The pitch meeting didn't happen over Zoom. No demo, no PowerPoint deck. Instead, engineers from WithAI sat across conference room tables from portfolio managers and analysts, notebooks open, asking how they actually worked—not how they wished they worked, but the real, messy processes that define how billions get allocated.
It's an old-school approach to selling cutting-edge technology, and it might be working.
WithAI, a young startup from Y Combinator's Spring 2026 batch, announced its Multiplier platform on May 20, positioning it as something different from the parade of AI tools flooding into finance. Rather than another chatbot promising to revolutionize research, the company deploys engineers directly into hedge funds to build custom AI systems—what it calls "command centers"—tailored to each client's specific workflows and hosted entirely on their own infrastructure.
The company says it crossed $100,000 in annual recurring revenue after approximately three months, live with multiple clients. For enterprise software targeting an industry famous for moving at glacial speed, that's a brisk start—though still early days by any measure.
What's Under the Hood
At its core, Multiplier aims to be the connective tissue between the sprawling collection of tools hedge funds already use: Bloomberg terminals, FactSet feeds, risk platforms, portfolio management systems. The company's website lists dozens of integrations, from AlphaSense to Snowflake to various order management systems, though it doesn't specify which are live in production versus aspirational builds.
The platform is designed to let funds research stocks, manage portfolios, and maintain financial models using AI agents—software that can act semi-autonomously within defined parameters. WithAI describes the interface as "more like an IDE than a collection of chatboxes," a nod to the development environments programmers use, emphasizing control and transparency. The idea: agents that can research "every stock on Earth, every day" and keep projections current, though those remain marketing claims rather than independently verified benchmarks.
What sets Multiplier apart, at least in theory, is that it learns a fund's proprietary investment process. How do they model cash flows? What signals matter? Where do they believe they have an edge? The system is supposed to enforce each firm's folder structure and internal taxonomy so that both AI agents and human analysts can navigate research in the same way.
It's a high-touch, high-trust proposition in a sector where both are in chronically short supply.
The Forward Deployment Gambit
WithAI's strategy revolves around what it calls "forward deployment"—a term borrowed from the Palantir playbook. The team conducts face-to-face interviews with fund staff, capturing the tacit knowledge that rarely makes it into onboarding documents, then builds out custom functionality based on those conversations and ongoing feedback loops.
It's labor-intensive, expensive, and difficult to scale. But it may be the only way to thread the needle between hedge funds' appetite for AI and their deep skepticism about handing over sensitive data or trusting black-box systems.
On security, WithAI emphasizes a clean pitch: "everything lives on your cloud—neither we nor Anthropic can see your data." The platform runs on client servers using what the company describes as secure inference endpoints, guardrails, and dedicated virtual machine infrastructure. The architecture reportedly leans on models from Anthropic's Claude family, though full technical details remain under wraps, as does the list of supported cloud providers.
Mercator Partners, one of the named clients, offered a testimonial that captures the promise: "Multiplier eliminates the need to search every 'haystack.'" Whether that holds up under sustained use remains to be seen.
The Founders and the Money
WithAI was founded by Ian McInnis and Ben Finch—backgrounds that hint at where the company's DNA comes from. McInnis spent time as an investor at Bridgewater Associates, the $150 billion hedge fund known for systematizing everything, and studied mathematics at Princeton. Finch was a founding researcher and chief of staff at Sentient Labs, with prior work on prompt security and multi-agent coordination—technical chops in an area most hedge funds are still figuring out.
The team has grown beyond the founders. The company's website lists Ryan Winkler as COO (formerly at StepStone Group), along with engineers Edison Zhu and Skyler Chan, among others.
Backing comes from Y Combinator and a roster of angel investors that includes Greg Jensen and Karen Karniol-Tambour, both co-chief investment officers at Bridgewater. Funding amounts haven't been disclosed—typical for this stage—but the Bridgewater connection is notable. Jensen and Karniol-Tambour don't invest lightly, and their involvement signals confidence in both the founders and the thesis.
A Market Still Finding Itself
WithAI is entering a market where optimism and confusion exist in roughly equal measure. A KPMG survey from early 2026 found that around 80% of large U.S. asset management and private equity firms—those managing over $1 billion—reported deploying AI agents. Yet a separate study by Carne Group in January painted a different picture: fewer than 20% had integrated AI into core operations.
The gap likely comes down to definitions. Is an analyst using ChatGPT to summarize earnings calls "deployment"? Probably not in the way WithAI means it. But it does suggest that the market is still figuring out what serious AI adoption looks like in finance.
Competition is emerging. Samaya.ai, Transient.ai, Blueflame AI, and Resiliq are all building AI-driven research or portfolio tools aimed at hedge funds. WithAI's bet is that white-glove service combined with private hosting—no shared infrastructure, no data leakage—will win over firms that care as much about control as capability.
The company is hiring "fast, tasteful software engineers who like stocks" for roles in New York, according to its careers page. That's one way to put it. Perhaps more telling: they're looking for people who can build in an environment where the stakes are measured in basis points, the clients are demanding, and the technology is still being invented.
Whether that combination scales remains the open question.
