Projects

Finance needs the liability caps across every active service agreement by Friday. The manual approach is always the same: open each document, find the clause, copy the value into a spreadsheet, repeat. A comparison across dozens of contracts can take days, and the answers are only as reliable as the person reading each one.

A Project does that reading for you. Pick the contracts, define the data points you want (governing law, expiry date, liability cap, a yes/no flag for a clause you care about), and the AI reads every document and fills a structured grid: contracts down the rows, your data points across the columns. Each cell carries the extracted value, the AI’s reasoning, and a quote from the source clause, so you can verify what it found.

In the left nav this is Projects, and it opens as a Tabular Review. Today it does one kind of analysis: extracting data points into a grid, one cell per contract per column.

Projects reads from contracts you already have, so you need contracts in the repository first. You can pull them straight from the repository, or upload files into the project (which also files them in the repository). With an empty repository there is nothing to analyze. For how contracts get there, see Contracts.

Projects are shared across your company. Every user in your company can see and edit every project, run analysis, and delete it. There are no private projects and no per-project sharing, so coordinate on a shared one rather than expecting it to stay yours.

What each person sees inside a project still respects the usual document access: a project surfaces only the contracts a user is allowed to see, so two people opening the same project may not see the same rows. To hand results to someone outside this access, export the grid to Excel and send the file.

This section also covers writing good column questions and checking and trusting AI results.

• Writing Column Questions
• Checking and Trusting Results
• Data Exports
• Contracts Overview