# Product Updates ## April 21–28, 2026 *** **1.** `Feature` — **MCP Raw Config Support** Users can now provide a fully custom raw JSON MCP server configuration, unlocking support for any MCP server regardless of transport type. * Supports HTTP, SSE, and Raw config types — selectable when adding or editing an MCP server * Raw mode provides a JSON editor directly in the UI for pasting full server config * Existing HTTP/SSE connections are unaffected Changes in: `Backend` `Frontend` `Agent` *** **2.** `Feature` — **GitLab CLI Integration** The AI agent can now authenticate to GitLab and run `glab` commands, mirroring the existing GitHub integration. * Supports both GitLab SaaS and self-managed instances * Credentials are scoped per session and cleaned up automatically when the session ends * Multiple GitLab hosts can be configured via workspace scopes Changes in: `Agent` *** **3.** `Feature` — **ASG Multiple Instance Types** Auto Scaling Groups can now be configured with mixed instance policies for better cost and availability control. * Choose multiple instance types or define requirements by vCPU and memory range * Control the split between on-demand and spot capacity in the advanced settings * Existing single-instance ASG configurations continue to work without changes Changes in: `Frontend` *** **4.** `Feature` — **Slack Integration Management for Tickets** Service Desk tickets can now be linked to a Slack channel and thread directly from the ticket panel. * View and update the Slack channel and thread associated with any ticket * Copy a ready-made association command to link an existing Slack thread * Connection status is visible at a glance from within the ticket management panel Changes in: `Frontend` *** **5.** `Feature` — **Agent Sandbox Enhancements** Agents can now access workspace and project directories automatically, and run Python-based MCP servers out of the box. * Workspace and project folders are mounted into the agent sandbox based on the platform context — no manual path configuration needed * Python MCP servers using the `uvx` pattern (e.g. `uvx mcp-grafana`) are now supported * The sandbox can be disabled via environment variable for deployments that require it Changes in: `Agent` *** ## April 7–14, 2026 *** **1.** `Feature` — **Cross-Workspace Report Sharing** Share AI reports with other workspaces directly from the report menu. * Reports can be shared with one or more workspaces via a multi-select dialog; shared reports appear in recipient workspaces with a clear indicator * Recipients can remove themselves without contacting the report owner Changes in: `Backend` `Frontend` *** **2.** `Feature` — **Workspace & Project Secrets Management** Sensitive credentials can now be stored securely at the workspace, project, and ticket level. * Secrets are encrypted at rest using AES-256 * UI provides dedicated modals for managing secrets across workspaces, projects, and service desk tickets Changes in: `Backend` `Frontend` *** **3.** `Feature` — **Dashboard Categories & Templates** AI dashboards are now organized into categories with tabbed navigation and publish controls. * Dashboards grouped under categories and navigated as tabs with configurable per-dashboard timeouts * Templates support a publish/unpublish workflow; admins can promote templates as global * Drag-and-drop support for dashboard organization, with YAML editing and link/investigation views Changes in: `Backend` `Frontend` *** **4.** `Feature` — **Provider Credential Encryption & Custom Fields** Provider credentials are now encrypted at rest and support structured, typed custom field definitions. * Supports arbitrary custom key-value fields with types: string, secret, certificate, and more * UI provides a scrollable modal for managing structured credentials with inline editing Changes in: `Backend` `Frontend` *** **5.** `Feature` — **Private Git Repository Skills** AI skills can now be loaded directly from private Git repositories. * Skills support a new Private Git Repository format type with configurable org, repo, branch, and folder path * Admin UI includes a Git repository configuration form within the Add/Edit skill workflow Changes in: `Backend` `Frontend` *** **6.** `Feature` — **Slack Integration** The AI agent can now receive and respond to messages from Slack. * Messages from Slack are processed with Slack-specific instructions * The agent can choose to stay silent on Slack when appropriate Changes in: `Agent` *** **7.** `Feature` — **Azure AI Foundry Support** The AI agent can now connect to Azure AI Foundry as a backend with support for multiple Azure scopes. * Agents on Azure infrastructure can use Azure AI Foundry as the AI provider * Multiple Azure scopes supported per agent configuration Changes in: `Agent` *** **8.** `Enhancement` — **Bulk Ticket Creation Redesign** Creating multiple AI tickets at once is now faster and more configurable. * New split-pane layout separates task selection from per-task configuration (persona, skills, permissions) * "Apply to checked" batch mode lets you configure multiple tasks at once * Tickets auto-trigger AI execution immediately upon creation Changes in: `Frontend` *** **9.** `Enhancement` — **Prompt Suggestion Auto-Submit** Clicking a prompt suggestion now immediately submits the request when a scope is active. * Suggestions auto-submit when a scope is already selected, removing an extra click * If no scope is selected, the scope dropdown highlights with an inline error message Changes in: `Frontend` *** **10.