Orchestrate LLMs, AI Agents, and RAG in Your Flows
For the complete documentation index, see llms.txt. For a full content snapshot, see llms-full.txt. Append.mdto anykestra.io/docs/*URL for plain Markdown.
Run AI as part of your flows: call models, run AI agents, ground answers in your data, and let external agents call your flows.
This section is for flow authors who put AI into production. Everything here runs during an execution, uses your model provider credentials, and shows up in execution logs and outputs. To get AI help writing or operating flows instead, see Develop with AI.
Choose an approach
| You want to… | Use |
|---|---|
| Summarize, classify, or extract structured data at a fixed step | LLM tasks |
| Give a model a goal and tools, and let it choose the steps | AIAgent tasks |
| Answer questions from your own documents or live web results | RAG workflows |
| Let Claude, Cursor, or another external AI agent run your flows | Flow MCP servers |
MCP in both directions
Kestra works with the Model Context Protocol both ways. As a client, an AIAgent task calls tools on external MCP servers through MCP client tools. As a server, a flow MCP server exposes your flows as tools that external agents call.
Neither is the Docs MCP server, which gives coding agents Kestra documentation and plugin schemas while you write flows.
LLM Tasks and AI Agents
Add AI to Kestra flows with LLM tasks for fixed steps and AI agents that choose their own tools. Compare both, set up providers, and adapt the examples.
RAG Workflows
Build Retrieval-Augmented Generation (RAG) workflows in Kestra to ground LLM responses in your own data or web search results.
Flow MCP Servers
Expose Kestra flows as MCP tools that Claude, Cursor, and other AI agents can call. Create MCP servers, choose authentication, and connect clients.
Was this page helpful?