and qualification agent](https://docs.aivax.net/learn/guides/sales-qualification-agent.md): Design a helpful sales assistant that gathers relevant needs, proposes a transparent qualification result, and hands control to people. - [Internal knowledge assistant](https://docs.aivax.net/learn/guides/internal-knowledge-assistant.md): Create an employee assistant that finds approved policies, respects departmental access, and makes uncertainty visible. - [Common use cases by industry](https://docs.aivax.net/learn/guides/use-cases-by-industry.md): Compare practical agent uses across eight industries and choose a first project with clear evidence, limited permissions, and manageable risk. - [Glossary and cheat sheet](https://docs.aivax.net/learn/guides/glossary-and-cheat-sheet.md): Look up essential agent vocabulary and use a compact checklist for prompts, knowledge, tools, safety, and cost. - [FAQ and learning paths by profile](https://docs.aivax.net/learn/guides/faq-and-learning-paths.md): Answer common questions about agent projects and choose a practical learning route for beginners, developers, or business owners.