​

API v0.17.0

​

Features

• 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.

​

Improvements

• Native Marketplace Integrations: The marketplace now highlights native integrations with dedicated cards, making them easier to discover and connect.

​

Bug Fixes

• Billing Address Country: You can now edit the billing address country on the payments page.
• Backend Stability: Resolved an issue that could cause brief service interruptions under heavy load.
• Web Crawl Reliability: Improved timeout handling and batching when agents crawl web pages, reducing failures on large sites.
• MCP Connection Recovery: MCP connections now automatically refresh expired tokens and display clear status messages when re-authentication is needed.
• Research Tool Caching: Fixed caching and validation in the research tool so repeated queries return consistent results.

​

Performance

• Todos Performance: Faster loading and smoother interactions across the tasks and todos experience.

​

API v0.16.0

​

Features

• 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

​

Bug Fixes

• Google OAuth Production Settings: Added missing client ID and secret to ProductionSettings for Google OAuth flows
• Dockerfile Ownership and Permissions: Corrected file ownership for the libs directory so setuptools can create package metadata
• Explore Workflow Payload Hardening: Fixed payload handling for the description/prompt split in explore workflows
• Deployment Script Updates: Improved service verification and environment setup in deployment configurations

​

API v0.15.0

​

Features

• 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

​

Bug Fixes

• User ID Fallback: Added user ID fallback from metadata in handoff and retrieval nodes, preventing orphaned conversations
• Subagent Web Search: Resolved KeyError in subagent web search tool and OpenRouter credit handling
• MCP OAuth Validation: Raises a clear error when OAuth tokens are missing in the MCP client instead of failing silently

​

Performance

• Agent Flow Latency: Optimized agent execution flow to reduce tool-calling latency across all agent nodes

​

API v0.14.0

​

Features

• 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

​

Bug Fixes

• Webhook idempotency for DodoPayments, preventing duplicate processing of already-handled events
• Timezone-aware scheduled tasks that reschedule past-due jobs instead of triggering Redis PSETEX errors
• MCP namespace normalization resolving schema breakage and connection failures with PostHog and Twitter integrations
• Integration sidebar fixes for display names and MCP-related rendering issues

​

API v0.13.0

​

Features

• 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

​

Bug Fixes

• Subagent Stream Cancellation: Users can now cancel long-running subagent and executor streams mid-flight instead of waiting for completion
• Integration Auth Data Sync: Mismatches between stored integration documents and MCP config for auth requirements are detected and auto-corrected
• Build Workflow Project Detection: Fixed affected-project checks in CI and added missing project type to the API configuration

​

API v0.12.1

​

Features

• 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

​

Bug Fixes

• Fixed streaming issues causing dropped or duplicated messages
• Resolved “no tool results” error when agent tools returned empty responses
• Fixed mobile redirect URI to use settings configuration instead of hardcoded values
• Temporarily disabled auto-loading of tools in production to prevent startup failures

​

API v0.12.0

​

Features

• 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

​

Bug Fixes

• Workflow generation from todos: Fixed workflow step generation that broke when creating workflows from todo items

​

Infrastructure

• Independent per-app versioning: Migrated from monorepo-wide versioning to per-app release-please configuration, enabling each app to release on its own cadence
• Docker Compose deploy path: Fixed production file path reference in the deploy workflow

​

v0.10.1

​

Bug Fixes

• File uploads in chat: Fixed an issue where messages with file attachments could not be sent, restoring full support for sending files alongside text in conversations
• Tool hashing and batch processing: Improved the hash computation logic for tool initialization and optimized how tool updates are batched into ChromaDB, reducing redundant writes and improving reliability
• Message validation: Tightened the message sending validation logic to correctly handle file upload payloads

