Informity AI v0.13.3 is out
v0.13.3
This release adds first-class Markdown export for chats and answers, standardizes the export API contract for future formats, and includes browser/dev stability polish.
What’s included
- Added Export Chat in the chat header to export full chats as Markdown.
- Added per-message export for assistant answers (export current answer as Markdown).
- Added deterministic Markdown rendering with structured output (title, metadata, template body, and normalized sources).
- Added a unified export endpoint:
GET /api/chat/chats/{chat_id}/export with standardized scope, template, include_frontmatter, and format.
- Kept backward compatibility for existing Markdown callers via
GET /api/chat/chats/{chat_id}/export/markdown.
- Standardized a forward-compatible PDF path:
format=pdf now returns 501 Not Implemented until PDF rendering is shipped.
- Improved browser dev consistency by pinning frontend/proxy loopback behavior (
127.0.0.1, strict port) to avoid localhost mismatch issues.
- Updated Settings iconography polish for Check for Updates.
- Added and expanded route/rendering tests for export behavior and contract coverage.
Notes
- Compatibility: no migration required; existing markdown export integrations remain functional.
- Export destination: download folder is controlled by browser/WebView/OS settings; app controls filename and content payload.
- Scope: export is per full chat or per single assistant answer (batch export remains out of scope for this release).
- QA status: validated with full quality gate (
make qa-full) including backend tests, frontend tests, lint, typecheck, and production build.
Tag: v0.13.3
1 Like
@anon7180143 Homebrew distribution is now available: GitHub - informity/homebrew-tap: Informity Homebrew · GitHub
You can install and upgrade Informity AI through this tap now. I’ll work on submitting to the official Homebrew repository down the line, but this should get you going in the meantime. Let me know if you run into any issues.
3 Likes
Fantastic! Thank you so much!
3 Likes
Informity AI v0.14.0 is out - now with Document Translation!
Informity AI v0.14.0 is a major expansion release. It adds a full document translation workflow, clearer Specializations, and a smoother local-first experience across the app.
What’s new
- Translate documents locally: new Translate workspace with OCR support for images, primary/additional language settings, tone selection, and export to Markdown or plain text.
- Keep translation history accurate: finished translations remember the language and tone they were created with, even if you later change settings.
- Use Specializations instead of Roles: the UI now uses clearer specialization language for chat expertise and the feature menu.
- Cleaner Files and Settings screens: better spacing, consistent buttons, improved language selection, less clutter from local upload rows, and more predictable section layout.
- Better local model performance controls: CPU Performance now lives in System and applies to local models used by Chat and Translate.
- More polished app behavior: improved markdown rendering, better translation state recovery, and clearer handling of new vs existing translation sessions.
Why it matters
- You can translate documents end to end without leaving the app.
- The UI is easier to understand and more consistent across screens.
- Chat, Translate, and Specializations now fit together as one coherent workflow.
- The app is more reliable across local and Ollama-backed models.
Compatibility
- Legacy role-based settings and chat metadata are still supported for existing users.
- New translations start from your current defaults; old translations keep their original metadata.
Thank you, CloakedNetizen for feature suggestion!
2 Likes
You mentioned my username, so I will be forthcoming and state that I am tracking this topic for Linux support.
I would love to deliver Linux distribution (.deb, .rpm) but I do not have a viable Intel computer to build the app on. What platform (x86? arm64?) and Linux distribution are you on? Perhaps I could supply you with instructions and you could build this yourself? Aside from some dependencies, the build should be just one command. Please let me know if you are interested in trying this out. Thanks!
1 Like
x86, Kicksecure (Debian derivative), containerize Informity with an OCI image for your own architecture first and you got a deal.
Informity AI v0.14.1 is out
Informity AI v0.14.1 is a feature-and-fix release since v0.14.0, with the main improvements landing in model cleanup, translation behavior, and general app polish.
What’s new
- Better model cleanup: legacy local model files can now be removed correctly even after you upgrade to a newer alias of the same model family.
