Most people in their early twenties don't walk away from Goldman Sachs to start a hedge fund. Jack Zumwalt and Mauricio Ortiz did exactly that—at ages 21 and 22—raised $10 million, and spent the next four years running Level III Capital—a systematic trading operation focused on volatility management. Then they did something stranger: they productized the fund's internal stack into a software product.
Kimpton AI, the result of that pivot, wants to deliver what sounds impossibly ambitious—daily trade proposals that land on a portfolio manager's desk before market open, complete with citations and tailored to their mandate. No hallucinated numbers. No generic stock screeners masquerading as intelligence. Just autonomous research that reads, the founders insist, like it came from a human analyst who actually did the work.
Whether the buy side believes them is another question entirely.
A Quant Fund That Became a SaaS Company
The origin story has the tidy logic of hindsight. Zumwalt was doing detection engineering at Goldman; Ortiz was a DevOps Engineer on a critical infrastructure team at the bank. Adrian Del Bosque, who'd become Level III's founding engineer, built systematic trading systems for digital assets. The three left Goldman together in 2021, still young enough to take the risk.
Level III operated as a quant fund, building automated infrastructure and volatility strategies in-house. At some point—perhaps more gradually than the founders now let on—they realized the stack they'd built to run money might be worth more as a product than as a proprietary edge. So they productized it.
The pitch is straightforward, almost deceptively so: AI does the research, humans make the decisions. Kimpton launched through Y Combinator's accelerator program with a simple premise. Upload your mandate and strategy documents, connect your brokerage account through Plaid, and the system starts generating tailored proposals every morning. Trade ideas, fully cited, ready to act on—or ignore.
What's Actually Under the Hood
Kimpton calls itself an "AI research platform for the buy side," targeting portfolio managers at hedge funds, mutual funds, and advisory shops. The core offering is those Trade Proposals—structured investment ideas with rationale and source citations. But the product extends beyond morning briefings.
It includes what the company calls Deep Research reports, pulling from SEC filings and earnings transcripts. There are natural-language dashboards tracking portfolio positioning, agentic charting tools that backtest strategies, and scheduled "Skills"—automated workflows for recurring analysis. The platform ingests data from more than 21,000 U.S. and international tickers, with over a decade of fundamentals, real-time SEC filings, Form 4 insider transactions, 13F holdings, earnings call transcripts, M&A activity, and live options Greeks.
Data flows in from FactSet, Tiingo, Massive, and Plaid. Even Polymarket's prediction markets are wired in, according to the company's Y Combinator profile. Everything, Kimpton emphasizes, is read-only. Every number gets cited to source—filings, transcripts, live feeds. The company's marketing materials hammer this point repeatedly, an implicit jab at AI tools that fabricate figures with confidence.
The platform isn't trying to replace Bloomberg Terminal so much as sit alongside it, handling the grunt work of idea generation that typically falls to junior analysts or outsourced research shops.
A Breakneck Shipping Cadence
Since launch, Kimpton has moved with the velocity of a team that's anxious to prove something. In recent months, the feature list has expanded aggressively: daily market insights, prediction market integration, live crypto coverage, real-time news signals. Brokerage connectivity through Plaid now supports thousands of brokerages. The company rebuilt its UI, accelerated its Deep Research function by a factor of ten (their claim), and added social signals from X, formerly Twitter.
More recently came real-time reasoning visualizations—so users can watch the AI think, essentially—along with futures and commodities data, new chart types, and what the company bills as "Cursor for capital markets," a dedicated workspace with live candlesticks, watchlists, indicators, and an AI research sidebar. PDF text extraction, file attachments, @-references for tickers and portfolios.
It's the kind of release tempo that either signals product-market fit or the frantic energy of a team still hunting for it. With a five-person team, per Y Combinator's directory, it's also the kind of pace that's hard to sustain.
Security Theater, or Actual Security?
Kimpton says it's pursuing SOC 2 Type II certification—a credential that matters to institutional clients who can't afford data breaches or regulatory blowback. The platform operates on a read-only basis, with no training on customer data, revocable connections, tenant isolation, AES-256 encryption at rest, TLS 1.2 or better in transit. A Trust Center link exists, though the details hide behind authentication walls.
The company's marketing leans into a build-versus-buy argument: days to production instead of 12 to 18 months building in-house, zero need for internal quants and data engineers, vendors consolidated into a single contract. These are Kimpton's claims, not independently verified benchmarks. Still, the argument resonates in an industry where back-office costs are scrutinized relentlessly.
A Crowded, Fast-Moving Market
Kimpton is hardly alone in chasing this opportunity. FactSet, which dominates institutional data delivery, has rolled out AI features to tens of thousands of users. Lightkeeper launched Lightkeeper Beacon earlier this year to deliver verifiable AI answers to investment questions. Boosted.ai enhanced its Alfa agentic AI platform and partnered with BX Partners to bring AI co-pilots to advisors and asset managers.
Then there's Aiera, providing event intelligence for institutions; Hamachi, which partnered with Modelist to embed AI-driven portfolio insights for RIAs; and Public.com, which announced plans last year for an AI-powered brokerage with an AI portfolio manager. Academic research on agentic AI for portfolio management has picked up steam as well, signaling broader industry interest in autonomous investment workflows.
The real question—perhaps the only question that matters—is whether a startup can compete with the distribution power of a FactSet or the institutional relationships of a Lightkeeper. Kimpton's founders are betting that former buy-side operators can build for buy-side needs better than vendors who've never run a book. It's a theory. Whether it holds in practice depends on execution, adoption, and whether portfolio managers actually trust the machine's ideas enough to put capital behind them.
What Comes Next
Kimpton offers a free tier to get users in the door and an enterprise track with SAML SSO, fine-grained access controls, and audit trails on the roadmap. Pricing details, predictably, sit behind authentication walls. The company lists its headquarters as San Francisco on Y Combinator's directory, though LinkedIn shows a Dallas address—likely a relic from Level III's Texas roots.
No named enterprise customers are publicly disclosed. No priced funding round has been announced beyond Y Combinator's participation, which typically involves a standard check and single-digit equity. Zumwalt posted recently that "Kimpton AI is officially live for all investors," showcasing Monte Carlo simulations and backtests. The platform exists. It's out there.
Whether portfolio managers will trade on its proposals—whether they'll trust a machine built by three former Goldman employees who spent four years running a fund most people have never heard of—remains to be seen. The technology is impressive. The pedigree is solid. But in a business where reputation and relationships still matter more than algorithms, building a better mousetrap doesn't guarantee anyone will use it.
For now, Kimpton is live, shipping fast, and making a bet that the future of investment research looks a lot more autonomous than the present. Whether the buy side is ready for that future is the trade proposal that matters most.