​

v0.10.0

​

Features

• Mem0 v2 migration: Migrated the entire memory backend to the Mem0 v2 API, enabling graph-based memory with richer relationship modeling, bulk delete support, and an updated graph visualization and export UI
• Redesigned onboarding flow: Added a multi-step onboarding experience with a progress indicator, a dedicated connections step, and an integration picker so new users can link their services during setup
• Onboarding OAuth persistence: Onboarding state now persists across OAuth redirects, so users are returned to the correct step after connecting a third-party account
• Subscription settings UI: Rebuilt the subscription management screen with a new layout, subscription illustration, and clearer plan details
• Discount code support: Subscription checkout now accepts discount codes and uses hosted checkout sessions for a smoother payment experience
• Gmail import expansion: Expanded Gmail import capabilities with async memory handling, letting GAIA ingest and remember more of your email context
• Integration tools for the agent: Added new integration-specific tools so the agent can interact with connected services more effectively

​

Improvements

• Settings menu redesign: Reorganized the settings menu with nested tooltips and polished iconography for better navigation
• Chat accordion controls: Added the ability to expand and collapse tool result sections in chat for a cleaner conversation view
• Pro button restyle: Updated the Pro upgrade button styling for better visibility and consistency with the design system
• Pricing card updates: Refreshed pricing bento cards with new labels, feature sets, and streamlined rendering logic
• Plan cleanup tooling: Added a backend script to clean up legacy GAIA Pro plans and update payment setup descriptions

​

Bug Fixes

• Fixed IndexedDB message syncing that caused messages to appear out of order or duplicate in the sidebar
• Fixed OAuth redirect security by validating the success parameter as a boolean string to prevent open redirect attacks
• Fixed memory service issues related to the Mem0 v2 migration, including type annotation corrections and unused import cleanup
• Fixed error handling for missing subscription or plan data on the payment summary screen
• Fixed subscription creation error logging in the payment service
• Fixed notification service import path in reminder tasks
• Fixed follow-up action error handling to prevent noisy parse exception logs

​

v0.9.1

​

Features

• Explore workflows: Added a curated “explore” section for featured workflows, served from a new backend endpoint with caching, replacing the previous static dummy data across the landing page, use-cases pages, and workflows UI
• ComparisonTable redesign: Enhanced the pricing comparison table with integration-aware props and an improved responsive layout
• Trello and ClickUp tools: Introduced backend tool definitions, orchestrator prompts, and OAuth configuration for Trello board/card/list management and ClickUp spaces/tasks/time tracking
• Thinking bubble: Added a ThinkingBubble component that surfaces the AI’s intermediate reasoning steps in the chat interface

​

Bug Fixes

• Gemini empty response handling: Fixed an issue where Gemini could return empty AI messages by adjusting subagent handoff patterns and preventing consecutive AIMessage sequences
• PostgreSQL checkpointer: Resolved a connection pool issue in the LangGraph checkpointer by adding autocommit and prepare threshold options to the pool setup
• Reddit comments parsing: Fixed a crash when Reddit returns unexpected dictionary structures in comments listings by adding defensive type checks in the after-hook
• Caching strategy for integrations: Disabled stale and garbage collection timers on integration status queries to prevent serving outdated connection states

​

Improvements

• Workflows header and layout: Added a dedicated WorkflowsHeader component, introduced reusable grid and section rendering helpers, and improved loading/error/empty states across the workflows pages
• Consolidated Reddit UI: Merged separate Reddit post and comment section components into a single unified component
• Sidebar cleanup: Removed the notifications button from the sidebar top buttons
• Import and lint cleanup: Removed unused imports, fixed import ordering, and resolved lint warnings across multiple frontend components