- More stable translation flow: language and tone defaults behave more predictably, and the translation lifecycle is less likely to get stuck in awkward intermediate states.
- Polish across chat and translate: small UI fixes improve menu behavior, spacing, and overall consistency.
- Compatibility and reliability updates: database schema handling, test expectations, frontend config behavior, and CI tooling were all tightened up.
Why it matters
- Upgrading no longer leaves behind legacy model files you can’t clean up.
- Translation feels more predictable and less fragile during day-to-day use.
- The app is a little smoother and more dependable without changing your workflow.
Compatibility
- Existing databases continue to upgrade automatically.
- Legacy Qwen aliases are still supported for compatibility, but you can now manage the active file and older downloads independently.
1 Like
Informity AI v0.15.0 is out
Informity AI v0.15.0 is a major quality-and-reliability release. The biggest visible changes are in extraction, indexing, retrieval, and startup behavior, with security hardening and a handful of polish fixes layered in as well. Re-indexing your files is not required, but it is highly recommended to take full advantage of the improved extraction and indexing pipeline.
What’s new
- Better extraction and indexing: document ingestion was tightened up with scan-path fixes, stronger PDF handling, and diagnostics that better reflect real corpus health.
- Better retrieval behavior: the 5Q classifier contract was consolidated, retrieval regressions were fixed, and metadata questions like “newest,” “oldest,” “largest,” and “smallest” are handled more naturally.
- Cleaner internals: hardcoded document-specific hooks were removed from diagnostics and runtime code so the app stays agnostic to your own files.
- Startup is more reliable: the app now waits for required model setup at launch and downloads the classifier automatically when needed.
- Clearer startup feedback: the boot overlay and startup error messages were simplified so setup and failure states are easier to understand.
- Bug fixes: translation file switching now resets correctly, chat scroll behavior is back to normal, and the startup spinner animates during model download.
- UI polish: chat source markers and tooltips were refined, source cards now look consistent across themes, and file names in Files open the underlying file directly.
- Security hardening: the frontend now uses a stricter Content Security Policy, and chat trace defaults were tightened for safer handling of sensitive traces.
- More stable foundations: compatibility shims and diagnostic baselines were cleaned up, and the full test suite stayed green through the release work.
Why it matters
- New and existing users get better answer quality and a more predictable first-launch/upgrade experience.
- When something is wrong at startup, the app now tells you what happened in plain language.
- Common metadata questions and retrieval patterns work more naturally without changing how you use the app.
- The app is also more explicit about model setup, which helps keep startup behavior consistent across installs.
Compatibility
- No manual migration step is required.
- Existing files, indexes, and settings remain local to your machine.
- If the classifier model is missing, Informity will fetch it during startup and then continue normally.
2 Likes
Informity AI v0.15.1 is out
Informity AI v0.15.1 is a focused follow-up release that makes semantic file search easier to use, improves file-discovery answers in chat, and removes a startup failure mode when the classifier model is not already cached.
What’s new
- Better semantic search in Files: search mode, search text, and semantic result limits now persist more reliably, and semantic queries can be opened directly from a link.
- Cleaner file-discovery answers in chat: file-discovery queries now return a clean deduplicated list, with a link to browse all matching files when chat shows only a partial set. That link now stays with the message when you reload the chat.
- Smarter routing for file-discovery questions: queries that are really asking for relevant documents now route to content-based retrieval instead of generic inventory responses.
- More reliable startup: if the required classifier model is missing, the app can fetch it during startup instead of failing immediately.
- Polish and consistency: semantic search presentation and source-related UI details were tightened up across the app.
Why it matters
- Semantic search now feels more like a first-class workflow instead of a special case.
- Chat answers are easier to scan when you’re trying to find files, not read a narrative summary.
- First launch and clean environments are less brittle because the app can recover by downloading the classifier automatically.
Compatibility
- No manual migration is required.
- Your existing files, indexes, and settings stay local to your machine.
- This release is a follow-up polish and reliability update rather than a schema change.
3 Likes