Token Security

A machine-first, agentless platform that discovers, owns, and right-sizes every non-human identity — service accounts, secrets, API keys, and now AI agents and MCP servers — across cloud, SaaS, and on-prem.

One-liner — Identity-first security for non-human and AI-agent identities: find every machine credential, attach a human owner, and enforce least privilege from one dashboard.

Categories — non-human-identity

What it does

Token Security tackles the problem that enterprises typically have far more machine identities than human ones (the vendor cites ~45x), and almost no one owns or governs them. The platform discovers non-human identities (NHIs) across the stack — cloud IAM roles and credentials, SaaS service accounts, API tokens/keys, secrets, and increasingly AI agents, MCP servers, and custom GPTs — then builds a unified graph correlating each identity with its secrets, permissions, usage, and a responsible human owner. From there it supports lifecycle management (creation through retirement), least-privilege “right-sizing,” behavioral anomaly detection, and remediation workflows.

The pitch a CTO repeats: “Hackers don’t break in; they log in.” The value is removing the blind spot around orphaned and over-privileged machine credentials — the kind that turn one leaked key into a breach — and extending that same governance to the explosion of AI-agent identities.

Where it sits in the stack

Primary category: non-human-identity (Foundation layer). Its job is discovery, ownership, and least-privilege governance of machine and agent identities — the identity-plane control that sits beneath secrets management and IGA. It does not inspect prompts or content; it constrains access to sensitive data and outbound reach indirectly, by ensuring an agent or service account can only reach the data and systems it is entitled to, and by killing orphaned/over-privileged credentials an attacker (or a compromised agent) would otherwise use to exfiltrate. It does not screen untrusted input — that is the job of ai-runtime-security guardrails. Trust-zone relevance: it polices who/what can cross zone boundaries at the identity layer.

Deployment & architecture

• SaaS, agentless. Connects via API/cloud connectors to map credentials across AWS, Azure, GCP, SaaS apps, CI/CD (e.g., GitLab), and on-prem systems — no endpoint agent required.
• Unified identity graph correlating non-human identities, their secrets/permissions, usage patterns, and human owners; least-privilege right-sizing and lifecycle automation on top.
• AI-agent / MCP coverage — discovers and audits AI agents, MCP servers, and custom GPTs. Ships an MCP Server so third-party AI tools (Claude, ChatGPT, Gemini, Cursor) can query Token’s NHI data, plus a native “Token AI Agent” in the platform UI.
• Integrations — major cloud providers, CI/CD, secrets managers, and identity providers; a centralized dashboard for posture, alerts, and remediation.

Positioning & differentiators

Token positions as machine-first / identity-first — built natively for NHIs and agentic AI rather than retrofitted from human IAM, and explicitly not a prompt/content guardrail. Its differentiator claims (marketing) are the unified identity-secret-permission-owner graph (“deep identity intelligence”), agentless deployment, and early moves into AI-agent and MCP governance (it markets an “industry-first” MCP server for NHI security).

Versus nearest neighbors in non-human-identity: oasis-security is the closest peer — another Israeli NHI-discovery/governance platform, better funded (raised ~$40M by 2024). astrix-security came at NHI from the third-party app/OAuth-integration angle (and per the registry is flagged as acquired by Cisco — verify). aembit is more of a workload IAM / access issuer (it brokers secretless workload-to-service access) rather than a discovery/posture tool. entro-security leads with secrets/NHI posture and secrets-security lineage. clutch-security is another NHI-lifecycle entrant. cyberark is the incumbent — secrets/PAM plus its Venafi (machine-identity/certificates) acquisition — and competes top-down from the enterprise PAM install base. Token’s wedge is breadth of agentless discovery plus the AI-agent identity story.

Ownership, funding & M&A

Independent, venture-backed. Founded 2023 by Itamar Apelblat (Co-Founder & CEO) and Ido Shlomo (Co-Founder & CTO), who met at Israel’s Unit 8200. Emerged from stealth ~mid-2024 with a $7M Seed (led by TLV Partners; angels including Shlomo Kramer). Series A: $20M on 2025-01-28, led by Notable Capital, with existing investor TLV Partners and executives from Palo Alto Networks, CrowdStrike, Check Point and Venafi — bringing total funding to $27M. HQ is Tel Aviv with a New York/US presence (CEO relocated to the US in Feb 2025). No acquisition found — the seed registry carried no M&A flag and research found none; ownership independent, confidence high for the funding/founding facts (multiple corroborating sources incl. TechCrunch), though figures are as of the Jan-2025 round and may be stale.

CTO / hedge-fund lens

NHI governance is generally Day-2 for a hedge fund — you stand up human SSO/MFA, secrets management, and logging first — but it climbs toward Day-1 the moment you run cloud-native infra with many service accounts/keys, or start deploying AI agents that hold their own credentials and act on systems. For a fund worried about an over-privileged or orphaned key (or a rogue/compromised agent) reaching positions, trade systems, or MNPI, Token’s discovery-plus-ownership-plus-least-privilege loop is the relevant control. It is not an SR 11-7 / model-risk tool and does not do comms surveillance or content DLP — pair it with those. Fit is medium: most valuable to larger, cloud- and AI-heavy shops; a small fund living mostly in SaaS with few machine identities may get enough from its IdP and secrets manager. As a Series-A startup, weigh vendor maturity/longevity against incumbents like cyberark.

Competitors / alternatives

oasis-security, aembit, astrix-security, entro-security, clutch-security, cyberark, natoma, silverfort

Open questions / to verify

• Post-Series-A funding/round (figures are as of Jan-2025); current valuation and headcount.
• Whether the legal/operating HQ has formally moved to the US or remains Tel Aviv with a US office.
• Depth of AI-agent/MCP governance in production vs. roadmap (marketing vs. shipped).
• Vendor-confirmed customer count and any SOC 2 / compliance attestations.
• Whether deployment is fully agentless for on-prem coverage or needs collectors.

Sources

• Token Security raises $20M (PR Newswire) — fetched 2026-06-28 — supports: $20M Series A (2025-01-28, Notable Capital), $27M total, founders/roles, HQ, agentless SaaS, customers (HPE, Hibob); confidence: high (vendor PR primary on funding).
• Token Security (vendor site) — fetched 2026-06-28 — supports: current AI-agent + NHI product framing, identity types (service accounts, API keys, secrets, MCP servers, custom GPTs), agentless SaaS, integrations, MCP server, differentiators; confidence: med (marketing).
• Hackers are targeting machine identities; Token Security raised $20M (TechCrunch) — fetched 2026-06-28 — supports: founded 2023, Unit 8200 founders, Apelblat CEO/Shlomo CTO, HQ/US relocation, competitor framing (Oasis, CyberArk); confidence: high (independent press).

History

• [2026-06-28] Stub created from seed registry.
• [2026-06-28] Researched; established founded 2023 (ex-Unit 8200 founders Itamar Apelblat CEO & Ido Shlomo CTO), HQ Tel Aviv + New York, $7M Seed (TLV Partners) + $20M Series A (2025-01-28, Notable Capital) = $27M total, independent/no M&A. Documented agentless SaaS NHI-discovery + AI-agent/MCP governance; positioned vs oasis-security, aembit, astrix-security, entro-security, clutch-security, cyberark. Set status: researched, confidence medium. 3 sources cached.