​

v0.9.0

​

Features

• Reddit integration: Built a full Reddit tool with backend hooks for processing search results, posts, and comments. Added dedicated UI components for displaying Reddit content in chat, including search cards, post cards, comment cards, and created-post confirmations, all rendered inline with the conversation
• Grouped tool rendering: Introduced a grouped tool data system in the chat bubble renderer so multiple Reddit tool calls are merged into a single cohesive section instead of appearing as separate blocks
• Request an Integration modal: Added a floating “Request an Integration” button on the integrations page that opens a prefilled support modal, making it easy for users to ask for new service connections
• Integration connect cards: Added inline connect cards for Google Calendar, Gmail, and chat integrations that appear contextually when GAIA needs an unconnected service, guiding users through setup without leaving the conversation
• Configurable LLM model fields: OpenAI and Gemini LLM clients now accept configurable model parameters during initialization, allowing dynamic model selection at runtime
• Redis Docker service: Added a Redis container with health checks to the Docker Compose development stack

​

Bug Fixes

• Google service identifiers: Normalized all Google integration identifiers to lowercase format across backend tool registry, OAuth config, and frontend icon mappings, fixing mismatches that prevented proper connection detection
• Slash command dropdown: Virtualized the slash command dropdown for smoother scrolling, added locked-tool and integration-connect UI states, and fixed the dropdown to only show the integrations card when opened via button rather than during search

​

Improvements

• Calendar tool category rename: Renamed the backend tool category from “calendar” to “google_calendar” with corresponding frontend alias mapping, improving clarity when multiple calendar providers are supported
• Pricing page refresh: Unified pricing card styling to a single blurred variant, updated the comparison table layout, and added a new background wallpaper for a cleaner visual presentation
• Rate limit rebalancing: Overhauled the feature rate limit configuration with rebalanced free and pro tier caps, renamed feature keys for consistency, and updated titles and descriptions
• Command menu enhancements: Added an Integrations entry to the command menu and renamed “New Chat” to “Create a New Chat” for clarity
• Sentry SDK integration: Enabled Sentry SDK logging in the backend initialization and added the LangGraph Sentry integration for better agent error tracking
• Link preview favicons: Chat message links now display inline favicons next to URLs for better visual context

​

v0.8.1

​

CI/CD Overhaul

• Reusable workflow architecture: Refactored GitHub Actions into composable workflow_call modules, replacing duplicated build and deploy logic with a single reusable pipeline
• Prek migration: Replaced the previous pre-commit framework with prek for faster, more reliable Git hook execution across the monorepo
• Auto-fix commit flow: CI now automatically commits linting and formatting fixes from prek back to the PR branch, eliminating manual re-push cycles
• pnpm caching: Added dependency caching for pnpm in frontend CI jobs, cutting install times on repeat runs
• Concurrency controls: Added and tuned concurrency groups on build triggers to prevent overlapping branch builds from wasting runner minutes
• Action version upgrades: Pinned and upgraded all GitHub Actions to v3 across the board, including setup-gcloud, docker/login-action, and docker/setup-buildx-action

​

Deploy Hardening

• Safe pull on GCP VMs: Deploy jobs now discard local changes on the target VM before pulling latest code, preventing merge conflicts during automated deploys
• Container lifecycle ordering: Deploy now stops running containers before pulling new images and removes the --no-deps flag, ensuring clean restarts with all dependent services

​

Housekeeping

• Lockfile tracking: Removed uv.lock from gitignore so Python dependency locks are now committed and reproducible across environments
• Pre-commit task definitions: Added dedicated mise tasks for running pre-commit checks locally, matching the CI pipeline behavior
• Typing fix: Corrected a TypedDict import to use typing_extensions for broader Python version compatibility

​

v0.8.0

​

Features

• Mise task runner: Adopted mise as the standard task runner and tool version manager for the entire monorepo, replacing ad-hoc scripts with a unified interface for running dev servers, linting, builds, and dependency installs across all apps
• Mprocs multi-process runner: Integrated mprocs alongside mise so developers can start the full stack [frontend, backend, workers] in a single terminal window with multiplexed output
• Integration connect UI: Added an in-chat interface that prompts users to connect required integrations before GAIA attempts to use them, guiding setup inline rather than failing silently

​

Bug Fixes

• Google Docs token handling: Fixed a token refresh issue in the Google Docs tool that caused authentication failures when reading or editing documents

