Recent improvements, new capabilities, and fixes across the platform.
Now, when you attach a large image, Intent clearly explains why it can't process it, preventing confusion and frustration. This makes reporting issues and sharing detailed info smoother and more transparent.
We fixed an issue where user identities were lost in bug reports that included images. Now, reporter identities are correctly shown, even with attachments, ensuring accurate tracking.
Now you can attach any file type to conversations through chat or changeset Files tab. All attachments are durable and AI can read them for better context in discussions.
We've fixed an issue where long AI chat responses were cut off. Now, you'll get the full answer without any abrupt endings, improving trust and clarity in our AI chat service.
We fixed an issue where long AI chats would permanently fail with an internal error, stopping the assistant from responding. Now, long conversations can continue without interruption.
We’ve added an Implementation Plan tab and chat mode to create detailed, step-by-step guides for software changes. This helps bridge the gap between design and execution, so developers have clear plans to hand to AI tools for direct implementation.
Now you can use AI coding agents like GitHub Copilot for automatic code changes. Simply toggle it on, and mark changesets for automatic implementation in your CI pipeline.
We fixed an issue where AI responses were missing in newly created changesets. This means your conversations will now look as expected with complete AI replies, improving clarity and trust in our split-changeset feature.
Introducing two easy-to-use copy buttons for specs and changeset details. Users can now quickly share content with deep links, reducing manual effort and errors.
Intent now shows fewer notifications when offline and automatically fetches missed notifications when you reconnect. This makes sure you don't miss important updates and reduces confusion with error messages.
Changing a changeset type doesn't trigger AI updates, leaving specs and descriptions in the wrong format. This can cause confusion and errors in project documentation.
Intent now features five distinct conversation modes for better clarity and focus in different types of work. Each mode has tailored prompts, and a new tool allows seamless switching between them.
Now you can easily mention team members in comments and receive notifications for important updates, assignments, and tasks. This helps keep everyone informed and on track with their responsibilities.
We've introduced team assignments for users to simplify their workflow and make team management easier. Users can now see only their team's work, reducing navigation time and enhancing focus.
Now, teams can be reassigned or cleared while a changeset is in progress, without needing to revert its status. This flexibility helps avoid unnecessary friction when team assignments need to change mid-flight.
We improved artifact versioning by introducing the read_artifact tool to avoid content loss during cross-changeset edits and enforcing lineageId to prevent duplicate artifacts. These changes streamline your version control and ensure accurate updates.
We've fixed an issue where tabs would clip and become inaccessible. Now, all tabs are visible and can be accessed, even on narrow screens.
We've changed how error notifications work. They now stay on the screen until you click to close them. This makes sure important messages don't get missed, improving your experience.
Branch names now correctly transliterate umlauts and other non-ASCII characters, and long titles are truncated more intelligently. This makes branch names easier to read and more traceable for all teams.
We've fixed an issue where the Team filter disappeared when switching to Code view. Now, you can apply team filters in Design view and have them remain visible as you switch to Code view, helping you stay on task without losing your filtering context.
We've refined our bug reporting widget to focus exclusively on bug reports, ensuring higher quality submissions. Now, it only accepts bug reports and conducts codebase research before asking questions, leading to more precise bug fixes.
Intent now lets you manage repositories, pull requests, and changeset statuses directly within your IDE. This means smoother workflows and less switching between tools.
You can now dynamically override credentials for each call to IntentWidget.open(). This flexibility helps intranet teams manage multiple projects and set credentials at runtime, without affecting existing integrations.
We now automatically condense lengthy conversations into summaries, ensuring the model stays within its context limit without losing important details. You'll still see the full conversation history.
Customers can now embed the reporting view in their own intranet for a seamless way to report bugs directly from users' workspaces. This allows users to create bug reports without needing an Intent account, ensuring that only authorized users can submit reports.
We fixed a bug making code blocks invisible in light mode. Now you can read all AI-generated code clearly, improving chat feature usability.
System API keys now have a stable identity linked to a synthetic user, ensuring consistent permissions and reducing security risks.
Now you can deactivate and reactivate user accounts without losing their history. This lets you manage access easily while keeping a record of all contributions.
We’ve made assigning tasks to PRs faster by making the assignee and reviewer slots on PR cards interactive. Now you can assign or clear assignments directly from the PR list without extra clicks, streamlining your workflow.
