Updated
Updated · TechCrunch · Sep 1
AIR Raises $50 Million to Vet AI Agent Skills as 27% of Add-Ons Fail Checks
Updated
Updated · TechCrunch · Sep 1

AIR Raises $50 Million to Vet AI Agent Skills as 27% of Add-Ons Fail Checks

2 articles · Updated · TechCrunch · Sep 1

Summary

  • $50 million in two seed rounds has brought AIR out of stealth, with the AI security startup targeting the tools, plugins and MCP servers used by enterprise AI agents.
  • AIR says companies need supply-chain oversight as agents gain broader access to databases, internal systems and the internet, creating risks from poisoned content and compromised add-ons rather than direct attacks.
  • Its platform discovers agents across a company, flags unapproved AI use, intercepts actions such as loading skills or fetching web content, and checks tools against a continuously updated whitelist; AIR says 27% of online add-ons it reviews are blocked.
  • More than 20 customers have signed on so far, about a quarter of them large enterprises, with strongest demand in regulated sectors including finance and pharmaceuticals.
  • Sequoia led AIR's $10 million first seed round and Greenoaks led a $40 million second round, funding hiring and U.S.-Europe expansion in a crowded market where Zenity recently raised $125 million and Noma raised $100 million.

Insights

With 27% of AI add-ons rejected, what hidden dangers are lurking inside the autonomous enterprise tools you already trust?
How did a fake AI skill bypass traditional security scanners to infiltrate 26,000 corporate agents entirely undetected?