​

Content

• Tool calling blog posts: Published two new blog posts covering tool calling patterns and best practices for AI agents

​

v0.7.0

​

Features

• PostHog analytics: Full product analytics integration with PostHog, enabling event tracking, feature flag support, and usage insights across the application
• Integrations page: New dedicated page where users can browse, connect, and manage all available third-party integrations from a single place
• Expanded Composio integrations: Added support for GitHub, Linear, Slack, and other services through Composio, giving the agent a broader set of actions across your tools

​

Improvements

• Batch sync for Dexie caching: Rewrote the Dexie.js data fetching layer to use batch synchronization, reducing the number of individual requests and improving perceived load times
• Sidebar state management: Sidebar now correctly reflects real-time conversation and navigation state after data syncs, eliminating stale entries and ordering issues
• Component quality fixes: Resolved schema mismatches, corrected TypeScript types, cleaned up duplicate variables, and fixed incorrect prop forwarding across multiple components

​

v0.6.0

​

Features

• Tavily web search: Replaced the deprecated Bing Search API with Tavily, delivering faster, more relevant search results with built-in content extraction
• Firecrawl page fetching: Swapped manual Playwright-based scraping for Firecrawl, providing cleaner and more reliable webpage content for the agent to work with
• Calendar page: Brand-new calendar interface with day, week, and month views, plus updated agent tools for creating, editing, and querying calendar events through natural language
• Sub-graph system: Introduced composable sub-graphs within the LangGraph agent, allowing complex tasks to be broken into modular, reusable execution paths
• Public workflows page: Separate browsable page for community workflows with programmatic SEO, making each workflow individually discoverable by search engines
• Dexie.js caching: Added an IndexedDB caching layer using Dexie.js for conversations, calendar events, and sidebar data, significantly reducing redundant API calls and improving load times
• Global command menu: Rebuilt the keyboard command palette with fuzzy search, categorized actions, and quick access to conversations, workflows, and settings
• Natural language todo input: The todo creation modal now accepts natural language descriptions and parses them into structured tasks with due dates, priorities, and labels

​

Improvements

• Background follow-up streaming: Follow-up actions now stream in the background instead of blocking the UI, making conversations feel faster and more responsive
• Type-safe caching decorators: Backend caching utilities now support generic Pydantic model types, improving correctness and reducing boilerplate when caching structured data
• Backend restructure: Reorganized the backend directory layout with clearer separation between services, configuration, and core logic, plus optimized application startup via parallelized initialization
• Tool registry lazy loading: Agent tools now initialize on demand rather than at startup, reducing cold start time and memory footprint
• Local OAuth icons: OAuth provider icons and model configuration images are now served from local assets instead of external URLs, improving reliability and page load speed
• Markdown-based blog system: Migrated the blog from a CMS-driven approach to file-based Markdown, simplifying content authoring and removing external dependencies
• Onboarding and memory refinements: Streamlined the onboarding flow and improved how the agent stores and recalls user preferences
• Landing page and SEO improvements: Updated landing page layout, added a contact page, and applied SEO optimizations across public-facing routes

​

Bug Fixes

• Fixed calendar events and emails not loading on the chat page after initial navigation
• Fixed Windows compatibility issue where uvloop was incorrectly required on non-Unix platforms
• Fixed a metadata key typo that caused silent execution mode to be ignored during graph streaming

​

v0.5.1

​

Observability and Logging

• Loguru-Based Logging System: Replaced the default logging setup with a structured Loguru integration, including contextual loggers that carry request metadata through the entire call chain.
• PyInstrument Profiling Middleware: Added opt-in profiling middleware powered by PyInstrument, enabling detailed per-request flame graphs for performance analysis in staging and development.
• Stackprinter for Exception Traces: Integrated stackprinter for richer, more readable exception output, making production debugging significantly faster.