We've updated the 'Forgot Password' feature to help users who haven't activated their accounts reset their passwords. Now, these users can request a new temporary password via email, ensuring they can access their accounts without needing help from an admin.
We've made the preflight check process clearer with real-time updates. Now, you'll see exactly what the AI is doing during the check, breaking up the wait and giving you a better sense of progress.
We’ve made code blocks easier to read with language-specific syntax highlighting. Now, all your code snippets appear in the correct color, improving clarity and user experience.
We fixed a bug preventing self-hosted Git tokens from being validated or saved. This allows you to use self-hosted instances without issues.
Now you can choose the AI's language output for each project, making it easier to collaborate in your preferred language and reducing translation barriers.
We've added 'Acceptance Criteria' to feature changeset descriptions for clearer completion definitions. This helps everyone understand what's needed to finish a feature, reducing confusion and ensuring everyone is on the same page.
We've reduced database connection timeouts by optimizing how job queues are managed. This change ensures smoother performance and prevents job processing issues, making your software more reliable under heavy load.
Now you can connect to your self-hosted Git instances in Intent, expanding its utility for users who need more control and flexibility over their Git hosting environment.
We've fixed problems causing changesets to misorder when moving sections or manually rearranging them. Now, changesets move to the bottom of new sections and stay put when reordered, improving workflow reliability.
We fixed an issue that caused emojis in changeset names, improving readability and professionalism. Now, changeset names are clean and error-free.
We've added a 'Forgot password?' link to help you regain access if you've forgotten your password. This feature makes it easy to recover your account without support, saving you time.
We've added automatic sitemaps and directives to improve search engine visibility for our marketing site, ensuring all public pages are easily discoverable and indexed.
We fixed search and short ID issues for changesets. Now you can find changesets by their ID and see the correct eight-character ID everywhere it's shown.
We've resolved an issue where switching between conversation and changeset views caused the AI response to appear multiple times. Now, the AI response displays just once for a smoother experience.
Now, the PR list updates instantly when a pull request changes, reducing confusion and improving project management efficiency.
Now, the browser tab title updates dynamically based on your activity, making it easier to identify what page you're on at a glance.
Now, each new conversation has a title prefixed with an emoji that matches its topic, making it much easier to quickly find and select the right conversation in the sidebar.
Intent now uses AI to automatically match changesets with features, reducing manual effort and making the process more efficient. The AI suggests features based on changeset content and asks for confirmation before assigning them.
We've introduced a lightweight way to group and track changesets under named initiatives. This makes it easier to see progress on larger projects and understand what’s in progress, completed, or pending.
Bug changesets now automatically appear in the frontend and list affected repositories, improving user experience when filing bug reports.
We fixed issues with shared links and message input. Now shared links show the right conversation and input is disabled during sending, avoiding message duplicates.
We added database indexes to speed up common queries, reducing response times for frequently used API endpoints. This ensures a smoother experience, especially as your data grows.
Intent now correctly manages bug changesets, avoiding unnecessary documentation and providing accurate guidance, enhancing the overall user experience.
Now you can switch between different workspaces without losing your chat drafts. Your in-progress messages are saved and restored as you move between contexts, helping you stay focused and uninterrupted.
Effort is now calculated at the changeset level for a more accurate assessment. The AI provides a holistic view, reflecting the complexity of the entire work package rather than individual specs.
We've added a new way to manage dependencies between changesets. Teams can now clearly see what needs to be finished before they can proceed, reducing confusion and errors in project workflows.
We've combined the Assignee and Reviewer filters into one Person filter for easier PR navigation. Now, you can quickly find all PRs where a specific person is involved, improving efficiency and reducing complexity.
You can now add extra pull requests to a repository within an active changeset. This helps manage large projects by splitting them into smaller, reviewable parts, all under one changeset.
Now everyone in a chat can see messages and AI responses in real time. This makes teamwork smoother and more effective, without needing to manually refresh the page.
We fixed issues causing silent AI chat crashes, ensuring smoother and more reliable conversations. The AI can now recover from certain errors and provide clearer feedback to improve its performance.
We've redesigned the inner tab bar in the Overview tab to make it cleaner and easier to use. Now, related content is grouped into distinct sections, making it simpler to navigate and manage different types of information without clutter.
Now you can attach files directly to changesets to keep all reference materials in one place. This makes collaboration easier and helps streamline your development process.
Improved visibility for pull requests with assignee and reviewer fields, plus a new PR list view for better oversight and filtering.