** `Enhancement` — **Persona Skills Provisioning** Skills defined on personas are now included when provisioning agent skills. * Persona-level skills merged into the main skills list before provisioning * Duplicates deduplicated to prevent double-provisioning Changes in: `Agent` *** **11.** `Enhancement` — **Optional Credentials in Scope** Credentials are no longer required for all scope types — only where they are necessary. * Cloud, Kubernetes, and Source Control provider types still require credentials * Other provider types can be scoped without credentials, reducing configuration friction Changes in: `Backend` *** **12.** `Enhancement` — **Azure AKS Credential Types** Azure AKS clusters can now be configured with Service Principal or Managed Identity authentication. * Two new credential types added: Service Principal and Managed Identity Changes in: `Frontend` *** ## March 31–April 7, 2026 *** **1.** `Enhancement` — **Keycloak SSO Login** Enterprises using Keycloak can now log into the AI helpdesk with their existing credentials. * Keycloak added as a supported identity provider alongside Microsoft SSO * Extra Keycloak parameters (realm, client ID, etc.) configurable via Helm Changes in: `Backend` *** **2.** `Enhancement` — **Configurable Google Login** Admins can now disable Google login entirely, giving tighter control over which authentication methods are available for your deployment. * Google login can be turned off independently per deployment * Useful for enterprises that require SSO-only access policies Changes in: `Backend` *** **3.** `Feature` — **Azure Deployment Support** The AI helpdesk can now be deployed on Azure infrastructure, expanding support beyond AWS. * Azure-specific credential and configuration handling added to the backend * Service Principal and Managed Identity authentication supported for AKS Changes in: `Backend` *** **4.** `Feature` — **Agent Dashboard Integration** Agent activity now feeds directly into dashboard metrics, giving a live view of what agents are doing across your workspaces. * Dashboard and agent views are more tightly integrated * Supports ticket and link views within dashboard panels Changes in: `Backend` `Frontend` *** **5.** `Enhancement` — **Pulumi CLI Support** The agent now has Pulumi CLI bundled, enabling infrastructure-as-code workflows using Pulumi. * Works alongside existing tools like `kubectl`, `aws`, and `gcloud` * Allows agents to manage cloud infrastructure declaratively via Pulumi Changes in: `Agent` *** **6.** `Enhancement` — **Multi-Origin Ticket Support** Tickets can now be created from and attributed to multiple sources — Slack, API, web UI — with origin tracked on each ticket. * Enables better routing, filtering, and audit trails across channels * Origin data available for reporting and analytics Changes in: `Backend` *** **7.** `Feature` — **AI-Assisted Markdown Editor** A new standalone markdown editor with built-in AI editing support is now available within the platform. * Useful for drafting and refining system prompts, project descriptions, and documentation * AI can assist with editing and improving markdown content inline Changes in: `Frontend` *** **8.** `Enhancement` — **Project Task as Opening Chat Message** When creating a ticket from a project, the task description is now automatically posted as the first message in the conversation. * The agent immediately sees the full context without requiring the user to re-type it * Reduces friction for project-driven ticket workflows Changes in: `Frontend` *** **9.** `Enhancement` — **Workspace Reports Enhanced** Workspace reports now include project files and ticket names, giving a more complete view of all activity and output within a workspace. * Project files included in report generation * Ticket names surfaced for better report organization Changes in: `Backend` *** **10.** `Enhancement` — **Cost Optimization: Smarter Agent Planning** The agent's planning and spec-writing prompts now only activate for ticket types that actually require them, reducing unnecessary token usage. * Cuts cost on routine tasks without affecting quality on complex ones Changes in: `Agent` *** ## March 24–31, 2026 *** **1.** `Feature` — **Stop Generation Mid-Response** You can now stop the AI agent while it's actively generating a response or running a task. * A stop button appears during streaming and immediately cancels the in-progress task * Works across both chat streaming and background agent task execution * No need to wait for a long task to complete before starting a new one Changes in: `Agent` `Frontend` *** **2.** `Feature` — **Scheduled Ticket Execution** Tickets can now be scheduled to run automatically at a defined time — useful for recurring tasks, nightly checks, or timed automation without manual triggering. * Schedule configured at ticket creation or update time * Runs the ticket silently at the scheduled time Changes in: `Backend` *** **3.** `Feature` — **Microsoft SSO Login** The AI helpdesk now supports Microsoft Single Sign-On for authentication, allowing users to log in with their existing Microsoft/Azure AD credentials. * Integrates with existing Microsoft/Azure AD identity providers * No separate password or account creation required Changes in: `Backend` *** **4.