We're excited to unveil a new Playground that showcase's Datalab's expanded feature set:

• Parsing: SoTA parsing with tunable quality, latency, and positional information.
• Refinement: Refine parse output using natural language prompts.
• Segmentation: Segment (or "split") PDFs by page automatically or by defining a schema.
• Extraction: Extract structured data from PDFs with a schema.

Here's the playground showcasing these features on a PDF using a "digitally-stapled" collection of testimonies from an NYC Council hearing, which you can download from the council's system of record, Legistar.

Give it a spin!

A limited version of the playground is available for free. You can also sign up to receive $5 in free credits*, which gives you full subscriber access to the playground while you evaluate Datalab.

In the coming weeks, we'll be making playground sessions persistent so you can return to them easily, share them with your team, and even open and examine recent API requests in our playground (as long as your data retention settings allow it).

‍

*Credit Card details required, but we won't charge you until you subscribe to a plan