Skip to main content

Welcome to LlamaFarm

LlamaFarm helps you ship retrieval-augmented and agentic AI apps from your laptop to production. It is fully open-source and intentionally extendable—swap model providers, vector stores, parsers, and CLI workflows without rewriting your project.

Found a bug or have a feature request?

Submit an issue on GitHub →

🖥️ Desktop App​

Get started instantly with our desktop application — no command line required.

The desktop app includes everything you need: visual project management, dataset uploads, chat interface, and built-in model management. See hardware requirements →

📺 Video Demo​

Quick Overview (90 seconds): https://youtu.be/W7MHGyN0MdQ

Get a fast introduction to LlamaFarm's core features and see it in action.

What You Can Do Today​

  • Prototype locally with Ollama or any OpenAI-compatible runtime (vLLM, Together, custom gateways).
  • Ingest and query documents using configurable RAG pipelines defined entirely in YAML.
  • Choose your interface – Use the powerful lf CLI for automation and scripting, or the Designer web UI for visual project management with drag-and-drop dataset uploads and interactive configuration.
  • Extend everything from model handlers to data processors by updating schemas and wiring your own implementations.
  • Give models superpowers with MCP (Model Context Protocol) – Connect your AI to local tools, APIs, and databases through a standardized protocol.

MCP (Model Context Protocol)​

LlamaFarm supports MCP – a standardized protocol for giving AI models access to external tools. Connect your AI to filesystems, databases, APIs, and custom business logic.

mcp:
  servers:
    - name: filesystem
      transport: stdio
      command: npx
      args: ['-y', '@modelcontextprotocol/server-filesystem', '/path/to/dir']

runtime:
  models:
    - name: assistant
      provider: openai
      model: gpt-4
      mcp_servers: [filesystem]

Key features:

  • Per-model access control – Different models get different tools
  • Multiple transports – STDIO (local), HTTP (remote), SSE (streaming)
  • Persistent sessions – Efficient connection management

Learn more about MCP →

Choose Your Own Adventure​

Get StartedGo DeeperBuild Your Own
Quickstart – install, init, chat, ingest your first dataset.Core Concepts – architecture, sessions, and components.Extending LlamaFarm – add runtimes, stores, parsers, and CLI commands.
Designer Web UI – visual interface for project management.Configuration Guide – schema-driven project settings.RAG Guide – strategies, processing pipelines, and monitoring.
CLI Reference – command matrix and examples.Models & Runtime – configure AI models and providers.Prompts – prompt engineering and management.

Philosophy​

  • Local-first, cloud-aware – everything works offline, yet you can point at remote runtimes when needed.
  • Configuration over code – projects are reproducible because behaviour lives in llamafarm.yaml.
  • Composable modules – RAG, prompts, and runtime selection work independently but integrate cleanly.
  • Flexible interfaces – Use the CLI for automation, the Designer for visual management, or the REST API for custom integrations.
  • Open for extension – documentation includes patterns for registering new providers, stores, and utilities.
Prefer Visual Tools?

The Designer Web UI provides a browser-based interface for managing projects, uploading datasets, and testing your AI—all without touching the command line. It's automatically available at http://localhost:8000 when you run lf start. Learn more →

🎥 In-Depth Tutorial​

Complete Walkthrough (7 minutes): https://youtu.be/HNnZ4iaOSJ4

Watch a comprehensive demonstration of LlamaFarm's features including project setup, dataset ingestion, RAG queries, and configuration options.

Ready to build? Start with the Quickstart and keep the CLI open in another terminal.