🦙 LlamaStack Integration for Agent Lab Web UI

[rawkintrevo]

Welcome to the discussion on integrating LlamaStack into the Agent Lab Web UI, a project maintained by The AI Alliance.

Right now, the app is tightly coupled with Google Vertex AI, but we're architecting for modularity, and the next frontier includes community-driven support for multiple backend providers.

This thread will track and coordinate efforts to:

• Scaffold and modularize LLM provider support
• Define interface contracts compatible with LlamaStack
• Discuss credentials, deployment expectations, and rate limits
• Share example use cases, bugs, and performance considerations

✅ Goal: Enable Agent Lab Web UI to send and receive completions using a LlamaStack backend, either self-hosted or via a third-party provider.

---

💬 Whether you're a LlamaStack power user, a contributor looking to help stub the implementation, or just curious — jump in! Questions, code snippets, or links to your repos are all welcome.

🔗 When a draft PR is ready or RFC emerges, it will be pinned in this thread.

— The AI Alliance

[rawkintrevo]

It probably makes the most sense to just build a helm chart that deploys llamastack+kserve or llamstack+istio so that all of the llamastack end points are exposed to the internet w/auth