When AI agents start booking your calendar, filing expense reports, and accessing company payroll data, who's watching the watchers?
That question landed Archestra—a London startup—a $10 million seed round led by Harry Stebbings' 20VC on June 2. The capital injection, which follows a $3.3 million pre-seed raised just nine months earlier, brings total funding to $13.5 million. Not bad for a team of seven building what amounts to a bouncer for enterprise AI systems.
The pitch is deceptively straightforward. As companies scramble to deploy autonomous AI agents that can actually do things—not just answer questions—those agents need access to sensitive internal tools. Slack channels. ServiceNow tickets. GitHub repositories. BambooHR payroll systems. Archestra's open-source platform functions as what CEO Matvey Kukuy and co-founder Ildar Iskhakov call an "enterprise MCP gateway," essentially a secure checkpoint that governs which AI agents can touch what, when, and how.
It's infrastructure for a problem that barely existed eighteen months ago but now keeps CTOs up at night.
The Security Arms Race Nobody Saw Coming
The investor lineup suggests the market takes this seriously. Beyond 20VC, the round drew backing from 20 Product (whose participants include Will Wu, chief product officer at Match Group), Visible Ventures, and Tenacity Capital, the firm launched by AJ Tennant after stints at Slack and Glean. The angels read like a reunion of enterprise security's greatest hits: Datadog's CEO Olivier Pomel, HubSpot CMO Kieran Flanagan, Nutanix Chief AI Officer Debo Dutta, and Gerhard Eschelbeck, Google's former CISO.
That roster matters, perhaps more than the founders expected. Archestra's focus on securing the Model Context Protocol—the emerging standard for how AI systems interact with external tools—positions it squarely in the path of what one investor called "the most enterprise-critical chokepoint in generative AI."
The competitive landscape bears that out. In January 2026, WitnessAI raised $58 million. Oasis Security landed a $120 million Series B in March. General Analysis pulled in $10 million in April. All are chasing variations of the same thesis: that agentic AI represents both enterprise computing's next frontier and its next major attack surface.
Archestra's approach hinges on role-based access controls, per-tool-call credential management, guardrails, and audit logging—the unglamorous plumbing that security teams demand before signing off on anything that touches production systems. The core platform is open source; enterprise features arrive via self-hosted Docker deployments, a distribution strategy that should feel familiar to anyone who's watched HashiCorp or GitLab navigate similar paths.
Fortune 500 Traction—Or a Head Start?
Here's where it gets interesting. Despite employing just seven people, Archestra claims multiple Fortune 500 companies are already running its platform in production. Dealroom, the market intelligence firm, pegged the count at four enterprise clients as of a recent Dealroom report in June 2026, though the company has declined to name logos.
That's either impressive early validation or a fragile dependency on a handful of design partners. Likely both.
Kukuy and Iskhakov aren't first-time founders stumbling into enterprise sales. They previously built Amixr, an incident management tool that Grafana Labs acquired in November 2021. Kukuy also co-founded Keep, reportedly acquired by Elastic—giving them both exits under their belts and a working understanding of how to sell infrastructure software to organizations that move slowly and demand reliability.
Still, scaling from seven employees and a few early adopters to the kind of market penetration that justifies a venture-backed trajectory won't be simple. The company faces not just direct competitors but also the looming possibility that cloud hyperscalers—AWS, Google Cloud, Microsoft Azure—could potentially decide to bundle similar governance capabilities into their AI platforms, a risk that has historically challenged infrastructure startups. Enterprise software history is littered with promising infrastructure plays that got commoditized by incumbents who could afford to give away features as loss leaders.
What Comes Next
The fresh $10 million will presumably fund hiring (engineers, go-to-market talent) and broader enterprise outreach as AI agent deployments accelerate. Industry observers have taken to calling this "the year enterprise AI finally gets to work," which may or may not age well as a prediction but certainly captures the current mood.
Archestra's open-source strategy offers some insulation from the build-versus-buy calculus that typically kills infrastructure startups. Developers can kick the tires for free; enterprises pay for governance, compliance features, and support when legal and security teams get involved. It's a familiar wedge.
Whether the company can scale fast enough to stay ahead of both competition and consolidation remains an open question. Seven people can move quickly. They can also get overwhelmed when Fortune 500 procurement cycles collide with the chaos of a market still figuring out what "agentic AI security" even means.
For now, Archestra has capital, credible backers, and a problem set that isn't going away. What it does with the next twelve months will determine whether it becomes the category winner or a cautionary footnote in someone else's origin story.