​

Developer Experience

• Beartype Runtime Type Checking: Added beartype as a dependency for optional runtime type validation, catching type mismatches early during development.
• Logging Decorators Overhaul: Rebuilt the logging and profiling decorators with cleaner interfaces, removing the legacy session logger utilities and profiler utils in favor of a unified approach.
• Health Check Enhancements: Expanded the health endpoint to surface project metadata and version information, simplifying deployment verification.

​

Agent and UI Polish

• Refined Agent Prompts: Updated the core agent system prompt to better define GAIA’s identity, mission, coaching style, and follow-up action suggestions.
• Workflow Card Refactor: Extracted a reusable BaseWorkflowCard component, consolidating duplicated card layouts across the workflows feature.
• Composer Keyboard Navigation: Added arrow key navigation to the slash command dropdown for faster command selection.
• Scroll Behavior Improvements: Reworked the chat scroll-to-bottom logic with a dedicated button component, replacing the previous multi-button scroll approach.

​

v0.5.0-beta

​

Sub-Agent Architecture

• Delegated Agent System: GAIA’s core agent can now spawn specialized sub-agents for distinct task domains. Each sub-agent operates with its own tool set and prompt context, reducing hallucination and improving accuracy on domain-specific requests.
• Streamlined Tool Routing: The orchestrator determines which sub-agent handles a request based on intent classification, keeping the main agent lightweight while specialists handle deep tasks.

​

Multi-Tool Execution

• Parallel Tool Calls: The agent can now invoke multiple tools in a single turn, dramatically reducing round-trips for complex requests like “check my calendar and draft an email about the meeting.”
• Unified Tool Output Schema: Restructured how tool call results are stored in the database, supporting arrays of tool outputs per message instead of a single result. This unblocks richer UI rendering of multi-step agent actions.

​

Workflow Triggers

• Event-Driven Workflows: Workflows can now be triggered automatically by external events, not just manual execution. Supported trigger types include scheduled times, incoming messages, and integration webhooks.
• Discord Workflow Integration: Fixed critical issues preventing Discord from sending messages within automated workflows, enabling cross-platform notification chains.

​

Notification Tools

• Agent-Initiated Notifications: GAIA can now proactively send notifications to users through configured channels. The notification tool supports structured payloads with titles, bodies, and action links.
• Notification Data in Chat: Tool messages in the chat schema now carry notification metadata, allowing the frontend to render notification cards inline with the conversation.

​

Token Usage and Credits

• Token Tracking System: Every LLM call now records prompt and completion token counts, attributed to the user and conversation. This data feeds into a credit-based usage system.
• Credit Accounting: Introduced a credit ledger that maps token consumption to a user’s balance, laying the groundwork for usage-based billing and fair-use limits.

​

Personalization

• User Preference Storage: Added a first pass at personalization, storing user preferences and behavioral signals that the agent uses to tailor tone, verbosity, and proactive suggestions.

​

Frontend

• Redesigned Landing Page: Shipped a completely new landing page with a tools showcase section, refreshed hero, and improved responsive layout. Integrated HeroUI components into the Tailwind setup.
• Contextual Loading Messages: Chat loading indicators now display context-aware messages relevant to the task being performed, replacing generic spinners.
• Conversation History with IndexedDB: Chat history is now persisted client-side using IndexedDB, enabling offline access to recent conversations and faster initial loads.
• Accessibility Improvements: Added aria-labels to interactive elements and iframes across the landing page and mail composer.

​

Infrastructure

• CI/CD Pipeline: Introduced automated build and deploy workflows with Docker image publishing, GCP deployment scripts, Discord status notifications, and health check verification.
• Cross-Platform Setup Script: Rewrote the project setup script to work reliably on macOS and Linux, with improved virtual environment creation and validation.
• Lazy-Loaded Dependencies: Converted all heavy Python imports to lazy loading, reducing cold start time and improving the self-hosting experience.