Now, switching between conversation history is smoother. The URL tracks your selections, enabling proper back and forward navigation, and the changeset panel now matches or clears based on the selected conversation. This makes it easier to follow your context.
Now you can quickly reference changes with shorter IDs and easily see who created and last updated each changeset, improving clarity and accountability.
Wide tables in chat and specs now fit within containers, with scrolling enabled for easy reading.
We've combined Explore and Design chat modes into one main workspace chat. This makes it easier for you to ask questions, brainstorm, and start changes without extra steps. Your past explore conversations are now easily accessible.
We fixed an issue where database errors caused the backend to crash, ensuring it stays up even with temporary connection problems.
Intent now handles interruptions during the 'Start Work' process more gracefully. Users can now retry without manual branch deletion, ensuring smoother and more reliable workflows.
We modernized the MCP server with a new explore tool for natural-language queries, streamlined codebase exploration, and added a read_file tool for easier file access.
Now, each user’s Git operations are traceable and respect their specific permissions. Plus, a new Getting Started checklist adapts based on user roles to simplify onboarding.
We now show the full name of the message sender in the chat, making it easy to follow conversations with multiple participants. This helps keep context clear, both in real-time and when reviewing past chats.
Worker processes now log only structured messages, making logs cleaner and easier to manage for monitoring and troubleshooting.
We've refreshed our homepage to make Intent's benefits clearer, especially for Product Owners and CTOs. Now, it's more engaging and highlights how Intent enhances software delivery efficiently.
We've refined the changeset detail panel to be more compact and user-friendly, especially for projects with many repositories and pull requests. The repository and pull request lists are now collapsed by default to reduce clutter, and key metadata is displayed side by side for quick access.
We introduced a new tool that lets you split your work into separate, focused changesets as you go. This makes managing complex projects easier, keeps each part clear, and avoids confusion.
We now include the current date in all system prompts, improving the AI's ability to understand time-related queries without needing users to specify the date each time.
Now you can click on specifications and changesets directly in AI chat. This makes it simple to find what you need without searching manually, saving you time.
We’ve introduced an AI pre-flight check to assess readiness when starting a new changeset. This replaces the old blocking system and shows AI-driven insights without preventing work. You can now decide whether to proceed despite flagged issues.
We fixed issues causing the AI stream to crash with `search_by_tags`, ensuring smooth conversations and clearer error messages when things go wrong.
We've fixed issues with artifact deletion and version history. Now, deleted artifacts can't compromise completed records, outdated files won't affect merged pull requests, and diffs will always compare the correct versions.
Our AI can now find and use external documentation to draft more complete specifications. This helps fill in gaps when specs involve third-party tools or integrations.
We've added a new Terms of Service page to our marketing site for better legal compliance and transparency. Users can easily find it from the main site footer.
We've locked changeset names once work starts to prevent mismatches, improving clarity and reliability.
We've added new legal pages to our marketing website and app to meet regulatory requirements and build user trust. Users can now easily access company information and data processing agreements, making compliance simpler and more transparent.
Now you can navigate all tabs easily, even with many specs. The tab bar scrolls horizontally, and important tabs keep their labels longer to improve workflow.
Now you can read both new and older parts of AI-generated chats without being automatically snapped back to the bottom. This improves your ability to review past messages seamlessly.
We added a warning to remind you of any unchecked todos before starting work on a changeset. This helps ensure you don't miss important tasks, without blocking your workflow.
We've added clear notifications to let you know when you're offline or back online. This helps you understand your connection status so you can take action quickly.
Now you can update repository details like name, URL, branch, and credentials directly in the project settings. Changes are saved without disrupting your work and only affected files are re-indexed.
Deleting a repository now marks it with a timestamp instead of removing it, preserving its links and history while freeing up storage. Deleted repositories are hidden from active views but remain visible in changeset history with a 'Deleted' badge.
Now, newly added repositories start indexing right away, eliminating the 15-minute wait. This change offers a much faster experience and better responsiveness for all users.
Now, changing the type of a changeset updates it instantly, and linked artifacts regenerate seamlessly for a consistent experience. This makes managing changes easier and more reliable.
We've streamlined changeset descriptions to give you a clearer, high-level overview. No more confusing per-repository details; now you get a single, easy-to-understand summary for each changeset.
We've added a new MCP server to allow seamless integration of AI tools and CI pipelines with Intent. Now engineers can access Intent's data without leaving their tools, boosting productivity and reducing context switching.
