​
Our story

tile.run was born out of our own needs. We are the team behind Kili, a document processing platform for business teams. While building Kili, we found that:
  • Structured data (i.e. JSON) was required for most workflows
  • Handling dense files (lots of tables or many pages) was challenging. These are the files where structured data extraction adds the most value.
  • Getting to accuracy that was meaningful enough for automation took a lot of iteration.
We spent a lot of time optimising our extraction engine. After talking to other developers, we realised other teams are facing similar challenges. tile.run is an attempt at helping them solve these challenges.

​
What is tile.run best for?

tile.run priorities accuracy and follows an async pattern. It performs particularly well on dense documents with lots of information like tables, forms and so on. tile.run may not be the best tool for the job if you want to extract data on the fly with low latency and low cost.