At Benefit Street Partners (BSP), portfolio manager Anant Kumar has developed internal AI tools to compare loan valuations and analyze credit agreements. The objective is to add review capacity and rigor within existing timeframes, not necessarily to increase deal volume.
BDC mark tracker: faster valuation comparisons
Kumar built a tracker that pulls public SEC filings from business development companies (BDCs) and compares how different holders value the same loan. The team can view how BSP’s marks stack up against other holders’ valuations.
He created the first version of the tracker in one day and used Codex for code review. After adding features and refining the interface over subsequent weeks, he estimates the total development time at about 10–12 hours.
Credit‑agreement analyzer: quick first‑pass reviews
The second tool analyzes credit agreements, which often exceed 300 pages. Investors need to identify clauses that could shift value away from lenders, a task that previously required outside counsel to review agreements page by page.
Kumar consolidated memos from leading law firms about clauses and risks to watch for and incorporated those into the analyzer, which he describes as a “second brain within our organization.” An uploaded credit agreement takes roughly 20 minutes to produce a first‑pass AI analysis, followed by an hour or more of human review.
BSP has refactored the analyzer with GPT‑6 Astra, and it now runs live on the company‑hosted portal.
Faster committee follow‑ups while preserving human judgment
According to Kumar, answering an investment committee’s follow‑up questions used to take days or weeks. With Codex, analysts can now turn a round of comments in a day and often respond during the meeting itself. Kumar notes: “The AI is very helpful in collating data, ingesting it, synthesizing analysis, but the ultimate investment decision needs to be made by a human.”
Underwriting typically proceeds in multiple rounds, each involving further diligence and a revised memo that addresses the committee’s questions. Deal timelines range from a few days to a few weeks, so the number of rounds varies. Kumar says, “We feel more confident in our decision making because we have more information at our disposal.”
Longer‑term aims: continuous underwriting and connected deal data
BSP is also working to connect information across deals. In the longer term, the team aims for continuous underwriting: monitoring loans as news arrives and assessing effects across the portfolio, rather than re‑underwriting only after events occur.
Kumar sums up the practical benefit: “Codex has unlocked my day, because I have many more hours in the day to do more high‑level work and less routine work.”
Key figures
- First BDC tracker: built in a day; ~10–12 hours total including later additions.
- Credit agreement first‑pass: ~20 minutes AI processing + typically 1+ hour human review.
- Faster responses to committee follow‑ups, often within meetings, per Kumar.
These tools extend BSP’s internal underwriting capacity and rigor without necessarily increasing deal throughput.
