Features

​

API v0.20.0

• Real-Time Voice Mode: Talk to GAIA in a live, full-duplex call powered by LiveKit. A dedicated voice worker bridges your audio to the same agent that runs text chat, so voice has the full toolset — GAIA acknowledges right away, then narrates the answer as it works.
• Smarter Paid Model: Paid plans now run on MiniMax M3 (free plans stay on Gemini), with reasoning enabled and provider routing hardened so tool calls and long answers stay reliable.
• Model Thinking in Chat: GAIA streams its reasoning alongside each step and shows it as a collapsible Thinking row inside the tool card, so you can see how it reached an answer.
• User Skills in Settings: Create, edit, and manage your own skills from a dedicated settings page.

​

Web v0.21.0

• Voice Mode: A new in-app voice experience with live transcription, token-by-token replies, and tool cards that render right inside the call.
• Voice Selection: Pick from a library of 100+ voices with country and language filters, starred favorites, and preview-on-select. Swap voices live, mid-call, from the “Customise voice” popover.
• Model Thinking: Reasoning now appears as collapsible Thinking rows in the unified tool card, at both the main and subagent level.
• Queue Your Next Message: Send a follow-up while GAIA is still replying and it waits in line — the composer’s Send button morphs to Queue, and the queued message shows as pending until it’s picked up.
• User Skills Management: A settings page to manage your skills.

​

Desktop v0.4.0

• Hey GAIA Wake Word: Say “Hey GAIA” to summon a Siri-style assistant popup and start talking without touching the keyboard.

​

API v0.19.0

• Local Memory Engine: Rebuilt memory from the ground up as a fully local engine, replacing the previous cloud memory provider. Recall makes no external calls on the read path, so it is faster and more private.
• Four Kinds of Memory: GAIA now organizes what it learns into facts filed in folders, a dated journal of what happened over time, auto-maintained core documents about you, and an entity graph of the people, tools, and projects in your life. See the memory guide for how each one works.
• Sharper Recall: Hybrid semantic and keyword search with local reranking surfaces the right memory even when your request never names it directly.
• Memory Tools for the Agent: GAIA can search your memory, journal, and past conversations, and add, update, or forget specific facts on request, each shown inline as a card in chat.
• Conflict-Aware Updates: When new information replaces something GAIA already knew, it updates the fact and keeps a version history instead of piling up duplicates.

​

Web v0.20.0

• Redesigned Memory Settings: A new memory page with Folders, Documents, Graph, Journal, and All views, so you can see exactly what GAIA knows and how it is organized.
• Entity Graph View: Explore the network of people, tools, and projects GAIA has connected, and jump from any node to the memories behind it.
• Memory Journal: A collapsible, day-by-day timeline of what GAIA recorded over time.
• Version History: See how a memory changed across updates, and delete anything you don’t want GAIA to keep.
• Email Previews with Avatars: Email addresses in chat now show the person’s name and photo, pulled from your contacts.

​

API v0.18.0

• Code Sandbox: Agents now run code in an isolated sandbox environment, making data analysis, scripting, and file generation safer and more reliable.
• Self-Managing Agent Workspace: GAIA maintains its own working files across a task, so multi-step work carries context from one step to the next.
• Notification Tools: Agents can proactively send you notifications when something needs your attention.
• Agent-Generated Files: GAIA can create files during a task and deliver them straight to you, including over Discord, Slack, Telegram, and WhatsApp.
• GAIA Self-Knowledge: A dedicated subagent lets GAIA answer questions about its own features and how to use them.
• Per-Integration Custom Instructions: Add your own instructions to any connected integration to tailor how GAIA uses it.
• Quick Actions and Community Workflows: Refreshed quick actions plus a seeded set of community workflows to start from.

​

Web v0.19.0

• What’s New in the App: A new sidebar card and settings timeline surface the latest release notes right inside GAIA.
• Redesigned Chat: A cleaner chat interface with less clutter and snappier interactions.
• Revamped Onboarding: A rebuilt onboarding flow with a guided intelligence pipeline and a smoother card reveal sequence.
• Hey GAIA Wake Word: On-device “Hey GAIA” detection so you can start a conversation hands-free.
• Todo Link Previews: Links in todo titles now show inline favicons for quick context.
• Landing Page Overhaul: Updated copy, navbar, pricing, comparison grid, and a live iPhone chat demo.

​

Mobile v0.6.0

• Redesigned Chat: The mobile chat interface picks up the same cleaner, faster redesign as the web app.
• Hey GAIA Wake Word: On-device “Hey GAIA” detection for hands-free use on the go.
• Notification Tools and Agent Files: GAIA can send you proactive notifications and deliver generated files on mobile.

