AMD’s GAIA AI agent framework (that previously stood for “Generative AI Is Awesome” albeit they seemed to have dropped promoting it as that name) for Ryzen AI hardware is out with a new version. AMD GAIA 0.17 introduces Agent UI as a new privacy-first web application for local AI agents.
AMD GAIA 0.17’s Agent UI is privacy-first for running AI agents on Ryzen AI hardware for analyzing documents, generating code, searching files, executing commands, and accomplishing tasks. The GAIA Agent UI is run strictly local without any cloud AI usage.
The Agent UI’s front-end is written in React/TypeScript and using an Electron shell. The GAIA Agent UI is promoted as initially offering the following capabilities:
Analyze your documents — Drag-and-drop PDFs, Word docs, or any of 53+ file formats and get answers with page-level citations, powered by local RAG
Execute tools safely — The agent can run shell commands, write files, and use MCP tools — but only after you approve each action
Search and browse files — The agent can find files, explore directories, and locate content across your projects
Access from your phone — Built-in ngrok tunnel lets you use your local GAIA instance from any device
Watch the agent think — Real-time streaming with block rendering shows the agent’s reasoning process inline
Pick up where you left off — Create, switch, and persist sessions with full history
Monitor performance — Hover tooltips show token counts, latency, and throughput metrics per response
Besides running everything locally without any cloud AI use, GAIA 0.17 also adds tool execution guardrails to be able to allow / deny shell commands before they are executed.
GAIA 0.17 also brings Ryzen AI and Radeon hardware detection improvements, improved messaging, system prompt optimizations, and other improvements and fixes.
This GAIA release pairs with the recently released Lemonade SDK 10.0 and FastFlowLM 0.9.35 for finally making Ryzen AI NPUs under Linux useful for running LLMs.
Downloads and more details on the GAIA 0.17 release via AMD’s GitHub.