Benefit Street Partners accelerates investment analysis with OpenAI Codex and GPT-6 Astra
Investment workflow transformation
Benefit Street Partners (BSP), a credit-focused investment manager, has significantly reduced the time required to address investment committee inquiries by leveraging OpenAI's Codex and GPT-6 Astra. Portfolio manager Anant Kumar reports that analysts can now complete rounds of comments within a single day, and frequently provide answers during live meetings, rather than waiting days or weeks. The firm utilizes these tools to increase the rigor and capacity of its underwriting process without increasing the total number of deals handled.
Custom dashboard development
Codex has enabled non-technical staff to develop internal software tools, including a tracker for Business Development Company (BDC) data. Previously, building such tools required dedicated software engineering resources to manage website development and data integration.
Using Codex to draft and review code, Kumar developed a BDC tracker that compares firm valuations for specific loans using public SEC filings. The initial version was completed in one day, with total development time—including subsequent refinements and feature additions—estimated at 10 to 12 hours.
Credit agreement analysis
BSP has deployed a tool to analyze credit agreements, which often exceed 300 pages. The tool functions as an internal "second brain," synthesizing memos from law firms regarding specific clauses and risks.
- Processing time: An uploaded agreement is processed into a first-pass analysis in approximately 20 minutes.
- Human oversight: Each machine-generated analysis is followed by at least an hour of human review.
- Technical stack: The tool has been refactored using GPT-6 Astra and is currently hosted on the company's internal portal.
Kumar emphasizes that while the AI is effective at collating and synthesizing data, the final investment decision remains a human responsibility. By automating routine data ingestion, the firm reports higher confidence in its decision-making due to the increased availability of synthesized information.
Future outlook
BSP is currently working to connect information across its various deals to enable "continuous underwriting." This long-term initiative aims to allow the team to monitor loan performance in real-time as news occurs, rather than relying on re-underwriting processes triggered only by specific events. According to Kumar, the integration of these tools has shifted his focus from routine tasks to higher-level analytical work.