All areas

Language & Documents

Reading and reasoning over text: questions answered from your own files, plain English turned into SQL.

Filter

Showing 4 of 4

Discrepancy Report Across Documents

Upload a contract, then the invoices billed against it, and see where they disagree. Mismatched amounts, dates and terms are flagged with both source passages side by side and an explanation of the conflict. Invoices are only ever checked against the contract, never against each other — they are supposed to differ.

RAGReconciliation
Open

AI-Powered Document Data Extraction

Upload an invoice, contract, resume, medical report or bank statement and get its fields back as structured data. The document type is identified automatically, each field is extracted with a confidence score, and a box is drawn on the page showing exactly where the value was found.

OCRLLM
Open

Tables & Figures as Citable Knowledge

Ask questions about a PDF and get answers cited back to the page they came from — including answers that live in a table or a chart rather than a paragraph. Tables are read as structured data and figures get an AI-written caption, so a number buried in a bar chart is still findable.

RAGMultimodal
Open

Natural Language → Database Queries

Ask a question in plain English and get SQL you can actually run. The agent writes the query, executes it against a real database, explains what came back, and retries itself if the query errors. Bring your own SQLite file or a PostgreSQL connection, or try it on the Chinook demo database.

SQLLLM
Open