The metric catches your eye immediately: four hours. That's how long institutional investors at four separate hedge funds are spending daily inside an AI platform that launched in May 2026.
It's the kind of number that cuts through the noise in enterprise software—enough sustained usage to suggest WithAI, a Y Combinator-backed startup out of San Francisco, might have figured out something others haven't. Or at least, that's what the early data from their first operating quarter appears to show.
The company builds what it calls Multiplier, a custom command center where portfolio managers and analysts collaborate with AI agents on stock research, portfolio monitoring, and the grinding daily workflows that define modern investing. This isn't a chatbot widget slapped onto existing terminals. According to WithAI, it's a unified workspace where frontier AI models—running entirely on each client's own servers—get woven into how funds actually work.
Whether that distinction matters in practice is precisely what makes the usage numbers worth watching.
What's Actually Under the Hood
Multiplier functions as connective tissue between AI agents and the sprawling tech stack hedge funds already use: Bloomberg terminals, FactSet feeds, AlphaSense research databases, Tegus expert transcripts, and the dozens of other sources institutional investors toggle between constantly. The platform sits atop that infrastructure, letting users run research queries, synthesize position data, monitor portfolios, and coordinate across tools without the usual system-switching gymnastics.
WithAI describes it as "agent harnesses for asset managers"—AI capable of researching "every stock on Earth, every day" while respecting the firm's folder hierarchies, naming conventions, and investment processes. The system is designed to absorb each fund's institutional knowledge: how they structure research, which dashboards matter, what metrics drive their particular strategies.
Everything runs on client infrastructure. WithAI emphasizes that neither the company nor Anthropic—whose Claude models power the agents—can access customer data. It's a security posture calibrated for an industry where proprietary research and trade signals represent core intellectual property.
The setup reflects a fundamental tension in enterprise AI: you can't plug powerful models into sensitive workflows without solving for data isolation and compliance first.
The Traction Question
WithAI appears to have launched publicly in recent months with four hedge fund clients already live. The company's Y Combinator materials claim users spend over four hours per day in the platform—a figure that, if it holds, would position Multiplier alongside essential workflow tools rather than experimental pilot projects.
The startup also disclosed crossing $100,000 in annual recurring revenue during Y Combinator's Spring 2026 batch—early revenue that doesn't necessarily indicate sustained or lasting ARR, though it's fast for a company that just launched. The market focus helps explain the traction: independent equity funds managing between $250 million and $5 billion in assets. These are shops with software budgets that can accommodate custom infrastructure, and where workflow improvements translate directly into competitive advantage.
Two clients have surfaced publicly through company materials. Scott Hobart, CIO at Mercator Partners, is quoted on WithAI's website saying his team's time allocation flipped from 80 percent gathering information and 20 percent acting on it to the reverse ratio. Michael Siliciano, co-founder of Verso Partners, is similarly quoted describing abandoning fragmented tools for the unified platform. These testimonials appear on the company's site but haven't been independently verified by third-party sources.
Client testimonials are marketing, of course. But sustained daily usage across multiple funds is harder to fake.
The Build Process
WithAI's approach involves embedding engineers with each fund to integrate tools, file systems, data feeds, and institutional context. The explicit goal, according to company materials, is avoiding a world where portfolio managers become "full-time Claude-wranglers" or prompt engineers.
The implementation includes secure inference endpoints, ontology maintenance, guardrails, and VM infrastructure, all hosted on client servers. The platform claims integrations with a sprawling roster of financial data providers, order management systems, cloud platforms, and collaboration tools. WithAI's website displays logos for Bloomberg, FactSet, MSCI, Aladdin, Interactive Brokers, and others—though these appear to represent technical connectors rather than formal commercial partnerships.
Once deployed, the system supposedly learns each fund's "DNA": how research flows, what questions analysts ask repeatedly, which metrics drive decisions. It's meant to codify the institutional knowledge that typically resides in Slack channels, shared drives, and the heads of senior analysts.
Whether AI can actually capture that tacit knowledge remains an open question. But the bet is worth watching.
Who's Behind It
The founding team brings credible institutional experience. Ian McInnis, CEO, studied mathematics at Princeton (reportedly graduating with a perfect GPA) and worked as an investor at Bridgewater Associates. Ben Finch, CTO and president, holds graduate and undergraduate degrees in electrical and computer engineering from Princeton and previously served as founding researcher and chief of staff at Sentient Labs.
Y Combinator backed the company. WithAI's website lists angel investors including Greg Jensen and Karen Karniol-Tambour, who serve as co-CIOs of Bridgewater, though these claims have not been independently verified. The company hasn't disclosed funding amounts or valuation. The team lists five members, including COO Ryan Winkler and engineers Edison Zhu and Skyler Chan.
A Suddenly Crowded Market
WithAI is entering a space that's gotten loud, fast. In recent months, multiple established players and startups have announced similar offerings. Arcesium launched an AI platform for institutional investment firms. Broadridge announced agentic AI in production, claiming operational cost reductions up to 30 percent. Symphony introduced its AI Agent Studio. Waton Financial opened a limited beta of its multi-agent investment platform.
The trend extends beyond financial services firms. Infrastructure providers like Datadog, Affinity, and Datasite launched MCP (Model Context Protocol) servers to connect AI agents to enterprise data with governance controls. The protocol allows agents to access structured systems while respecting permissions and compliance requirements—exactly the infrastructure layer that makes institutional deployment realistic.
Even fellow Y Combinator companies are staking territory. KelAI, from the same batch, positions itself as an autonomous research loop for funds, capturing the investment process from initial idea to portfolio manager feedback.
The competitive landscape points to a broader industry shift. A recent Broadridge study found that 26 percent of financial firms are already deploying agentic AI, with 51 percent of those in active production. Academic research has proposed agentic architectures where researcher and meta-agents iteratively improve forecasting and portfolio construction—theoretical framing for a trend already playing out in product roadmaps.
What Happens Next
WithAI's early metrics—four clients, sustained daily usage, six-figure ARR—suggest the company has found product-market fit in a specific vertical. But it's operating in a window where multiple players are racing to define institutional AI infrastructure. The on-premises deployment model, white-glove implementation, and fund-specific customization position it as bespoke infrastructure rather than standardized SaaS.
The open question is scalability. Can this approach expand beyond early adopters, or does the market eventually consolidate around more plug-and-play platforms once agentic workflows become standard? For now, the hours-per-day usage metric indicates WithAI has built something portfolio managers actually integrate into daily work. Whether that translates to the next institutional tier—larger funds, multi-strategy platforms, family offices—depends on how quickly the company can replicate its deployment model and how the competitive field evolves.
The focus on independent equity funds managing $250 million to $5 billion carves out defensible territory: sophisticated enough to justify custom infrastructure, nimble enough to move quickly on new technology.
It's the kind of market timing that can make or break early-stage companies. Right product, right buyer, right moment—maybe. The usage data suggests WithAI is onto something. Whether it's durable advantage or just early-mover momentum is the question every enterprise software startup eventually has to answer.
