The Flatfile team
New Feature Release Workspaces Embeds (Portal V3)

Stand on the shoulders of giants

AUTHOR: The Flatfile team

Yes, we think anyone dealing with importing data should be considered for sainthood. We'll save you the pain with...

ML powered Matching in Portal V3 + Workspaces


Field matching in our Portal V3 product is now powered by a machine learning model trained on Flatfile’s knowledge of over 100 million manual matching choices. Now, if matches are not remembered from a previous interaction by that user or organization, global matching kicks in to do the heavy lifting. Our testing shows that over 90% of imports have all headers suggested correctly.

Portal V3 also has a new engine! This change:

  • improves loading time for all steps of import as well as hooks and validations

  • allows us to support files of any size (even over 1GB!)

  • laid the foundation for our improved matching and other upcoming data automation projects - stay tuned!


Flatfile is now hosted in the EU! Contact us if you’re interested in learning more.

Other changes

Embeds (Portal V3) and Workspaces

  • Import data table now includes metadata

  • Validators now support multiple values in arrays

  • All unique enum values are now shown for matching in Portal, no longer first 1000

  • We added REST and GraphQL endpoints for adding, updating and deleting rows to Workspaces

  • Data template JSON editor now shows error if JSON is invalid

  • Show data previews for excluded fields at Matching

  • Failed imports will now be retried automatically up to 20 times before the import is canceled

  • Users will now be notified when trying to add an empty template to an Embed

  • Improvements to automatic detection of a file’s delimiter

  • Uniqueness validator no longer highlights the original instance of the duplicate

  • Improved stability of our data table

Powered by LaunchNotes