We now validate repository URLs and branches before saving. Errors are shown immediately, so you can fix mistakes right away, improving reliability and user experience.
We've added a 'todos' list to track non-code tasks required for a changeset. This helps ensure all necessary steps are completed before marking a changeset as done.
We've standardized refactor descriptions to match bug and feature formats, making them easier to understand and use as technical records. This consistency helps everyone stay on the same page.
We now have a clear way to describe bug fixes and new features, making it easier to understand changes and their impact. This helps our team work more efficiently and transparently.
Now you can delete Draft and Backlog changesets, search results exclude cancelled specs, and we've renamed 'Not Implemented' to 'Cancelled' for clarity. Plus, a new colour-coded status badge on the Overview tab helps you quickly see changeset statuses.
We now show clear, consistent toast notifications for failed actions, like updates or deletions, to ensure users are always informed about errors, improving their experience and reducing support queries.
You can now reliably reorder changesets with drag-and-drop. This fix prevents them from snapping back after a refresh.
We’ve added a separate Status column to make it easier to see the indexing status and time of repositories. This change clears up the main table and helps you quickly understand your project’s status.
Images in chat messages now stay saved and visible when you revisit conversations, helping you recall important details.
We now support image attachments and raw document pastes to help you convey complex ideas more effectively. Attach up to five images per message or paste large blocks of text directly into the chat to streamline your communication with the AI.
Inline code blocks now appear correctly in AI chat and editor without unwanted bold formatting or backticks, improving clarity and professionalism.
Now you can view detailed changeset descriptions directly in your linked repositories, making it easier to understand changes without leaving your codebase.
We've added a 'Description' tab to the changeset artifact panel to make changeset summaries visible. This helps you understand the purpose and impact of each change.
Intent now generates specs that accurately reflect the current state of the software. This ensures that documentation stays up-to-date and serves as a reliable reference for developers, making it easier to understand what the software does today.
We've fixed an issue where AI chat replies were lost on page reload. Now, in-progress replies stay intact even if you refresh the page, ensuring important context isn't lost.
We've fixed the distracting red underlines in chat and editor areas. Now, technical writing is smoother and less distracting.
We fixed how Markdown tables display in the spec viewer, so they now appear structured and clear. This makes spec content easier to read and understand.
We've added a 'Close as not implemented' action to manage deprioritized changesets. It replaces the delete button and keeps records clean without deleting, while also abandoning linked pull requests.
We removed unnecessary tools in explore mode and fixed bugs in report mode, ensuring accurate behavior and correct focus on bug reporting.
We now save changeset summaries and task lists, making them always up-to-date. Also, pull request descriptions now include clickable links to specs, making reviews easier.
Intent now prevents automatic repository additions during active changesets. It informs users and lets them add repositories manually, ensuring intentional changes and avoiding scope creep.
Intent now automatically generates and updates pull request descriptions using AI. This gives reviewers a detailed overview without needing to leave their git provider, streamlining the review process.
We've fixed a delay in visual feedback during AI artifact generation, now showing an indicator immediately when the AI starts. This helps reduce uncertainty and improves the user experience.
Spec files now correctly overwrite instead of creating duplicates, ensuring only the latest version is present in the repository.
We added a toggle to easily compare current specs with previous versions. This helps you quickly see changes and understand updates, improving clarity and assessment.
Now, once all pull requests are merged, changeset statuses update to 'Done' without needing a page refresh. This makes sure your view always matches what's happening behind the scenes.
We've streamlined navigation with a bottom-to-top flow in the changes panel and added clearer interactions for the progress stepper to help you manage work more intuitively.
We've added collapsible help panels in the management tabs to make it easier for users to understand and navigate the system. The panels offer guidance and explanations, always visible on larger screens, but hidden on smaller ones to keep the interface clean.
We added a real-time progress indicator to show AI's spec creation and saving progress, improving transparency and user trust in the system.
We optimized our code indexing to save on unnecessary tokens and introduced a new `.intentignore` file for developers to control what gets indexed, improving search relevance and efficiency.
We've fixed an issue where finalized specifications could still be edited. Now, once a changeset is complete, its specifications are automatically locked to prevent any changes, ensuring accurate and reliable data.
We've created a dedicated portal and user role for easy bug reporting. This makes it simple for users to report issues and request features without accessing the full product, keeping everything secure.
Now your specification files stay up-to-date with Intent updates, ensuring developers always have the latest information without needing manual updates.