Skip to main content

​
Roadmap

Work in Progress - This roadmap is actively being refined and may change.
Our focus for the next few quarters, driven by community feedback and usage analytics.
Vote on Features - React with 👍 on the GitHub issue for each plan to show your interest and help us prioritize.
Timelines are estimates and may shift based on community feedback and technical constraints.

​
What’s Next

​
Timeline

​
Q1 2026: Lightweight Installer

One-command installation for GAIA 
irm https://amd-gaia.ai/install.ps1 | iex
Fast onboarding (< 2 minutes from install to first chat) with automatic updates via gaia update. → View detailed plan • Vote with 👍 on GitHub

​
Q1 2026: Chat Desktop UI

Privacy-first desktop chat with document Q&A A lightweight replacement for the previous RAUX desktop UI, focused on chat and document Q&A. Drop PDFs, code files, and documents into a local chat interface. Your data never leaves your device. Key capabilities:
  • 50+ file format support (PDF, TXT, MD, code files, etc.)
  • Streaming responses with source citations
  • Session management and export
  • Shared state between CLI and desktop app
  • Lighter and faster than RAUX
→ View detailed plan • Vote with 👍 on GitHub

​
Q1 2026: MCP Client Mixin

Computer Use Agents for GAIA An MCP Client Mixin that enables GAIA agents to connect to external MCP servers, transforming them into Computer Use Agents (CUA): 
from gaia.agents.base import Agent
from gaia.mcp import MCPClientMixin

class DesktopAgent(MCPClientMixin, Agent):
    def __init__(self):
        super().__init__()
        self.connect_mcp_server("windows", "uvx windows-mcp")
Key capabilities:
  • Connect to any MCP server (Windows, browser, database, etc.)
  • Desktop automation via Windows MCP
  • Auto-registration of MCP tools as agent tools
  • Multi-server orchestration
→ View detailed plan • Vote with 👍 on GitHub • See CUA for the first production implementation

​
Q1 2026: Computer Use Agent (CUA)

AI-powered desktop automation through natural language The first production agent built on GAIA’s MCP Client infrastructure. CUA lets you control your desktop through natural language commands: 
gaia cua "turn on dark mode"
gaia cua "check my battery status"
gaia cua "prepare my laptop for a meeting"
Built as a reference design, CUA demonstrates how to create computer-use agents that connect to MCP servers (like Windows MCP) for desktop automation. Key capabilities:
  • Natural language desktop control
  • Battery conservation workflows
  • System health monitoring
  • Meeting preparation automation
  • Windows settings management
CUA validates the MCP Client Mixin framework and serves as a template for building other computer-use agents. → View detailed plan • Vote with 👍 on GitHub

​
Q1 2026: Docker Containers

Self-contained images for running and developing GAIA Official Docker images for running GAIA in isolated environments. All images are self-contained—connect from your host terminal via docker exec -it. 
docker run -dit --name gaia -e LEMONADE_BASE_URL=<lemonade-url> amd/gaia:linux
docker exec -it gaia zsh
Key capabilities:
  • Linux image: Run GAIA on any Docker-compatible host via pip-installed package
  • Windows image: Native Windows container for Windows Server environments
  • Development image: Full dev environment with UV, Node.js 20, GitHub CLI, and Claude Code with sandboxing
  • Consistent ports: All images expose the same ports (3000, 5173, 8000, 9229, 9222) for API, debugging, and frontend dev
  • Environment config: Set LEMONADE_BASE_URL at container creation to connect to a Lemonade Server
→ View detailed plan • Vote with 👍 on GitHub

​
Q1 2026: SD Agent

Intelligent Stable Diffusion optimization assistant An AI agent that optimizes both prompts and generation parameters for Stable Diffusion, producing professional-quality images from simple descriptions. Built with AMD NPU-accelerated LLMs. 
# Analyze and enhance prompt, optimize parameters
gaia sd generate "a mountain landscape"

# Search your generation history
gaia sd search "show me all cyberpunk images"

# Browse gallery with ratings and annotations
gaia sd gallery
Dual optimization approach: enhance text prompts AND recommend optimal SD parameters (model, size, steps, cfg_scale). Includes searchable database with gallery UI for managing your creations. Key capabilities:
  • LLM-powered prompt enhancement and parameter optimization
  • SQLite database with natural language search
  • Gallery UI with chat interface for image creation
  • Rating system that learns your preferences
  • Template library with proven prompt+parameter combinations
  • Terminal image display for immediate feedback
  • Reference-based generation using your top-rated images
Transform “a cat” into professional SD images through intelligent optimization of all generation parameters, with a searchable gallery that learns your aesthetic preferences. → View detailed plan • Vote with 👍 on GitHub

