New York cybersecurity startup Astelia used Black Hat USA 2026 to launch a suite of agentic AI agents that automate the noisiest, most manual parts of vulnerability management. The pitch: only about 1% of vulnerabilities in a typical enterprise are actually reachable and exploitable, and the rest are burning security cycles.
Agents That Close The Exposure Loop
The new agent suite runs the vulnerability lifecycle end-to-end. It ingests newly disclosed and zero-day CVEs, scores their reachability against Astelia's network model, assesses operational impact and coordinates fixes across security and IT teams. Astelia says the agents draw on more than 100 MCP integrations, with human approvers built into every major decision and full audit logs on every action.
Reachability Beats CVSS
Astelia's core platform maps enterprise networks read-only, then uses agentic AI to correlate topology with the technical prerequisites for exploiting each CVE. That evidence-based approach produces attack-path graphs showing exactly how an attacker would traverse a network from an exposed host to "crown jewel" systems, letting teams pick the fix that shuts down the most routes.
Funding And Customer Traction
Astelia recently raised $35 million in combined seed and Series A funding led by Index Ventures and Team8, with participation from Holly Ventures. Customers include a Fortune 50 energy company, a Fortune 100 financial firm, Bilfinger, Syniverse, UCT and AlphaSense. The company was founded in 2024 by former Israeli national red-team leaders Alon Noy, Nadav Ostrovsky and Roy Rajwan.
Crowded Agentic Security Market
Astelia's launch lands in a rapidly consolidating agentic-security field. This same week Tanium unveiled Atlas MCP and attack-path mapping at Black Hat, and Anthropic just closed a $10 billion compute deal with Volta to power the models many of these agents rely on.
Reporting based on Astelia press materials and coverage from SecurityWeek, GovInfoSecurity and Newswire.