​

Bots v1.4.0

• Media and Voice on WhatsApp and Telegram: Send and receive images and voice messages directly in your bot conversations.
• Agent-Generated Files: Files GAIA creates during a task are now delivered straight into your bot chat.
• Proactive Bot Messages: Background tasks and notifications can reach you on your connected bot platforms.

​

API v0.17.0

• WhatsApp Bot Integration: Connect GAIA to WhatsApp so you can chat with your agent directly from your phone’s messaging app. See bots overview for setup details.
• MCP Apps Builder: Build and run custom MCP-powered apps inside GAIA conversations, letting agents render interactive UI components on the fly.
• Deep Research Tool: Agents can now perform multi-step web research with source citations. Free-tier users get search fallbacks so everyone benefits.
• Artifacts: Agents can generate and display rich content like documents, tables, and code in a dedicated panel alongside the conversation.
• Calendar Attendees and Google Meet: Create calendar events with attendees and auto-generate Google Meet links, all from a single prompt.
• Custom Integration Context: Integrations can now pull in relevant context automatically, giving agents better awareness when using your connected tools.
• Streaming Calendar Data: Calendar information now streams to the frontend in real time, so you see results as they arrive instead of waiting for a full response.

​

Web v0.18.0

• MCP Apps Builder: Build and preview MCP-powered mini-apps directly in the chat interface.
• Artifacts Panel: View agent-generated documents, code, and tables in a dedicated side panel.
• Workflow Toast Actions: Workflow completion toasts now include an “Open” button so you can jump straight to the result.
• Status Banner: A new banner alerts you to service status changes and scheduled maintenance.

​

Desktop v0.3.0

• Artifacts Support: The desktop app now supports the new artifacts panel for viewing agent-generated content.

​

Mobile v0.5.0

• WhatsApp Integration: Chat with GAIA on WhatsApp, bringing your agent to your favorite messaging app.

​

Bots v1.3.0

• WhatsApp Bot: GAIA is now available on WhatsApp in addition to Discord, Slack, and Telegram.

​

API v0.16.0

• System Workflows for Gmail and Google Calendar: Auto-created workflows when connecting Gmail or Google Calendar, delivering ready-to-use automations out of the box
• LangChain Middleware, VFS, and Skills: New middleware layer for LangChain integrations, a virtual file system abstraction, and first-class skills support for agents
• Workflow Prompt and Description Split: Display descriptions and execution prompts are now separate fields, giving users clearer context while preserving precise agent instructions
• Docker Image Optimization: Slimmed down API images by pruning unused dependencies and improving layer caching

​

Web v0.17.0

• Vercel Web Analytics: Production usage tracking integrated into the Next.js frontend
• Cloudflare Deployment Support: Added Wrangler configuration, SSR chunk analysis scripts, and deploy/preview commands for Cloudflare Pages
• Server API Base URL Centralization: Consistent API base URL handling across all server components via a shared utility and new provider architecture
• OAuth Success Toast: Reintroduced global OAuth success notifications for a smoother post-auth experience

​

Desktop v0.2.0

• Deep Linking: Protocol handler registration enables opening GAIA links directly in the desktop app
• Branded Splash Screen: New splash screen with refined appearance during app startup
• Auto-Updater: Built-in automatic update support keeps the desktop app current without manual downloads
• Modular Window Management: Refactored window lifecycle into a modular architecture for better maintainability

​

Mobile v0.4.0

• LangChain Middleware and Skills Support: Mobile app now benefits from the shared LangChain middleware layer, VFS abstraction, and agent skills system
• Dependency Updates: Updated package dependencies for improved stability and compatibility

​

API v0.15.0

• Bot Platform Endpoints: Comprehensive API endpoints for bot platform integration, supporting Discord, Slack, and Telegram authentication and messaging
• Streaming Keepalives and Retries: Server-sent keepalive events and client-side retry logic for transient streaming errors, with adjusted timeout thresholds
• Multi-Channel Notifications: Push notifications to Telegram and Discord channels, letting GAIA proactively reach users on their preferred platform
• Secure Platform Linking: Token-based platform linking flow with WorkOS SSO return URL handling, enabling users to connect bot accounts to their GAIA identity
• Bot API Key Authentication: Dedicated middleware for bot API key auth, isolating bot traffic from user session auth
• Integration Status in Responses: API responses now include integration connection status, surfaced in bot settings UI

