Benefit Street Partners speeds up credit-agreement analysis with Codex
According to an OpenAI customer story, Benefit Street Partners uses Codex-assisted tools to produce a first pass on credit agreements in about 20 minutes and answer committee questions faster.
AT A GLANCE
Key points
- The first BDC tracker version was built in a day
- Total development time is estimated at 10–12 hours
- A first-pass credit-agreement analysis takes about 20 minutes
- More than an hour of human review follows each first pass
What is changing?
OpenAI described Benefit Street Partners’ use of Codex on October 16. Portfolio manager Anant Kumar built a BDC tracker that compares how different firms value the same loan using public SEC filings. The first version was completed in a day; he estimates 10–12 total development hours after later refinements.
Kumar’s other tool scans credit agreements, which can exceed 300 pages, for risky clauses. According to OpenAI, a first-pass analysis takes about 20 minutes and is followed by more than an hour of human review. The system was refactored with GPT‑6 Astra and now runs in a company-hosted portal.
BSP says it can respond to investment-committee comments in a day rather than days or weeks, and sometimes during the meeting. The firm says it uses the tools to add review capacity and rigor within the same timeframe, not to do more deals. The figures are user claims in an OpenAI customer story, not independent measurements.
Explore the original sources
Source date: 16 October 2026.
Published by AI Radar: 17 Oct 2026, 13:17 (Istanbul, UTC+3).
FOLLOW THE TOPIC
Topics and tags
Share your discovery.
Send this story to a friend.
Reader comments
What do you think about this development? Share your experience or question. Comments are published after editorial approval.
Your name/nickname and comment become public if approved. Read: Privacy notice.
Loading comments…