TLDRocket
Sign in

AIR raises $50M to help companies vet the skills and add-ons AI agents use

TechCrunch Ram Iyer Covered by 2 sources

AIR just raised $50M to police the tools AI agents plug into. It says companies need to vet that supply chain before an agent turns into a quiet breach point.

Based on reporting by TechCrunch, Ram Iyer — read the original for the full story.

Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error

AI agents are getting access to more company systems, and that has created a fresh problem: the software they pull in along the way. Skills, plugins, MCP servers, add-ons — all the little pieces that let an agent do real work — are becoming their own supply chain. AIR thinks that chain needs watching, so the startup is coming out of stealth with $50 million raised across two seed rounds.

The company was founded by Yair Saban, its CEO, and Niv Hoffman, its CTO. Both come out of Israel’s Unit 8200 and worked on offensive cybersecurity there. AIR’s pitch is straightforward. It finds agents inside a company, keeps checking the skills and components they use, and can block anything that fails its security rules. It also runs a marketplace of vetted add-ons and skills.

The two rounds closed within weeks of each other. Sequoia led the first, a $10 million round. Greenoaks led the second, a $40 million round. AIR also drew in a long list of angels, including Zach Frankel of Cognition, Yinon Costica of Wiz, Ofir Erlich of Eon, Anne Neuberger, Omer Adam, and Varun Anand of Clay.

Saban’s argument is that AI agents are starting to behave a bit like operating systems, but without the old discipline that came with signed drivers and other checks. The risk, he says, is not just direct attacks on the agent itself. It is poison in the content the agent consumes, or in a tool that looks fine until a package changes or a developer account is compromised.

AIR says its platform works in three layers: visibility to find agents and unapproved AI tool use, enforcement that intercepts actions in real time, and a whitelist that decides which tools and add-ons are allowed. The company says it currently filters out about 27% of the add-ons and skills it finds online. It also says it has more than 20 customers, with roughly a quarter of them large enterprises, and it is seeing the strongest demand in financial services and pharmaceuticals. The space is already crowded, though, with names like Noma Security, Zenity, Astrix Security and Operant AI all selling their own versions of agent control.

AIR says the new money will mostly go to researchers and to sales in the U.S. and Europe. The company has around 40 employees now. Sequoia’s Bogomil Balkansky framed the bet neatly: this is not a scanning problem, it is a continuous re-verification problem.

My take — AI-written commentary, not fact-checked reporting

This is the sort of security category that appears the moment companies decide they’d rather automate than think. AIR is probably right that the real business is not spotting agents, it’s constantly re-checking everything those agents touch, because that’s where the mess lives. Also, “we’ll fix it inside the model later” is a lovely slogan right up until procurement has to buy something real.

Read more about this at: TechCrunch

Related stories

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.