​

v0.4.0-beta

​

Workflows

• Workflow engine launched with a full UI for creating, managing, and executing multi-step automations, including trigger and integration selection
• Workflow error boundary added to gracefully catch and surface runtime failures in the workflow interface
• Workflow worker optimizations with dedicated update functions for steps and metrics, reducing unnecessary database writes
• Database cleanup script introduced to purge stale and invalid workflow steps automatically

​

Integrations

• Composio tool platform integrated, giving GAIA access to hundreds of third-party actions and triggers out of the box
• Dodo Payments fully wired up with webhook handling, signature verification via Standard Webhooks, subscription management, and a post-checkout success modal with auto-redirect
• Google OAuth overhaul with unified integration handling, improved scope management, and server-side caching for integration configuration

​

Observability

• Sentry added across both frontend and backend for error tracking, performance monitoring, and a global error boundary
• PostHog configured for product analytics on the frontend

​

Performance

• Application startup optimized across three successive passes: secrets extraction, parallelized MongoDB initialization, and streamlined lifespan management, cutting cold-start time significantly
• ARQ queue for email processing implemented so inbound mail webhooks are handled asynchronously instead of blocking the request cycle
• Cloudinary uploads parallelized alongside summary generation for faster file-attach flows

​

Chat and UX

• Cerebras AI model added as a selectable LLM provider with dedicated configuration
• Chat stream reliability improved with abort controller management, save-on-interrupt for incomplete conversations, and a fix for the infinite “error in chat stream” loop
• ScrollToBottomButton and card stack UI introduced for smoother navigation in long conversations and a new dashboard grid with upcoming events and unread emails

​

Other

• Resend audience sync so new signups are automatically added to the email marketing list
• Databuddy SDK integrated into the app layout for in-app analytics and feedback

​

v0.3.0-beta

​

Features

• Calendar Integration: Full calendar support with event fetching, creation, deletion, and a dedicated Calendar page with loading states and empty state handling
• Recurring Events: Create repeating calendar events with recurrence rules directly from the event creation dialog
• Email Composition: Compose, send, and reply to emails from within GAIA, with a dedicated mail layout and improved email body rendering
• Email Threads: View and navigate full email threads with expandable thread cards and click-to-search for deeper thread fetching
• Unified Todos API: Consolidated todos endpoint with optimized counts, inline editing across all pages, completion filtering, and date-grouped sorting
• Workflow Generation: Attach AI-generated step-by-step workflows to any todo, with real-time status updates and timeout handling
• Follow-Up Actions: GAIA now suggests contextual follow-up actions after completing tasks or conversations, with dedicated UI components
• WorkOS Authentication: Migrated login and signup to WorkOS with OAuth redirection, token repository for Google auth tokens, and a polished redirect loader
• Integration Connection Handling: Declarative integration decorators with scope validation, connection-status indicators, and guided prompts when a tool requires an unconnected service
• Notifications System: Unread notification count badges in the sidebar, bulk mark-as-read, and an enhanced notifications page
• Composer Context: Shared Composer context lets any component append text to the chat input, enabling seamless cross-feature interaction
• Public Beta Landing Page: Redesigned landing page with updated navbar, pricing section, blog post modal, and YouTube integration

​

Improvements

• Chat Streaming: First message now renders immediately during streaming instead of waiting for the full response
• Todo Caching: Smarter cache-first fetching reduces redundant API calls and eliminates edit lag on todo checkboxes
• Goal Management: Improved goal caching logic, updated goal data types, and better roadmap task filtering
• Sidebar Overhaul: Refreshed sidebar styling with accordion-based chat list, system-purpose chat icons, and cleaner navigation
• Base Message Types: Introduced shared base message types to eliminate redundant type management across chat and user bubbles
• Google Docs Markdown: Google Doc creation now accepts Markdown input for richer document formatting

