Introducing Agent Stack (formerly BeeAI Platform)

As our ecosystem has grown, we've heard feedback that having both "BeeAI Framework" and "BeeAI Platform" can be confusing. To make things clearer, we're renaming BeeAI Platform to Agent Stack—a name that more accurately reflects its purpose as open infrastructure for deploying and sharing agents without framework or vendor lock-in.

Why the Change?

As BeeAI has evolved, we ended up with two projects with similar names:

1. BeeAI Framework – for building reliable, intelligent agents
2. BeeAI Platform – for deploying and sharing those agents

While both are essential parts of the BeeAI ecosystem under the Linux Foundation, the similar naming has led to some natural confusion about which tool does what. We've heard from community members that the distinction between "Framework" and "Platform" isn't always immediately clear, especially for newcomers.

Agent Stack makes the purpose explicit—it's your deployment stack for agents.

What is Agent Stack?

Agent Stack is open infrastructure for deploying AI agents to production, regardless of how you built them. It gives you everything needed to go from "works on my laptop" to "other people can use it" - hosting, web UI, runtime services, and multi-tenancy - without framework or vendor lock-in.

Agent Stack gives you:

• A server you can self-host that runs your agents
• A web UI where users interact with your deployed agents
• A CLI tool to deploy and manage your agents
• Infrastructure services that your agents can use:
  • Runtime-configurable LLMs - Switch between 15+ providers (Anthropic, OpenAI, watsonx.ai, Ollama) without code changes
  • Embeddings & vector search - For RAG and semantic search
  • File storage - S3-compatible uploads and downloads
  • Document text extraction - Using Docling
  • External integrations - Connect to databases, APIs, Slack, Google Drive, etc. via MCP protocol with built-in OAuth
  • Secrets - Encrypted credential storage for API keys
• An SDK that lets your agents request these services using standardized A2A protocol extensions
• A HELM Chart for custom infrastructure configuration (plug in your own S3 storage, database, auth, etc.)

Build your agent using agentstack-sdk with any framework you want inside (LangGraph, CrewAI, BeeAI Framework, or custom logic) - the SDK handles requesting infrastructure services at runtime.

The workflow:

1. Build your agent with agentstack-sdk and framework logic
2. Deploy to Agent Stack with a single CLI command
3. Users interact via the web UI

Why This Matters

The agent ecosystem is evolving rapidly. Teams are experimenting with LangGraph, CrewAI, custom code—choosing different tools for different problems. But when it comes to deployment, you face a choice: commit to one framework's hosting, integrate with an enterprise platform, or build everything yourself.

Agent Stack explores a different path: open, self-hostable infrastructure that works across any framework.

This matters if you:

• Want to experiment with multiple frameworks without committing to one
• Need to deploy agents quickly for internal use and testing
• Have different teams using different tools
• Want infrastructure you control and can customize
• Value keeping your options open as the space evolve

What This Means for You

For Existing Users

This is purely a rename—the only changes are to repository URLs, CLI commands, and package names. The migration is straightforward:

• Repository: github.com/i-am-bee/agentstack
• CLI: agentstack prefix
• SDK: agentstack-sdk
• Docs: docs.beeai.dev
• Starter Template: github.com/i-am-bee/agentstack-starter

For New Users

If you're building agents and wrestling with deployment, Agent Stack might be worth exploring!

Get up and running in one line:

sh -c "$(curl -LsSf https://raw.githubusercontent.com/i-am-bee/agentstack/HEAD/install.sh)"

Then check out our Quickstart Guide.

Migration Guide

The steps below cover all user types, whether you’re running locally, building agents, or hosting your own stack.

If You Run It Locally

If you were using the old BeeAI Platform CLI or local server, you’ll just need to reinstall and switch to the new command prefix.

1. Uninstall the old version

beeai self uninstall

2. Install Agent Stack

sh -c "$(curl -LsSf https://raw.githubusercontent.com/i-am-bee/agentstack/HEAD/install.sh)"

3. Use the new CLI

agentstack [OPTIONS] COMMAND [ARGS]

All previous beeai commands now use the agentstack prefix.

If You Build Agents

If you’ve built agents that run on BeeAI Platform, you’ll just need to update your SDK and imports.

1. Install the new SDK

pip install agentstack-sdk

2. Update your imports

from agentstack_sdk import Server

3. Rebuild your container images to apply naming and dependency changes - no code logic changes are required

If You Host Your Own Stack (via Helm)

If you’re running your own instance using Helm, this release requires a clean reinstall. Simply uninstalling isn’t enough because Helm leaves persistent volumes behind. A full reinstall avoids configuration conflicts and ensures compatibility with the new version.

Note: Old conversations, agent data, and configurations won’t carry over. Start fresh for a clean environment.

What’s Next

Agent Stack is in incubation—this means we're actively exploring what agent deployment infrastructure should look like and learning from actual usage. We're focused on:

• Making multi-framework deployment practical. We want to make it genuinely easy to deploy agents from different frameworks without rebuilding infrastructure. This means continuing to refine A2A protocol extensions and improving the deployment experience.
• Improving the development workflow. The CLI and web UI should get out of your way. We're gathering feedback on what works and what doesn't, and iterating based on real usage.
• Building ecosystem connections. As more frameworks adopt A2A, we want Agent Stack to work seamlessly with them. We're also exploring better ways to discover and share agents.
• Understanding real needs. Incubation means we don't have all the answers yet. We're learning what teams actually need from deployment infrastructure—what's essential, what's nice-to-have, and what we got wrong.

If you're building production agents, we'd love to hear what challenges you're facing with deployment. Your feedback shapes where Agent Stack goes next.

Questions or Feedback?

We’d love to hear from you. Share your thoughts, issues, or ideas on Discord or GitHub.