Turning contracts into searchable data at OpenAI

This is part of our series sharing internal examples of how OpenAI is using its own technology and APIs. These tools are being used internally, only at OpenAI, and are shared here as illustrative examples of how frontier AI is supporting use cases across our teams. We’re also sharing the internal tool names for a clearer look at how frontier AI helps our teams get work done.

When contracts became the bottleneck

Every enterprise deal comes with a signed contract. Each one has start dates, billing terms, renewal clauses.

At first, the process was manageable: read line by line, retype into a spreadsheet, move on. But when volume doubled and doubled again, this manual approach broke.

“In less than six months, the team went from reviewing hundreds of contracts each month to more than a thousand. And yet we’d only hired one new person. It was obvious that the process wasn’t going to scale,” says Wei An Lee, AI Engineer.

Building a smarter workflow

Instead of throwing more people at the problem, our finance and engineering teams built a contract data agent. The design principle was simple: take the repetition out of contract review, keep experts firmly in control.

The Agent works in three steps:

• Ingest data: PDFs, scanned copies, even phone photos marked up with handwritten edits. What used to be dozens of inconsistent files now flow into one pipeline.

• Inference with prompting: Using retrieval-augmented prompting, the system parses contracts into structured data. It doesn’t dump a thousand pages into context; it pulls only what’s relevant, reasons against it, and shows its work.

• Review: Finance experts review the structured output, complete with annotations and references for any non-standard terms. The agent highlights what’s unusual; humans are then looped in to review.

“We’re not just parsing, we’re reasoning—showing why a term is considered non-standard, citing the reference material, and letting the reviewer confirm the ASC 606 classification.”

Siddharth Jain, AI Engineer