Evo AI

What is Evo AI?

Evo AI is an open source platform for building and managing AI agents that you self-host. It wraps Google’s Agent Development Kit (ADK), LangGraph, and CrewAI in a backend and web UI so you can create agents, give them tools over the Model Context Protocol (MCP), and have them communicate through Google’s Agent-to-Agent (A2A) protocol.

What is Evo AI best for?

Evo AI suits teams that want to run their own agent infrastructure instead of a vendor’s managed service, and who care about interoperability. Because every agent publishes a standard A2A agent card, it’s a good fit when you need agents from different systems to call each other, or when you want composable agent types — LLM, sequential, parallel, loop, workflow, and task — assembled as sub-agents under one control plane.

What can Evo AI do?

• Build seven agent types — LLM, A2A, Sequential, Parallel, Loop, Workflow, and Task — and nest them as sub-agents.
• Implement Google’s A2A protocol so agents publish agent cards and talk to agents from other systems.
• Connect tools and data through the Model Context Protocol (MCP), plus custom tools.
• Stay model-agnostic across OpenAI, Anthropic, Gemini, Groq, and Cohere via your own API keys.
• Build complex flows visually with a LangGraph-backed workflow agent and a ReactFlow canvas.
• Manage users and clients with JWT auth and email verification, and store API keys encrypted.
• Trace prompts, model responses, and tool calls through native Langfuse integration.

Where does Evo AI fall short?

CrewAI framework support is still listed as in development, so Google ADK is the primary, fully supported path — if your team is standardized on CrewAI, expect gaps. It’s also a younger project than the managed platforms it competes with, so documentation and community are thinner, and you should expect breaking changes between releases. Finally, it’s self-host only: there is no first-party managed cloud, so someone owns running the FastAPI backend, PostgreSQL, and Redis.

Is Evo AI free?

Yes. Evo AI is released under the Apache 2.0 license and is free to self-host for commercial or personal use. There is no paid edition or managed cloud tier — your only costs are the infrastructure you run it on and the tokens you spend on whichever model providers you connect.

What does Evo AI replace?

Evo AI is an open source alternative to the big cloud agent platforms — Microsoft Copilot Studio, Google Vertex AI Agent Builder, and Amazon Bedrock Agents. Instead of building agents inside a single vendor’s cloud, you run the platform yourself, keep your keys, and stay model-agnostic. If you want a visual builder more focused on LLM apps and RAG, Langflow and Flowise are related open source options.

FAQ

Is Evo AI open source? Yes. It’s licensed under Apache 2.0, which permits commercial use, modification, and redistribution with attribution. The full backend and frontend source lives on GitHub.

Can I self-host Evo AI for free? Yes. The repository ships a Docker Compose setup that starts the backend, PostgreSQL, and Redis. You pay only for your own hosting and model-provider API usage.

Is Evo AI a good Copilot Studio or Bedrock Agents alternative? It’s a strong fit if you want to avoid vendor lock-in and stay model-agnostic across OpenAI, Anthropic, Gemini, Groq, and Cohere. The managed platforms offer deeper first-party cloud integration and support; Evo AI trades that for control and portability via open A2A and MCP.

What do I need to run Evo AI? The backend needs Python 3.10+, PostgreSQL 13+, and Redis 6+; the frontend needs Node.js 18+. Docker and Make simplify setup, and you supply API keys for whichever model providers you use.