​

Web v0.16.0

• Bots Landing Page: Dedicated landing page introducing GAIA bots with platform highlights, setup instructions, and invite links for Discord, Slack, and Telegram
• Discord Bot Landing Page: Standalone page for the Discord bot with feature overview and one-click server invite
• Platform Linking UI: New settings panel where users can link and manage their Discord, Slack, and Telegram accounts
• Docs Structure and Styling: Overhauled documentation site with improved navigation hierarchy, component styling, and content organization
• CLI Installation Page: Dedicated page walking users through CLI installation and first-run setup
• Multi-Channel Notification Settings: UI for configuring which platforms receive proactive notifications from GAIA

​

Bots v1.1.0

• Discord Bot: Full-featured Discord bot with slash commands, DM support, context menu commands, rotating presence status, and a welcome message for new DM conversations
• Slack Bot: Slack integration with threaded conversations, streaming responses, and workspace-level authentication
• Telegram Bot: Telegram bot with serialized message updates, Markdown parsing, and streaming message break handling
• Streaming Responses: Real-time streamed replies across all three platforms with dynamic message editing and new-message-break logic for long responses
• Platform Linking: Secure account linking from any bot to a GAIA account using one-time link tokens and SSO
• Help and Settings Commands: /help and /settings slash commands displaying available actions and current configuration per platform
• Stop and Unlink Commands: /stop to end the current conversation and /unlink to disconnect a bot account from GAIA
• New Command: /new command to start a fresh conversation without carrying prior context
• Centralized Bot Configuration: Unified configuration with Infisical secret injection across all three bots
• Nx and CI Integration: All bots integrated into the Nx workspace, CI pipeline, and release system with per-bot Docker builds
• E2E Test Package: Dedicated end-to-end test package for cross-bot integration testing

​

CLI v0.2.0

• Init Command: Guided initialization flow that clones the repo, installs dependencies, configures environment variables, and starts services
• Status Command: Real-time dashboard showing running services, ports, and health status
• Stop Command: Graceful shutdown of all GAIA services with cleanup
• Branch Selection: --branch flag on init to clone a specific Git branch for development or testing
• Docker Support: Full Docker Compose integration with prerequisite checks, detailed error messages, and service management
• Port Conflict Detection: Automatic port availability checks during init with clear conflict resolution guidance
• Logging Dashboard: Live log streaming UI during service startup with per-service filtering
• Post-Start Health Verification: Automatic health checks after startup to confirm all services are responding
• npm Distribution: Migrated packaging from Bun to npm for broader compatibility and simpler installation

​

Bots v1.0.0

• Shared Adapter Pattern: BaseBotAdapter in @gaia/shared establishing the unified command dispatch system, GAIA API client, and platform-agnostic RichMessageTarget interface used by all three bots
• Bot API Key Authentication: Dedicated middleware for authenticating bot-to-API requests independently from user session auth
• Nx Monorepo Integration: All three bot packages integrated into the Nx workspace with build, lint, type-check, and release targets
• Docker Support: Dockerfiles for each bot for containerized deployment alongside the rest of the GAIA stack
• E2E Test Package: Dedicated bots-e2e test package with Vitest infrastructure for cross-bot integration testing

​

CLI v0.1.0

• Core Commands: init, setup, start, stop, status, and logs commands covering the full self-hosting lifecycle
• Interactive Terminal UI: Built with Ink (React-in-terminal) for a rich, step-by-step setup experience with live progress display
• Docker Compose Integration: Full service orchestration via Docker Compose with prerequisite validation and detailed error messages
• Self-Host and Developer Modes: start supports both a full self-hosted stack and a developer mode that runs only infrastructure services locally
• --branch Flag: Clone a specific Git branch during init for development or testing against unreleased code
• npm Distribution: Published to npm as @heygaia/cli, installable via npx or a curl install script

​

API v0.14.0

• PostHog analytics unified across frontend and backend for end-to-end event tracking
• Automatic workflow creation lets GAIA generate workflows from natural language descriptions
• Execution history tracking and storage for all workflow runs
• MCP orchestration overhaul with improved tool indexing, namespace normalization, and integration cleanup
• Pricing update reflecting the new $30/month tier with adjusted usage limits

​

Web v0.15.0

• Interactive demo sections on the landing page showcasing chat and workflow capabilities with live animations
• Deep linking via query parameters, enabling direct navigation to specific views from external links
• FAQ structured data and SEO enhancements for improved search discoverability
• Execution history UI displaying workflow run status, timing, and output details
• Blog and manifesto updates for the “Tool Calling Works” post with minor landing page improvements

