Turning mixed files into structured data
Could any kind of data be used right away, just by connecting it?
Data in every format, such as spreadsheets, PDFs and logs, usually has to be cleaned up before AI can use it. We want to build that cleanup step as open source and share it with others who do the same work by hand. As with KITE, we would like to keep it open and refine it together.
Spreadsheets · PDFs · databases · logs
We aim to recognize formats and clean duplicates and errors, then have agents infer the structure and check each other's work.
Cleaned, structured data
| Field | Open source |
|---|---|
| Status | Planned |
| Input | Spreadsheets · PDFs · databases · logs |
| Output | Cleaned, structured data |
| Delivery | GitHub open source (goal) |