
A fast-growing GP with private equity and credit arms ran its deal pipeline through a CRM, Excel trackers, and email attachments. Every deal pulled from a sprawl of disconnected sources (CIMs, internal models, prior memos, CRM, portfolio monitoring, external research), but none of it was connected or queryable together.
Cursory use of Claude on individual documents helped summarize at the margins but couldn't do the work that mattered: first-pass analysis across all the inputs at once. That bottleneck hit hardest at screening, where both arms needed efficient first-pass reads to decide what to advance.
Credit was screening high opportunity volume and couldn't keep up. Equity was screening fewer but wanted to get a deeper view quicker on the deals worth pursuing. Deal teams were rebuilding the picture from scratch on every opportunity, by hand, before any judgment could happen.
Soal Labs built a deal screening and memo-generation engine for the GP's private equity and credit arms. Using Azure OpenAI and Anthropic models, it ingests every deal input, extracts and validates the deal's shape against firm business rules, stores it in a structured deal database, and drafts each memo from pre-screening through investment committee.
Soal Labs started by sitting with the deal team to map how an opportunity actually moves through the firm and where the bottlenecks sat on the credit and equity sides. That shaped a deliberately thin first slice: ingest a CIM, extract the deal's shape, run it through a basic screening checklist, and draft a pre-screening memo. The structured deal database was built in parallel so every extraction was stored from day one rather than retrofitted later, and by the end of month one the firm could run a real deal end to end. The next three months expanded the system in three directions: the rest of the memo lifecycle, from refined screening through investment committee; a validation layer of business rules to catch bad extractions; and a workflow layer for stage routing and audit trails. The order was intentional — memos first because that was where analysts spent their time, validation second because trust in extractions had to be earned, and workflow last. Throughout, the team kept the deal team in the build loop on a weekly cadence.
Private equity and credit GPs running active deal flow at enough scale that manual screening and memo work has become a real bottleneck. Most relevant for firms with:
- Sufficient deal volume to make automation worth the build (hundreds of opportunities per year).
- A real deal archive (5+ years, hundreds of opportunities) currently sitting in folders and CRMs rather than as a queryable asset.
- Leadership willing to document the firm's analytical lens before automating it.
Less relevant for solo GPs, very early-stage funds, or firms without a settled view on what their screening process actually is.



-p-500.jpg)


