When Vinod Khosla wants to buy "every available dollar" of a financing round, founders take note. That's exactly what happened with Runlayer, a seven-month-old startup building what it calls the governance infrastructure for corporate AI—a pitch that convinced Felicis to lead a $30 million Series A before anyone else got a serious look.
The round, which closed in late June, brings the New York and San Francisco-based company's total raised to $42 million. It's the kind of velocity that reflects both investor enthusiasm and, perhaps, a creeping unease among enterprises now scrambling to deploy AI agents without a clear playbook for keeping them in check.
Jake Storm, the Felicis general partner who led Runlayer's $11 million seed round back in November, preempted this Series A—venture capital shorthand for moving so fast that competitors never get a chance to bid. Khosla Ventures still managed to squeeze in. According to Fortune, which first reported the deal, Khosla himself lobbied hard for allocation.
A Governance Layer for the Agent Economy
What Runlayer sells is less a product than a response to an emerging problem: as AI agents proliferate inside companies, somebody has to make sure they don't run amok. The startup describes itself as "the default infrastructure layer governing how every corporate employee interacts with AI agents"—admittedly a mouthful, though one that captures the ambition.
Founded in 2025 by Andrew Berman, Tal Peretz, and Vitor Balocco, the company built a control plane for managing Anthropic's Model Context Protocol servers, AI skills, plugins, and agents. Think of it as middleware that sits between employees and the AI tools they're starting to use dozens of times a day. It binds identity through Okta and Microsoft Entra, enforces granular policies, detects threats at runtime, and logs everything in a tamper-evident trail.
Berman brings a track record. He previously founded Nanit, the AI-powered baby monitor that reportedly pulled in north of $100 million in revenue last year, and Vowel, a meetings tool Zapier snapped up in 2023. Peretz led machine learning efforts in the Israeli Air Force. Balocco, a staff AI engineer at Zapier and an MCP security specialist, rounds out the technical core.
In April, Runlayer achieved something called Extended Conformance (R1-R9) with the Cloud Security Alliance's Autonomous Action Runtime Management framework—a designation that, as of mid-June, made it the only vendor publicly listed at that tier in the AARM builder registry. Whether that certification translates to competitive moat remains to be seen, but it signals the company's willingness to play the compliance game early.
Traction Among Unicorns (and at Least One Fortune 500 Bank)
Customer momentum, at least on paper, looks strong. Fortune reports that Runlayer now counts more than a dozen unicorn customers, including Instacart, Gusto, Opendoor, dbt Labs, and AngelList. Gusto is among the companies using the product, according to the report. A Fortune 500 bank—unnamed, as is often the case with financial institutions testing new tech—is monitoring over 100,000 employees across 200,000 devices with the platform.
Since the seed round, the startup has announced partnerships with 1Password, Box, and Cursor. In December, it joined the Autonomous AI Interoperability Framework as a founding member alongside Anthropic, OpenAI, Google, AWS, and Microsoft—the kind of consortium membership that lends credibility, even if the practical impact of such alliances can be hard to measure.
Gartner projects that 40% of enterprise applications will embed AI agents by year-end, up from less than 5% in 2025. If that forecast holds, the market for tools to govern those agents could grow fast. Then again, Gartner forecasts tend toward optimism.
Competitive Pressure Looms
The urgency investors see in Runlayer's pitch isn't happening in a vacuum. Large incumbents—Wiz, Palo Alto Networks, Okta—are already building their own agent governance features. Whether a startup can carve out lasting share against that kind of competition depends on execution speed and whether enterprises view governance as a discrete layer or something bundled into existing security stacks.
Runlayer didn't disclose its valuation or how it plans to deploy the fresh capital. The company's LinkedIn profile suggests a headcount between 11 and 50 employees, though precise figures weren't provided. Storm and Khosla's eagerness to back the team suggests confidence that the window to define this category remains open—but perhaps not for long.
For now, Runlayer has the capital and the customer logos. What it doesn't yet have is proof that governance infrastructure will remain independent rather than absorbed into the platforms enterprises already use. That's the bet investors just made, and the one Berman and his co-founders will spend the next twelve months trying to prove out.
