• Add an upload option to the chat, which simply adds a document to the doc store.

• Add an option to submit an interaction as part of a support request: basically, please check that this is right, which sends it to a human for review.

• Ability to lookup CVR

• Status page, which shows an overview

• Detailed breakdown page

• Little circle in the corner which takes you to a fresh chat

Extract a short description of any document: image or pdf.

Extract metadata from the document, e.g. location of image. Which can help the LLM match. Translate the GPS coordinate to a human readable string, e.g. "Paris, France".