Deterministic import, rich text overhaul, Max mode, bug fixes, and platform hardening

You can now read release notes directly in the app — click the icon in the top-right corner.
This is one of our largest updates yet. We added deterministic file import, a major overhaul of rich text, Max mode for image sourcing, our own CDN, significantly better statistics, and hundreds of fixes behind the scenes. Below is everything from the biggest changes down to the details.
Headline features
Deterministic file import
Alongside extraction and empty tables, there is now a third way to get supplier data into Lasso — direct import.
Many of you told us that extraction sometimes failed to build a table correctly, even for formats you use over and over. Import solves that: a known supplier format lands in your enrichment table clearly, deterministically, and the same way every time.
When to use what:
- Extraction — new supplier tables with a specific format you do not want to handle manually.
- Import — a repeating, well-known format that should always map the same way.
- Empty table — when you want to paste data in yourself.
Import is the second option in the product listing flow.

Rich text — major overhaul
The rich-text editor gained new capabilities and better performance:
- Support for non-breaking spaces directly in formatted text.
- More HTML export options — inline styles, classes, and custom formats on request.
- HTML templates can include variable sections based on content length.
- General improvements to features, speed, and day-to-day editing.
- When placing images, Lasso now looks at their content and writes stronger surrounding copy.
- You can duplicate, delete, create, and preview custom templates before generating content. We can also add your own export format on request.
- Image placeholders — when Lasso does not have an image yet, it suggests where one should go and lets you fill it in later.
- Internal link building — add your sitemap and Lasso can insert internal links to other pages on your site while generating text.

Images — Max mode and generation
Max mode (experimental) Ultra-precise image sourcing where the Lasso agent uses its own browser. It can click, type, pass CAPTCHAs, and open complex galleries. Much better for instructions like “on site X, find every image for product Y.” It is still expensive to run (100 credits), and we will bring the cost down as the technology matures.
Manual image generation
- Available in galleries, rich text, and when editing existing images.
- Multiple cheaper models to choose from.
- Worth trying — for example, turning reference images from the web into polished lifestyle shots for your products.
Other image improvements
- Better standard image sourcing (the 2-credit flow).
- Validation and automatic adjustment — minimum resolution, file format, compression, and more.
- Reorder images in galleries.
- More metadata shown, including resolution and file format.
- On request, we can connect image sourcing to your own database or FTP.

Dedicated CDN
Downloaded images are now stored on our CDN. You can keep images hosted through Lasso on your site without worrying that older assets will disappear.
Enrichment
- Type
/allin an enrichment field to pull the full row into the agent’s context. - Several AI models updated in the background.
- Better language detection for translations and unit conversions.
- A fully redesigned AI reasoning panel (question-mark icon) — see more detail about how the AI thought through a response. For selected features, including video and screenshots of its reasoning steps.

Statistics — major improvements
- More useful metrics, including conversion to Czech koruna.
- A new API endpoint — pull statistics into Power BI or your own automations.
- Better and more accurate date selection.
Smaller but welcome updates
- Never export — column settings now let you exclude a column from every export. Useful for helper columns used for validation or lookup that should stay in the table but never ship in the final file.
- Tags and enum column values can be edited after creation.
- xlsm file support (Excel with macros) for extraction and import.
- New API endpoints for attribute management.
- Updated API documentation — see the API quickstart.
Simplification and design
We keep simplifying the app — refining copy, contrast, and small details across attributes, sign-in, settings, table work, and more.
Bug fixes and performance
In our last changelog we promised better reliability. Here is what shipped:
- 300+ automated tests and 24/7 AI agents now hunt for new issues before we or you notice them. Reliability remains our top priority.
- If you received a mistaken low-credits email — sorry about that; it is fixed.
Enrichment and performance
- Fixed slow loading and general enrichment behavior.
- Faster deletion of cells and large row batches.
- Fixes for undo/redo and saving column values when several people work in the same table.
- Fixed waiting on enrichment for a field that had already finished in the background (previously required a refresh).
- Fixed enrichment results that sometimes did not write back to the cell.
- Clearer error messages when cell enrichment fails.
Extraction
- Custom extraction reads tags from a single field more reliably.
- Extraction indicator loads faster.
Table work
- Invalid enum values can now be corrected in place and added to the allowed list.
- Fixed search in long enum dropdowns.
- Fixes for pasting thousands of rows from Excel into Lasso.
- Fixed paste overwriting cells in unrelated columns.
- Fixed Cmd+Z on Mac.
- Fixed a white screen that could appear after an hour or more of work.
Schemas and attributes
- Fixed attribute key display on new tables.
- Fixed attaching existing attributes to a schema.
- Fixed file and image export settings on empty schemas.
- Fixed reordering columns while editing a schema.
Export and views
- Clearer saved-view management.
- Fixes for CSV export and segmentation.
- Batch mode fixes.
Other
- Fixed Czech/English switching for texts that stayed in the wrong language.
- Fixed duplicated tutorial videos.
- Gallery fixes.
- Fixed credit usage indicator.
- Security improvements.
- …and many other issues we caught along the way.
What we are working on next
- More roles for different people and more teams.
- Continued bug fixes and performance work.
- Relationships for products and metaobjects (long promised — getting close).
- Views — we want to invest much more here, especially when several people on one team work in the same table on different columns. Easier sharing, reuse, and organization.
- Integrations to your favorite services.
- Enrichment sequences (some steps earlier, some later).
- Catalog improvements and more.
Have a great summer ☀️