​

Mobile v0.3.0

• Automatic workflow creation mirroring the web experience for on-the-go automation
• MCP tooling and execution history support aligned with backend orchestration improvements

​

API v0.13.0

• Background Task Management: Overhauled background task lifecycle with improved scheduling, status tracking, and message persistence handling
• MCP Bearer Token Authentication: Integrations that require bearer tokens can now be connected directly, bypassing the OAuth flow entirely
• Integration Suggestion Search: Smarter word-based matching when suggesting integrations, filtering stopwords and searching across name, description, and category
• Optional Bot Configuration: Bot platform keys [Discord, Slack, Telegram] are now optional in production settings, so the API starts cleanly without all bot tokens present

​

Web v0.14.0

• Animated Loading Indicator: New chat loading component with Framer Motion transitions and a polished shine animation for better perceived responsiveness
• Tool Output Persistence: Tool call outputs now stream to the frontend in real time and persist across page reloads

​

API v0.12.1

• Voice Agent added as a new monorepo app with shared Python and TypeScript libraries for real-time voice processing
• Discord, Slack, and Telegram bots introduced with initial implementations and supporting API endpoints
• MCP Marketplace with public integrations page for discovering and connecting third-party tools
• Push notifications backend with device token registration, deregistration, validation, and per-device limit enforcement
• WebSocket notifications with secure authentication via subprotocols for real-time event delivery
• Bearer token authentication for mobile clients with configurable WorkOS redirect URIs
• Memory prompt optimization for improved agent recall and context management
• System prompt management node replacing message deletion, preserving memory messages across agent graph hooks

​

Web v0.13.0

• Real-time notifications via WebSocket with notification provider, permission prompts, and in-app notification center
• MCP Marketplace page for browsing and enabling public integrations directly from the dashboard
• Retry message action allowing users to regenerate a failed or unsatisfactory assistant response
• Voice mode improvements including chat description auto-generated in the sidebar when a conversation starts in voice mode
• WaveSpinnerSquare animation component for richer loading states during voice interactions
• Memory prompt optimization for more accurate and relevant agent recall in conversations

​

Mobile v0.2.0

• Chat streaming with SSE client integration for real-time message delivery on mobile
• Push notifications with Firebase Cloud Messaging, custom notification sound, and device token lifecycle management
• Real-time WebSocket notifications with secure subprotocol-based authentication
• Tool data cards and interactive follow-up actions rendered inline within chat messages
• Bearer token authentication replacing cookie-based auth for native mobile sessions
• Zustand state management integrated across the app for consistent global state
• Settings sheet with redesigned chat UI components and styling utilities
• Connect drawer for quick access to integrations from the chat input
• Chat history with dynamic loading, multi-part message support, and progress indicators
• Playful AI thinking messages with animated loading states in chat bubbles
• App logo and splash screen updated to new branding

​

API v0.12.0

• Unread conversations: Conversations with new messages are surfaced with unread indicators, including after onboarding flows and workflow executions
• Reply to message: Users can reply directly to specific messages within a conversation, adding threaded context to chat interactions
• Abstract workflow steps: Workflow step generation now uses abstract action descriptions instead of specific tool names, making workflows more portable and readable

​

Web v0.12.0

• Keyboard shortcuts: Navigation now supports keyboard shortcuts for faster access to conversations, settings, and core actions
• Tools We Love page: New curated page showcasing recommended integrations and tools that complement GAIA

​

Mobile v0.1.1

• Chat with AI: Full chat feature on mobile with hooks, context management, and streaming AI responses
• WorkOS authentication: SSO integration for mobile with authenticated routing, session management, and a user dropdown menu
• Theme system: Light and dark theme toggle with NativeWindCSS, reusable UI components, and a design-system-aligned color palette

​

Desktop v0.1.0

• Initial release: Electron desktop app bundling the Next.js standalone build for native desktop usage, bringing the full GAIA web experience to macOS, Windows, and Linux

​

Mobile v0.1.0

• Expo SDK 54: Built on Expo SDK 54 with Expo Router v6, React Compiler enabled, and the new React Native architecture turned on
• UI Component Library: 18+ reusable components built on @rn-primitives, avatar, button, card, checkbox, dialog, dropdown-menu, input, popover, raised-button, select, skeleton, toggle, tooltip, and more
• NativeWind Styling: Tailwind CSS for React Native with a surface color system, custom spacing, and full light and dark theme support
• Icon System: @hugeicons/react-native and Lucide React Native for consistent iconography across the app
• Feature-Based Architecture: Code organized into features/auth/ and features/chat/ with co-located components, hooks, utilities, and types