## Why this matters GitLab is evolving beyond being just a DevSecOps platform to becoming an AI-native platform where intelligent agents collaborate seamlessly with development teams. This transformation represents a fundamental shift in how users interact with their development workflows, moving from manual processes to AI-assisted and AI-delegated tasks. By reimagining our platform with AI agents at the core, we enable developers to focus on high-value creative work while agents handle routine tasks, troubleshooting, and process automation. ## Problem to solve Our current platform structure was designed for human users interacting with traditional development tools. As we introduce more AI capabilities, we're experiencing several challenges: * **Fragmented AI experience**: AI features are scattered across different areas without a cohesive interaction model * **Limited agent capabilities**: Current AI interactions are primarily chat-based and don't integrate with the full development lifecycle * **Navigation complexity**: The existing UI structure doesn't accommodate the new paradigm of AI agents working alongside human users * **Workflow friction**: Users need to context-switch between AI assistance and their regular development tasks ## Proposal Transform GitLab into an AI-native DevSecOps platform by introducing a comprehensive agent platform that provides multiple interaction methods and seamlessly integrates AI agents into every aspect of the development workflow. This initiative encompasses: **Core Agent Platform Components**: - **AI Side Panel Evolution**: Transform the current chat interface into a comprehensive AI interaction hub with Chat, Sessions, and Suggestions views - **Agent Catalog Area**: Provide discovery, creation, and management capabilities for AI agents and workflows - **Multi-Modal Agent Interactions**: Enable agent invocation through direct chat, @mentions, assignee selection, event triggers, and slash commands - **Panel-based UI**: Evolve the platform navigation and layout to accommodate AI-first workflows while reducing overall complexity **Key Capabilities**: - Agents can participate directly in issues and merge requests through comments and suggestions - Event-driven agent activation for automated responses to pipeline failures and system events - Flexible agent modes (Collaborate vs Delegate) to match different workflow needs - Session management for tracking and resuming agent interactions across the platform This transformation will establish GitLab as the leading AI-native DevSecOps platform, where intelligent agents work as seamless team members rather than external tools. #### Enable New UI Toggle for Users We can enable `New UI` toggle within user dropdown by running following chatops commands; ``` /chatops run feature set tailwind_container_queries true --user=user_name /chatops run feature set paneled_view true --user=user_name ```