Skip to main content

Release Notes — v1.4.0

August 25, 2026

This release adds version history for both artifacts and Skills and greets you with personalized chat suggestions based on what you've been working on. It also brings Grok as a model option, much faster report refinement, and project cards that show where work is happening.

New Features​

Artifact Version History​

Artifacts now keep a full version history. Every content change records an immutable version, whether it came from chat, refinement, the editor, or the API, so a refinement that makes a report worse is recoverable.

The Version History panel lives in the artifact page's right rail (desktop). Click a version to view a read-only snapshot rendered exactly like the live page, and use Restore to bring it back. Viewing history requires view-content permission; restoring requires manage-content. Existing artifacts start at version 1.

Learn more: Artifacts Overview

Skill Version History​

Skills get version history too. Every save that changes how a Skill runs (content, model configuration, custom variables, tools, attachments, or web search) records an immutable version with author and timestamp; cosmetic edits don't. The Skill details page gains a Version History section with read-only snapshots of each version.

Execution always uses the live Skill; versions are a record of how the Skill has changed over time, and there's no rollback or diff view yet.

Learn more: Skills Overview

Chat Suggestions​

The chat landing page now offers personalized suggestions for your next conversation, generated from your recent activity: chats, artifacts, documents, and the Skills and Agents available to you. Each suggestion is a ready-to-send prompt, sometimes pointing at a specific Agent or Skill. Clicking one prefills the composer and enables the tools the task needs, and Show new ideas regenerates the set.

Suggestions are cached, so they render instantly and only change when your activity does or you ask for new ones. Everything they draw on is permission-filtered; they never surface activity you can't already see.

This feature is behind a feature flag. Contact your Revelry Ai team to enable.

Activity Snapshot Tool​

The AI can now look up what you've been working on instead of asking you to restate it. A new read-only tool returns a permission-filtered snapshot of your recent chats, artifacts, documents, collections, and recently used Skills, scoped to the current project or your whole workspace. This is the same activity summary that powers chat suggestions, available to the AI mid-conversation.

This feature is behind a feature flag. Contact your Revelry Ai team to enable.

Learn more: MCP Tools

Grok Model Support​

SpaceXAI (Grok) joins the supported LLM providers. Organization admins connect it from the LLM Providers page with an xAI API key, which makes Grok 4.6 available as a chat and Skill model. Admins can also set a reasoning effort per model configuration (Low, Medium, High, or Extra High).

Faster Report Refinement​

Refining a report no longer regenerates the whole document for every change. The refinement agent now applies targeted find-and-replace edits, so small changes complete in seconds instead of rewriting everything, and a full rewrite still happens when the change genuinely calls for one. Edits apply atomically: if an edit doesn't match, nothing changes.

Learn more: MCP Tools

Onboarding Chats Carry Over​

The conversation you have during onboarding, before creating an account, is now saved as you go. When you sign up, it's imported as a real chat in your first project, so you pick up exactly where you left off, and it seeds your first chat suggestions.

Account Transfer Between Instances​

Accounts can now move between Revelry Ai instances. Upgrading from a free instance walks you through export, sign-up on the destination instance, plan selection and payment, and a live import progress screen, with your chats, artifacts, Agents, and Skills carried over.

This feature is behind a feature flag. Contact your Revelry Ai team to enable.

Project Cards That Show Activity​

The Projects page now tells you where work is happening. Each project card shows its chat and artifact counts and a "last active" time, with the exact date and time in a tooltip and an indicator dot for projects active in the last 24 hours. The grid orders projects by most recent activity instead of alphabetically, so your busiest projects surface first.

Global Indicators​

Content provided by the platform is now labeled. Global Skills show a Global pill in the Skills list, and globally available documents, data sources, and collections show the same pill in the Knowledge lists, so you can tell platform-provided content from what your team created.

UI Improvements​

  • New/Edit Project pages. Create a group, category, or collection inline without leaving the form, with typeahead sorting, filter reset, a slimmer category modal, and dark mode popover fixes.
  • Tool execution cards name Skills by slash command. A Skill run shows as /create-artifact on the card instead of a generic "load skill" label, with a link to the Skill's detail page.
  • Failed tool executions show their error. Failed cards expand to show the actual error message, and canceled executions are labeled as canceled instead of reading as failures.
  • Interior page polish. Page tabs, a mobile drawer, mobile tooltips, and collection descriptions.
  • Chat header. The app now uses the full width of large screens, and the default chat puts the project switcher right in the header.
  • Reload prompt after updates. When a new version of the app is deployed, open tabs show a banner offering to reload, so you're never running on stale styles and scripts.
  • Mobile bottom sheets. Mobile flows now open in a consistent bottom sheet, with smoother scrolling behavior behind overlays.
  • Chat layout fixes on mobile. Chat content no longer scrolls behind the input when it's focused, plus fixes for horizontal scrolling, input padding, artifact flyout positioning, and chat width on small screens.
  • Integrations page. The tab subheading no longer takes up a card slot, and cards wrap instead of overflowing on narrow viewports.
  • Onboarding polish. Logo placement, organization name display, and a fix for text overlapping in the chat box.
  • Mobile navigation sizing, nav height, and button color corrections.
  • Sign out and credit widget adjustments in the account menu.

Bug Fixes​

  • Fixed artifact type and category issues when creating and editing artifacts: the category can now be cleared, category selection is correctly scoped to your organization, and the artifact type is retained in the edit view.
  • Long organization names now ellipsize in the organization switcher instead of clipping, with the full name shown on hover.
  • Removed a duplicate tool execution card that could appear while a tool was preparing.
  • Chat suggestions that come back short or generic are automatically retried.
  • Fixed a crash that could make a chat permanently unloadable when a tool call's arguments arrived in an unexpected format.
  • The AI's background helpers (report conversion, image analysis, Skill and workflow runs) now receive the same instance, organization, user, and project context as chat, so their output matches your setup.
  • Global collections no longer show "Unknown" for Created By.
  • Fixed an error when saving Edit Project Settings with certain category selections.
  • Account transfer imports now carry over your last viewed project and no longer leave behind a redundant "Untitled Project."

Under the Hood​

  • Removed the legacy Jira data source and export path.
  • Provider file uploads are now tracked and cleaned up automatically when they expire.
  • Model usage records now capture which user spent the credits.
  • Superadmin tooling for exporting and importing an organization, with a dry-run audit that validates the data before a real transfer.
  • A shared foundation for building HTTP integrations, standardizing tool definitions, request parameters, and OAuth flows across providers.
  • Groundwork for sectioned analysis of large PDFs, splitting a document into page ranges analyzed in parallel.
  • Expanded analytics coverage for artifact creation and updates and for Skill executions.
  • CI pipeline for sandbox template builds, plus dependency updates.

See v1.3.3 for the previous release.