Every proof-of-concept engagement uncovered something malicious. Not most. Every single one.
That's the pitch Shahar Bahat has been delivering as he takes Pluto Security to market, and it seems to be working. The startup emerged from stealth last week with $42 million in funding—an $8.5 million seed led by TLV Partners followed by a $33 million Series A from Jump Capital. Already, Pluto claims to protect more than 1.5 million endpoints for customers including Oscar Health, Workato, Sixt and AlphaSense.
The trajectory reflects broader investor appetite for anything that promises to secure AI agents before they go rogue. Rein Security raised $25 million in early October for runtime agent protection. Hush Security pulled in $30 million in July with Akamai backing it. AIR Security launched in September with $50 million for what it called an "AI agent firewall." Harmonic Security added another $50 million days after Pluto's announcement, focused on governing how employees use AI tools.
Bahat, who previously held senior roles at Microsoft and served in Israel's elite Unit 8200 intelligence unit, frames the challenge bluntly. "We're moving from user behavior to agent behavior," he told TechCrunch.
Pluto's platform monitors AI agents, tools, skills, MCP servers and employee-built automations across endpoints, identifying security risks and enforcing policies in real time. It ingests data from existing endpoint detection and mobile device management systems while deploying what the company describes as lightweight sensors. The system can intervene mid-session, forcing an AI agent to request user confirmation before executing over-privileged commands or blocking risky actions outright.
The founding team brings considerable technical pedigree. Co-founder and chief product officer Tal Ofer Regev led cloud application security engineering at Palo Alto Networks following the Bridgecrew acquisition. CTO Gil Maman also has roots in Bridgecrew, Microsoft and Unit 8200. It's a tight circle—the kind of founding roster that tends to open doors in enterprise cybersecurity.
Pluto says it began commercial sales earlier this year and has since crossed multi-million dollar annual recurring revenue with more than 40 enterprise customers. Headcount sits at more than 50 employees split between the U.S. and Israel.
Seventy percent of its sales pipeline, the company claims, flows from inbound interest tied to Plutonium—a free risk catalog and MCP scanner it released alongside its go-to-market push. Plutonium has drawn thousands of users, Pluto said, though it declined to share more granular adoption metrics.
In September, Pluto's research team published findings showing 179 exposed MCP servers, 147 of which accepted unauthenticated requests. The research served dual purposes: it highlighted a genuine vulnerability in the emerging Model Context Protocol ecosystem, and it provided a stark proof point for Pluto's broader thesis.
Bahat repeated the claim to TechCrunch about malicious components surfacing in every proof of concept. "In general, I'll say there was not one POC in which we entered and didn't find malicious stuff," he said. It's the kind of statement that raises questions about baseline security hygiene at the companies being audited, but it also underscores how nascent—and exposed—many organizations remain as they bolt AI agents onto existing infrastructure.
The investor roster extends beyond the lead firms. Skinos Ventures, AlphaDrive Ventures and Ansa Capital participated, along with angels including Armis founders Yevgeny Dibrov and Nadir Izrael, Descope co-founder Slavik Markovich and Andreessen Horowitz partner Assaf Ezra, according to Calcalist.
"What stood out to us about Pluto was a fundamentally different approach to a rapidly growing problem," Rona Segev, co-founder and managing partner at TLV Partners, told the Israeli business publication.
Pluto plans to deploy the capital toward product expansion, security research and building out its U.S. presence. Jump Capital framed the investment as backing "the next generation of endpoint security built for the agentic era," according to a LinkedIn post reported by PYMNTS.
Whether the market can sustain this many well-funded players chasing overlapping problems remains an open question. For now, the money keeps flowing into anything promising to keep AI agents on a leash.
