A portfolio manager builds his own credit-agreement analyzer and loan-mark tracker
At credit manager Benefit Street Partners, portfolio manager Anant Kumar built an AI tool that produces a first-pass analysis of a credit agreement in about 20 minutes, followed by an hour or more of human review. He also built a tracker comparing how other firms value the same loans.
How it works
Credit agreements often run 300 pages or longer, and finding clauses that could shift value away from lenders used to mean asking outside counsel to go through them page by page. Kumar collected memos from leading law firms on the clauses and risks to watch for and built an analyzer around them; it was later refactored with GPT-6 Astra and now runs in the company-hosted portal. Separately, he built a dashboard from public SEC filings of business development companies so the team can compare its loan marks with other holders' valuations, using Codex for code review; the first version took a day and about 10-12 hours in total. The team says follow-up questions from the investment committee that used to take days or weeks can now often be answered during the meeting, and the final investment decision stays with a human.
Source
OpenAI customer story, October 16, 2026