GuideLast checked 16 October 2026 with AI-assisted research. Editor: Akshay Dubey.MethodologyCorrections log

This group secures what AI agents do, not only who they are: discovering agents and the tools they can call, applying guardrails at runtime, and testing for prompt injection and data leakage. It is adjacent to identity, and we list it because agent governance needs both. An agent with a clean identity can still take a harmful action, and a guardrail cannot attribute an action to an accountable person without identity underneath it.

13 companies in this category have a sourced profile; 3 of them are now owned by another company. The table restates each company's own profile and is not a ranking. Order is alphabetical. The last three columns use the capability labels defined in our methodology: Documented by vendor means the company's own public documentation describes the capability; Independently reported means a third party describes it; Not publicly verified means we found no public documentation, which is not the same as the product lacking it. Owner assignment means assigning an accountable owner to each non-human identity; it is different from acting on behalf of a user.

Companies

Scroll sideways to see all columns.

CompanyOwnershipHeadquartersFoundedFundingAI agent supportNHI discoveryOwner assignment
Aim SecurityAcquired by Cato Networks (announced 3 September 2025)Tel Aviv, Israel (as stated in June 2024, before the acquisition)2022$28M ($10M seed and $18M Series A)Documented by vendorNot publicly verifiedNot publicly verified
AIRIndependent, venture-backedNew York, USA (as reported by SiliconANGLE)2026 (as reported by SiliconANGLE)$50M (company press release); TechCrunch reports two seed rounds of $10M and $40MDocumented by vendorNot publicly verifiedNot publicly verified
Geordie AIIndependentLondon, United Kingdom; Fortune also reports a New York office2025 (company site); early 2025 per Fortune$36.5MDocumented by vendorDocumented by vendorNot publicly verified
Lasso SecurityIndependentTel Aviv, Israel and New York, USA (release dateline)2023More than $37M, as reported by SiliconANGLEDocumented by vendorDocumented by vendorDocumented by vendor
NeoIndependentBoston, Massachusetts, USA, with most staff in IsraelNot publicly disclosed$100M (seed and Series A combined)Documented by vendorDocumented by vendorNot publicly verified
Noma SecurityIndependentNew York, USA (company release); development centre in Tel Aviv (CTech)2023$132M (Series A and Series B, as reported by CTech and Globes)Documented by vendorDocumented by vendorNot publicly verified
Operant AIIndependentSan Francisco, California, USANot publicly disclosed$13.5M reported as at September 2024Documented by vendorDocumented by vendorNot publicly verified
Pillar SecurityIndependentTel Aviv, Israel and Miami, USA (release datelines)2023 (October 2023 per CTech)$9M seedDocumented by vendorDocumented by vendorNot publicly verified
Prompt SecurityAcquired by SentinelOne. Agreement announced 5 August 2025; completed 5 September 2025.Tel Aviv, Israel (per SecurityWeek, November 2024); the company's own 2024 release carried a New York datelineAugust 2023$23M ($5M seed, $18M Series A)Documented by vendorNot publicly verifiedNot publicly verified
Protect AIAcquired by Palo Alto Networks (agreement announced 28 April 2025, completed 22 July 2025)Seattle, Washington, USA (before acquisition); offices in Berlin and Bangalore2022$108.5M, as stated by the company in August 2024Not publicly verifiedNot publicly verifiedNot publicly verified
StraikerIndependentMountain View, California, USA; office in Bengaluru, IndiaNot publicly disclosed; emerged from stealth on 27 March 2025$85M (company-stated total after the Series A)Documented by vendorDocumented by vendorNot publicly verified
TumerykFounder-led; no parent company or acquisition found; investors not publicly disclosedNot publicly disclosedNot publicly disclosedNot publicly disclosedDocumented by vendorNot publicly verifiedNot publicly verified
ZenityIndependentR&D in Tel Aviv; operations and go-to-market in New York2021About $180M, as reported by SecurityWeekDocumented by vendorDocumented by vendorDocumented by vendor

Recent funding in this category

  • Announced 1 Sep 2026: AIR, Seed (second round), $40M
  • Announced 1 Sep 2026: AIR, Seed (first round), $10M
  • Sep 2026: Lasso Security, Funding round (series not stated), $30M
  • 3 Aug 2026: Zenity, Series C, $125M
  • 20 Jul 2026 (announced): Neo, Series A, $75M (reported)
  • 29 Jun 2026: Straiker, Series A, $64M
  • May 2026: Geordie AI, Series A, $30M
  • Sep 2025: Geordie AI, Seed, $6.5M

All rounds: NHI funding tracker.

Guides and frameworks for this category

Full list across all categories: company directory. How we research: methodology.

Frequently asked questions

How should you evaluate an AI agent security vendor?

Test it in a proof of concept, not on paper. Ask how it treats an AI agent differently from a service account, whether it can show that an agent acted for a specific human under a specific consent, whether it discovers agents you did not deploy, how it handles MCP connections, and which draft standards it implements at which version. The RFP question bank lists these questions with what each one tests.

Is AI agent security the same as AI agent identity?

No. Agent identity products issue and authorise identities for agents. The agent security products grouped here discover agents and the tools they can call, apply runtime guardrails and test for prompt injection and data leakage. Governance needs both: an agent with a clean identity can still take a harmful action, and a guardrail cannot attribute an action to an accountable person.

Monthly change log

Know when the NHI landscape changes.

One email a month from this reference: new deals, standards status changes and corrections. No sponsored content. Unsubscribe from any email.

We first email you a link to confirm. Privacy policy.