** `Enhancement` — **Multiple Personas per Workspace** Workspaces now support multiple personas, giving more flexibility in how the AI behaves across different use cases within the same workspace. * When creating a ticket, you can choose which persona to use for that conversation * System prompt field removed from ticket creation — persona handles this instead Changes in: `Frontend` *** **5.** `Feature` — **Skills on Tickets** Skills can now be assigned to tickets at creation time and managed throughout the ticket lifecycle. * Gives precise control over which capabilities the agent can use in each conversation * Skills are protected from accidental removal in the selection UI Changes in: `Frontend` *** **6.** `Feature` — **GRC (Governance, Risk & Compliance) Provider Tab** A new GRC provider category tab is now available in the admin providers panel, joining the existing IT and GTM categories. * Allows GRC tool integrations to be organized separately from IT/GTM providers Changes in: `Frontend` *** **7.** `Feature` — **Auto Ticket Title Generation** Tickets now automatically generate a descriptive title based on the conversation content, so you no longer need to manually name every ticket. * Title is generated by the agent after the first exchange * Displayed in the ticket list and header immediately after generation Changes in: `Agent` `Backend` *** **8.** `Feature` — **AKS (Azure Kubernetes Service) Provider Support** The agent now supports Azure Kubernetes Service as a provider type, expanding the range of cloud infrastructure it can work with. * AKS clusters can be connected and managed via the agent * Works alongside existing AWS and GCP provider support Changes in: `Agent` *** **9.** `Feature` — **GCP System Credential Support** GCP system-level credentials are now supported, enabling deeper integration with Google Cloud infrastructure without per-user credential setup. * System credentials apply across the workspace rather than per user * Simplifies GCP-heavy deployments Changes in: `Backend` *** **10.** `Feature` — **Reports & Analytics API** New reporting APIs are now available, laying the groundwork for analytics and reporting features within the AI helpdesk. * Provides the data foundation for future reporting dashboards * API-first design allows external tools to query report data Changes in: `Backend` *** **11.** `Enhancement` — **HTML Artifact Viewer Improvements** HTML artifacts can now be viewed in fullscreen mode, and external links show a confirmation before navigating away. * Dedicated expand button opens artifacts in a fullscreen overlay * Prevents accidental navigation away from the current session Changes in: `Frontend` *** ## March 17–24, 2026 *** **1.** `Feature` — **Real-Time Streaming Chat** AI responses now stream to your screen as they are generated, rather than waiting for the full response to complete. * Fixed intermittent cases where streaming would stall or produce malformed output * Collaborative messages from other users in the same session now also appear in real-time Changes in: `Backend` `Frontend` *** **2.** `Feature` — **System Prompts for Workspaces, Projects & Personas** You can now define custom system prompts on workspaces, projects, and personas to shape how the AI behaves in that context. * System prompts render with full markdown formatting * Viewable inline on the detail page; longer prompts open in a modal for easier reading * Prompts defined at each level are automatically passed to the agent during conversations Changes in: `Backend` `Frontend` *** **3.** `Feature` — **Long-Lived API Tokens** Users can now generate personal API tokens to authenticate with the AI helpdesk programmatically — useful for integrations and automation without interactive login. * Tokens have configurable expiry dates; the last 4 characters are shown as a hint for easy identification * Tokens can be revoked (active) or permanently deleted (expired/revoked) by the user * Admins can view and manage all tokens across users from the Access Control panel Changes in: `Backend` `Frontend` *** **4.** `Enhancement` — **Ticket Name Prefixes** Workspaces now support a short name field used as a prefix when generating ticket names, making tickets easier to identify across workspaces. * Short names must be unique within a workspace and start with a letter * Editable on both create and edit workspace forms Changes in: `Backend` `Frontend` *** **5.** `Feature` — **Ticket Summaries** Tickets now automatically generate a compacted summary of the conversation, giving you a quick overview without scrolling through the full history. Changes in: `Backend` *** **6.** `Feature` — **Edit Ticket Scopes from Chat** You can now reassign which integrations/tools (scopes) a ticket has access to directly from the chat interface — no need to recreate the ticket. * At least one scope must remain selected * Changes take effect immediately with success feedback Changes in: `Backend` `Frontend` *** **7.** `Enhancement` — **User Identity in Chat Messages** Chat messages now show who sent them — display name (with fallback to email) and a color avatar. * Makes it easier to follow conversations with multiple participants * Mentions in activity chat now show user avatars and display names in the dropdown Changes in: `Backend` `Frontend` *** **8.** `Enhancement` — **Unread Message Badge** The activity sidebar now displays an unread message count badge so you can see new messages arriving without keeping the chat panel open. * Badge updates in real-time as new messages arrive * Clears when the chat panel is opened Changes in: `Frontend` *** **9.