StartupVenture-backed
AIR
An inline firewall for AI agents and the skills, plugins and MCP servers they use.
AIR builds what it calls an inline firewall for AI agents and the skills, plugins and MCP servers they rely on, sitting between agents and their add-ons to stop malicious behaviour. It came out of stealth on 1 September 2026 with $50M led by Sequoia and Greenoaks.
People
Co-founder & CEO
MCP & tools · Security & sandboxes · Skills, plugins & mods
Co-founder & CTO
Security & sandboxes
Neighbours in the directory
New York, NY · Est. 2023
Red-teaming and robustness for AI agents.
Haize Labs stress-tests agents with automated adversarial attacks — prompt injection, tool misuse, unsafe actions — and ships an SDK for continuous safety evaluation in CI.
San Francisco, CA · Est. 2024
Automated evaluation and guardrails for LLM systems.
Patronus AI scores outputs against policy and business criteria in CI and production, catching regressions before users do — 'unit tests for AI'.
United States
Human approval and context engineering for agents.
HumanLayer builds tooling for human-in-the-loop agent workflows and authored '12-Factor Agents' — the widely-cited methodology for building production harnesses. Now focused on agentic code review tooling.
Connections
Relations from the Harness Hub knowledge graph, as charted in the State of Harness reports.
Built on · 1
- Model Context ProtocolProject
Sources
Verified Oct 29, 2026