​
Q2 2026: Vision SDK

Unified document processing pipeline with VLM-powered OCR A comprehensive SDK for processing any document type—medical forms, legal logs, technical manuals—using Vision Language Models running locally on AMD hardware. 
from gaia.vision import VisionSDK

vision = VisionSDK()

# Medical forms → structured data
result = vision.extract("form.pdf", schema=medical_schema)

# Legal logs → tables + visuals
result = vision.extract("logs.pdf", extract_tables=True, extract_visuals=True)

# Technical manuals → RAG indexing
result = vision.extract("manual.pdf", pages="all")
Key capabilities:
  • Multi-page document processing (1 to 1,200+ pages)
  • Table and visual element extraction
  • Handwriting recognition (validated)
  • Structured data extraction with schemas
  • Batch processing and expense reporting
  • Seamless RAG integration
  • Agent mixin for vision-powered tools
  • Production-ready with checkpointing and parallel processing
Impact: Reduce vision agent code by 60%, enable document automation workflows, power legal/medical/technical document processing. → View detailed plan • Vote with 👍 on GitHub

​
Q2 2026: C++ Framework Production Readiness

Make the native C++ agent framework ready for third-party integration The C++ Agent Framework shipped in Q1 2026 as a native C++17 port of the base agent — same agent loop, tool registry, and MCP support, compiled to a standalone binary with no Python runtime. The next phase focuses on production readiness: 
// Cancellation support
gaia::CancellationToken token;
std::thread([&] { agent.processQuery("diagnose network", 20, token); }).detach();
token.cancel();  // graceful stop from UI thread

// Structured event callbacks for GUI integration
agent.addEventListener(std::make_shared<MyUIEventHandler>());

// Tool security policies
toolRegistry().registerTool("restart_service", desc, callback, params,
                            gaia::ToolPolicy::CONFIRM);
Key deliverables:
  • Cancellation & timeout — CancellationToken for graceful abort, configurable HTTP timeouts
  • Event callback system — typed AgentEvent structs for embedding in desktop applications (WPF, Qt, Electron)
  • Tool security — per-tool allow/confirm/deny policies, argument validation callbacks, path traversal utilities
  • API versioning — semantic versioning, stability guarantees, deprecation lifecycle
  • Performance benchmarks — binary size tracking, loop latency, memory footprint, CI regression detection
  • Pluggable logging — Logger interface for ETW, syslog, or custom telemetry
  • Streaming responses — SSE parsing, token-by-token callbacks for responsive UIs
  • Runtime configuration — JSON config files, environment variables, dynamic model switching
→ View C++ framework docs • View milestone on GitHub

​
Q2 2026: MCP Docs Server

AI assistant context for GAIA development An MCP server that gives AI coding assistants (Claude Code, Cursor, GitHub Copilot) intelligent access to GAIA SDK documentation: 
gaia mcp docs start
Key capabilities:
  • Semantic search across all GAIA documentation
  • Working code examples on demand
  • Version-aware API information
  • Works with Claude Code, VSCode, Cursor, and any MCP-compatible client
Make AI assistants expert GAIA developers out of the box. → View detailed plan • Vote with 👍 on GitHub

​
Q2 2026: AI PC Agents Hub

The App Store for AI PC agents—discover, try, compete, and run locally A community platform where developers publish agents, users discover solutions, and the community decides what’s best:

Agents Hub

Browse and install AI PC agents

Agent Arena

Try agents on AMD Strix Halo cloud

Competitions

Compete for prizes and recognition
Key capabilities:
  • One-click “Try in Agent Arena” on AMD Strix Halo hardware
  • gaia agent install <name> for local deployment
  • Themed competitions with real prizes
  • Community voting and leaderboards
  • Developer profiles with badges and recognition
→ View detailed plan

​
Recently Shipped

​
v0.15 - Framework Evolution

January 2026
Repositioned as pure SDK/framework for AI PC agents. Added Linux support, 20+ components, and expanded documentation.

​
v0.14 - Agent Expansion

December 2025
EMR agent with desktop UI, Blender automation, enhanced Code agent.

​
v0.13 - MCP Integration

November 2025
Model Context Protocol support for external integrations.

​
How We Prioritize

We prioritize based on community votes and needs:
How You Can InfluenceWhat It Affects
👍 Vote on GitHub issuesFeature priority
💬 Share your use caseWhat we build next
🐛 Report bugsWhat we fix first
🤝 Contribute PRsFaster delivery
The more votes an issue gets, the higher priority it becomes. Your input directly shapes the roadmap.

​
Get Involved

Report Issues

Found a bug or have a feature request?

Contribute

Help build GAIA

Share Your Use Case

Tell us how you’re using GAIA
Updated: February 27, 2026