​

Bug Fixes

• Reminder Timezones: Fixed timezone handling so reminders fire at the correct local time
• Toast Deduplication: Eliminated duplicate toast notifications triggered by rapid actions
• WebSocket Error Handling: Improved resilience for WebSocket disconnections and calendar event deletion race conditions
• URL Validation: Tightened URL validation logic in the metadata preview hook to prevent false positives
• Slash Command Dropdown: Corrected z-index and positioning so the dropdown renders above all overlays

​

Infrastructure

• 100% Mypy Compliance: Achieved full type safety across all 232 Python source files
• Infisical Secrets Management: Integrated Infisical for centralized secret injection across environments
• Docker Compose Profiles: Added service profiles and corrected backend port mapping for streamlined local development
• Google Analytics: Integrated GA tracking across the web application
• React Strict Mode: Enabled strict mode to surface lifecycle and side-effect issues during development
• CI Quality Checks: Added conditional frontend checks and ESLint autofix to the CI pipeline
• License Update: Moved from PolyForm Shield to PolyForm Strict license

​

v0.2.1-beta

​

Documentation

• Code of Conduct: Added community Code of Conduct for open-source contributors
• Branch naming: Corrected documentation references from main to master

​

v0.2.0-beta

​

Infrastructure

• Release automation: Configured Release Please for automated versioning, changelog generation, and GitHub releases from the master branch
• Manual release dispatch: Added support for maintainers to trigger releases on demand
• Environment example file: Added .env.example to help new contributors set up their local environment quickly
• Docker image pinning: Updated backend container to pull the latest image from GitHub Container Registry

​

Improvements

• Calendar component: Rebuilt with proper event handling, loading states, and a cleaner layout
• Email composition data model: Restructured for clarity, making the mail tool easier to extend

​

Documentation

• Contributing guide and security policy: Published for open-source readiness

​

v0.1.0-beta

​

Features

• Todo Management: Full task management system with create, edit, delete, and list operations directly from chat, organized by projects and labels with semantic search
• Goal and Roadmap Generation: Set high-level goals and let the AI generate detailed roadmaps with actionable steps and progress tracking
• Persistent Conversation Memory: Mem0-powered memory system that retains user preferences, past conversations, and key details across sessions, with a dedicated settings panel
• Document and File Handling: Upload PDFs and documents for summarization, analysis, and question answering using retrieval-augmented generation
• Secure Code Execution: E2B sandbox integration enabling the assistant to execute code, analyze data, and generate dynamic charts and visualizations
• Email Thread View: View full email threads, compose new emails with AI assistance, and manage recipients directly in the chat interface
• Calendar Management: Full create, read, update, and delete support for Google Calendar events, including recurring events and timezone handling
• Google Docs Integration: Create and query Google Docs, Sheets, and Slides directly within the chat
• Subscriptions and Payments: Pro subscription plans with tiered features, powered by Razorpay for secure payment and subscription management
• Guided Onboarding: Personalized onboarding experience that tailors the assistant to your needs and communication style from signup

​

Improvements

• Redesigned Interface: Complete UI overhaul of the chat interface, landing page, settings pages, and sidebars for a modern, intuitive experience
• Smarter Agent Responses: Improved context understanding, tool usage accuracy, and follow-up action suggestions after each tool interaction
• Notification System: Proactive notifications for events, reminders, and user engagement with bulk mark-as-read support
• WorkOS Authentication: Migrated to WorkOS for robust and secure OAuth 2.0 authentication flows
• Streamlined Settings: Reorganized settings section for managing profile, connected integrations, memory, and subscription in one place

​

Infrastructure

• Monorepo Architecture: Full project restructure into an Nx monorepo separating frontend and backend for scalable development and deployment
• Docker Optimization: Multi-stage Docker builds with dependency caching for smaller images and faster build times
• CI/CD Automation: GitHub Actions workflows for automated linting, quality checks, builds, and dependency management
• Database Stack: PostgreSQL, MongoDB, Redis, and ChromaDB integrated to handle relational data, documents, caching, and vector search respectively
• API Rate Limiting: Tiered rate limiting by subscription plan with enhanced secret management via Infisical