StoryHouse's $50,000 Fund I SAFE into Precision Finance (fka Cabernet.ai) is impaired: the position was written down 50% in Q1 2026 on underperformance and roughly two months of runway, and founder discord had already peaked in September 2025 when CEO Alex Berman was pushed out. Precision Finance is an AI co-pilot for investment bankers and PE firms: it turns unstructured counterparty documentation into detailed, source-linked financial models, cap tables, and deal memos, then flags conflicting data and benchmarks each deal against prior ones. The original bet was a small, high-optionality check into a category with a genuine secular tailwind, alongside a pedigreed team and a brand-name seed syndicate.
At entry the signals were credible: a Scenic Advisors LOI (~$3,000–$5,000 per seat) and a Guggenheim verbal, a $3M round at a $15M cap led by Ben Casnocha at Village Global with $1M from 10VC (the Coors family office), and a technical co-founder described in the 2024-03-27 note as a former CTO of Marcus and head of engineering at Stripe. The company never converted that early interest into durable revenue: it stayed pre-revenue, the market proved crowded against far-better-capitalized incumbents (Rogo, Hebbia, Bloomberg), and by September 2025 the founding team fractured, with the CEO ousted after an ultimatum from the CTO. SH context: $50,000 on a $15M cap SAFE, StoryHouse Fund I, single check, now carried at a 50% discretionary markdown.
The tailwind is real: generative AI in financial services was roughly $1.7B in 2023 and is projected to reach ~$16.0B by 2030 at a 39.1% CAGR, with investment banking among the leading application segments Web. The problem was never the market: it is that the analyst-copilot niche filled fast with far-better-funded players (Rogo raised $300M+ and reached a $2B valuation), leaving a $50K seed check with no revenue in a category being consolidated by incumbents.
| Player | Positioning | Funding / Stage | Edge vs. them |
|---|---|---|---|
| Precision Finance | AI co-pilot generating source-linked financial models, cap tables, and deal memos from unstructured counterparty docs for banks & PE | Seed, $3M / $15M cap; pre-revenue, marked down 50% | — |
| Rogo | AI financial analyst purpose-built for banking / sell-side workflows | Series D, $160M at $2B valuation; 35k+ users across 250+ institutions | Rogo is orders of magnitude ahead on capital, distribution, and enterprise footprint |
| Hebbia | Document analysis at scale for data-intensive diligence, source-grounded answers | Well-capitalized growth stage | Overlaps Precision's core doc-to-model wedge with far more resources |
| AlphaSense | Market intelligence and premium external content | Late-stage, multi-billion valuation | Adjacent, but owns the research-content surface analysts already use |
| Bloomberg / FactSet / S&P | Entrenched terminals and structured-data platforms adding AI | Public / scaled incumbents | Deep data moats and universal bank adoption are a steep barrier to a seed entrant |
Moat: The intended edge was verified, source-linked models that flag inconsistencies straight from raw counterparty documents; against better-funded incumbents and banks' more-robust-than-expected internal tools (per the Evan Schnidman reference), that wedge never proved defensible.
Sector M&A is active (fintech deal value reached roughly $499B in 2025, up ~40% year over year) and AI-finance platforms remain acquisitive, with strategics like Bloomberg, FactSet, and S&P and scaled players like Rogo the plausible buyers Web. Given ~2 months of runway, a pre-revenue product, and a fractured founding team, the realistic base case is an acqui-hire or impairment rather than a return-multiple exit on the $50K entry.
Alex Berman told Josh that, a week earlier, Jeff and Andrew had informed him they were letting him go as CEO — a decision that blindsided him.
Jeff and Andrew surprised Alex a week ago and told him they'd be letting him go. Alex had been pushing the CTO hard to make technical progress, which in Alex's view wasn't happening.
The CTO told Jeff and Andrew that it was either Alex or the CTO; Alex had received no word of this. Alex has managed huge teams and had never fired someone where it was a surprise to them, so he was very surprised. On the bright side, he plans to reflect and think about what's next.
Alex feels the company is doing ok: they've found sales traction with real pilots and made technical progress. He will likely try to offload his shares to the team or existing investors. He holds about 10% of the company in common.
Andrew Aw emailed Miles to resolve a background-check discrepancy, reporting that both online and in-person court searches returned no matching cases.
Update from Andrew Aw re: issues with a background check. He looked into the issue by asking a GC colleague to dig into the information shared and by checking the OC Superior Court online search portal for case numbers and name.
He also physically went to the OC Superior Courthouse to speak with a clerk and perform the search at the courthouse. All methods yielded negative results (No Matching Cases Found), aligning with historical background checks and the prior day's discussion; he attached photos from the courthouse kiosk for additional color.
He asked whether the provider could fix the error in his profile by removing it, so there are no future background-check issues.
Alex gave a business and round update: early LOIs, a firmed-up $3M round at a $15M cap led by Village Global with 10VC, and roughly $500K left open for StoryHouse.
Business updates: expecting an LOI from Scenic Advisors the next day, paying $3K–5K per seat; Guggenheim a verbal yes; in conversations with Warburg Pincus. Scenic proposed a stock swap between their bank and Cabernet (e.g., 10% of the bank for 10% of Cabernet) to be the AI-enabled boutique investment bank; per Alex this could generate several million in cashflow for them if closed (Scenic did $20M in revenue last year). This deal is still in discussion.
Round update: they decided to take our advice and are now doing a $3M round at a $15M cap, taking a $1.5M check from Ben Casnocha at Village Global and $1M from 10VC (Coors family office), with ~$500K left for us to write a $50K–300K check. MVP timeline: get an MVP this summer to Scenic and Guggenheim. The team will rent an Airbnb in Jackson Square and work out of Scenic together a few times a week. On technical readiness: Jeff and Danielle will be coding, with one more technical hire (a top infrastructure lead at Happy Money who verbally agreed), and possibly a front-end hire to make the product slick.
Summary: was more impressed with Alex's soberness about building the GTM side; liked that they'll work in person in sprints and that they have an initial paying pilot ($3–5K/seat). Village is a stronger lead than the prior offer and a $15M post is more reasonable than the earlier $20–30M. Next steps: introduced them to LP Fred Cummings, who runs a large bank-focused hedge fund and spent a full day with the CEO; speak with Fred, ask to speak with Ben at Village, and discuss at IC.
Evan Schnidman (reference on the concept / founder Alex Berman, 2024-05-30): bearish. He called it a crowded space and noted that banks' internal tools are more robust than you'd expect. He was skeptical you can fully automate the workflow of a senior banker directing an analyst: the end user is still the analyst, who already has other tools the product must out-compete. He pointed to MD psychology (career bankers aren't motivated to make juniors' lives easier) and argued software mainly lets banks run with fewer juniors, while conceding the team isn't wrong that a massive inefficiency exists.
Source: Meeting Notes reckOzNCgXSxDVEO6 · Referenced founder: Alex Berman