** `Feature` — **Observability & Security Dashboards** New dashboard views for Observability and Security metrics are now available in the admin panel. Changes in: `Backend` `Frontend` *** **10.** `Enhancement` — **Command Output in Chat** Long command outputs in chat are now collapsed by default for a cleaner view, with a clear expand control to see the full output when needed. * Expandable/collapsible with hover tooltips Changes in: `Frontend` *** ## DuploCloud AI Suite Release — September 5, 2025 ### Overview We're excited to introduce **DuploCloud AI Suite**, a comprehensive artificial intelligence platform that transforms how you manage cloud infrastructure. This inaugural release brings intelligent automation to DevOps workflows through specialized AI Agents that work alongside your team to solve complex infrastructure challenges. ### What's New #### AI Studio **Build and Deploy Custom AI Agents** AI Studio provides everything you need to create, customize, and deploy AI Agents tailored to your organization's needs: * **Agent Specification Builder** - Define your Agent's capabilities and behavior * **Vector Database Support** - Enable Agents to access and search your documentation and knowledge base * **One-Click Deployment** - Automatically deploy Agents to Kubernetes * **Agent Registration** - Seamlessly register deployed Agents into HelpDesk for immediate use #### HelpDesk **Intelligent Support with Human Oversight** HelpDesk transforms traditional IT support through AI-powered assistance while maintaining complete human control: **Smart Ticketing System** * Create tickets and get matched with the most appropriate AI Agent * View, search, and filter tickets * Track Agent assignments and ticket status in real time **Collaborative AI Assistance** * **Human-in-the-Loop Approval** - All Agent actions require your explicit approval before execution * **Shared Canvas** - Work side-by-side with AI Agents in a collaborative workspace * **Interactive Terminal** - Share a live terminal where both you and the Agent can run commands * **File Collaboration** - Agents can generate configuration files that you can edit and refine * **Rich Content Support** - Full Markdown rendering with interactive Mermaid diagrams **Advanced Security Controls** * **Double Approval** - Sensitive commands require additional confirmation for extra security * **Credential Proxying** - Agents use your permissions, never their own * **Smart Prompt Suggestions** - Get started quickly with Agent-specific conversation starters **Flexible Access** * **Standalone Interface** - Full-featured HelpDesk application * **Chat Bubble Integration** - Quick access from anywhere in the DuploCloud platform ### Available AI Agents #### Kubernetes Agent Your intelligent Kubernetes troubleshooting companion that can: * Diagnose deployment issues and container problems * Create new deployments and services * Perform cluster maintenance and optimization tasks * Provide expert guidance on Kubernetes best practices #### Observability Agent Powered by DuploCloud's Advanced Observability Suite, this Agent helps you: * Retrieve and analyze metrics and logs for any microservice * Identify performance bottlenecks and anomalies * Get intelligent insights from your OpenTelemetry data * **Coming soon:** Trace analysis and performance profiling #### CI/CD Agent Intelligent pipeline support that automatically: * Monitors your Jenkins and GitHub Actions pipelines * Creates support tickets when builds fail * Attaches relevant logs, configuration, and error details * Provides expert troubleshooting assistance for pipeline issues #### Architecture Diagram Agent Transform complex infrastructure into clear visuals: * Generate architecture diagrams using natural language descriptions * Automatically map relationships between services, databases, and infrastructure * Create shareable Mermaid diagrams of your entire technology stack #### Private GPT Agent Secure AI assistance for security-conscious organizations: * Private ChatGPT-like experience using AWS Bedrock * Complete data privacy with enterprise-grade security * Perfect for teams who need AI assistance without third-party data sharing ### Supporting Infrastructure #### DuploCloud Cartography * **Automatic Discovery** - Continuously maps your AWS, Kubernetes, and DuploCloud resources * **Relationship Mapping** - Understands dependencies between microservices and infrastructure * **Custom Dependencies** - Define application-specific relationships for complete visibility #### DuploCloud Presidio * **Data Protection** - Automatically redacts sensitive information in AI conversations * **Customizable Rules** - Configures what data should be protected based on your security policies * **Privacy-First** - Ensures sensitive data never leaves your environment ### Key Benefits **Intelligent Automation** - AI Agents that understand your infrastructure and can take action with your approval **Security First** - Human oversight for all actions with your credentials, never autonomous access **Faster Resolution** - Collaborative workspace where you and AI work together to solve problems **Better Insights** - Automatic diagram generation and intelligent analysis of your systems **Enterprise Ready** - Private AI models with complete data privacy and security controls --- # Agent Instructions: Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://docs.duplocloud.com/docs/product-updates.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.