Token Security
Non human identity security platform that discovers and governs machine identities and AI agents across cloud, SaaS and on premises systems, with posture management, threat detection, least privilege, a conversational Token AI Agent and MCP Server, and Enzo for building identity workflows in natural language.
Token Security is a non human identity security company, headquartered in Tel Aviv with a New York presence, that secures the machine identities and AI agents spreading across enterprises. Founded in 2023 by Itamar Apelblat and Ido Shlomo, it has raised about twenty seven million dollars across a seven million dollar seed round led by TLV Partners and a twenty million dollar Series A in January 2025, backed by Notable Capital and SNR, and was named to The Information's list of the fifty most promising startups of 2025.
Non human identities such as service accounts, API keys, OAuth tokens, secrets and AI agents often run with broad, permanent access and no clear owner. Token discovers them across cloud, SaaS, on premises and CI/CD systems, maps ownership and access paths, and enforces least privilege, with posture management, identity threat detection and response, access reviews, and policies for how the customer's own AI agents are created and what they may reach.
On the agent side, the Token AI Agent and the Token MCP Server, launched in May 2025, let teams query inventory, risk and blast radius in plain language, and the MCP Server works with Claude, ChatGPT, Gemini and Cursor and can generate remediation scripts and initiate actions. Enzo, launched in June 2026, lets security and identity teams build their own live applications, workflows and automations in natural language, including Slack and email notifications, ticketing, lifecycle management and remediation actions, with the platform also triggering remediation on dynamic risk thresholds.
Token's trust center lists ISO/IEC 27001:2022 and SOC 2 Type 2. Enzo recommends remediation and offers one click actions, and its actions are logged. Token names no approval step before an automated action runs, no run trace of its own agent, no endpoint or tool list for the MCP Server and no hosting location for the service. Token does not name the model behind its agent. It fits a security team that needs to inventory and govern the machine identities and AI agents across a mixed estate and wants to build its own identity workflows on that inventory. A team looking for a general purpose agent builder or a secrets only vault will find it focused on identity governance.
Vendor details
Canonical URL
https://token.security
Category
Security / SOC agent
Subcategory
Non human identity and AI agent security
Funding status
Independent, headquartered in Tel Aviv with a New York presence, founded in 2023 by Itamar Apelblat and Ido Shlomo. Token Security has raised about twenty seven million dollars across a seven million dollar seed round led by TLV Partners and a twenty million dollar Series A in January 2025, backed by Notable Capital and SNR with angel investors including Kevin Mahaffey, founder of Lookout, and Shlomo Kramer, co-founder of Cato Networks. It was named to The Information's list of the fifty most promising startups of 2025.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
Token Security's integrations catalog covers AWS, Azure and GCP, identity providers such as Okta and Microsoft Entra ID, source control and CI/CD, secrets vaults such as HashiCorp Vault and CyberArk, ticketing such as Jira and ServiceNow, Slack and Microsoft Teams, SIEM and log tools, and AI platforms whose agents it discovers. Enzo applications act through notifications, ticketing, lifecycle and remediation workflows, and the Token MCP Server exposes the identity inventory to Claude, ChatGPT, Gemini and Cursor.
In practice
AI agents and service accounts are multiplying across your cloud and on premises systems with broad permissions and no owner, and human style controls do not catch them. Token Security discovers every non human identity across those systems and enforces least privilege before one is abused.
A leaked API key or over privileged token is the entry point for a growing share of cloud breaches, but your team cannot even inventory them. Token Security maps ownership, scores risk, and triggers remediation on dynamic thresholds.
Security teams drown in fragmented identity data and cannot answer which inactive machine identities are riskiest. The Token AI Agent lets them ask in plain language across inventory, posture, secrets, and threats and get guided remediation.
Sources & related URLs
Agentic Index coverage score
6.5 / 14 capabilities · 46%
| Integrations & Tool Calling | Full |
|---|---|
|
Enzo applications send Slack notifications, run email workflows, open tickets, handle lifecycle management and run remediation actions against connected systems. The Token MCP Server can generate remediation scripts and even initiate actions. Token integrates with ticketing (Jira, ServiceNow), chat (Slack, Microsoft Teams), the major clouds, identity providers and secrets vaults. Sourcetoken.security/blog/introducing-enzo-by-token-security-the-ai-native-identity-security-application-builderread 2026-09-28 |
|
| Workflow Orchestration | Full |
|
Enzo lets security teams describe and build their own live applications, workflows and automations in natural language, with workflow triggers and recurring identity operations. The platform also triggers remediation workflows on dynamic risk thresholds. Sourcetoken.security/blog/introducing-enzo-by-token-security-the-ai-native-identity-security-application-builderread 2026-09-28 |
|
| Knowledge Grounding & RAG | Full |
|
The Token AI Agent, the MCP Server and Enzo all draw on the same live identity context, which Token keeps current across the customer's own estate. It maps human and non human identities, AI agents, cloud and SaaS accounts, secrets and credentials, each with its owner and blast radius. Sourcetoken.security/blog/introducing-the-first-nhi-mcp-server-ai-powered-smarter-driving-fast-remediationread 2026-09-28 |
|
| Human Oversight & Guardrails | Partial |
|
Remediation in Enzo comes as a recommendation with one click actions, so the decision stays with the team. Approvals are among the workflows Enzo can support. Token names no approval step that holds an agent action until a person signs off. Its policy controls govern the customer's own AI agents, not Token's. Sourcetoken.security/blog/introducing-enzo-by-token-security-the-ai-native-identity-security-application-builderread 2026-09-28 |
|
| Security, Identity & Governance | Partial |
|
Token is ISO/IEC 27001:2022 certified and holds SOC 2 Type 2. It names no customer facing SSO, roles or admin audit log for its own console. Sourcetrust.token.securityread 2026-09-28 |
|
| Observability & Auditability | Partial |
|
Actions taken by Enzo applications are audited with full audit logging, so Token's own automation leaves an action log. Token names no run trace of the Token AI Agent or of Enzo's steps. Its immutable logs and audit trails record what the customer's AI agents do, not what Token's own agent does. Sourcetoken.security/blog/introducing-enzo-by-token-security-the-ai-native-identity-security-application-builderread 2026-09-28 |
|
| Memory & State Persistence | Not documented |
|
No memory is named for the Token AI Agent or Enzo, and Token gives no scope or lifetime for one. The identity inventory and behavioral baselines are Token's data model of the customer's estate. Sourcetoken.security/productread 2026-09-28 |
|
| Deployment & Data Residency | Not documented |
|
No hosting region is published for the service itself, and Token names no option to run in the customer's environment. Its coverage of on premises, hybrid and cloud environments is where it discovers identities, not where it runs. Sourcedocs.token.securityread 2026-09-28 |
|
| Prebuilt Agents, Templates & Packs | Partial |
|
Token ships one conversational agent, the Token AI Agent, and the MCP Server. Example Enzo applications include a Clean Exit Checklist and access review workflows. Token names no library of prebuilt agents or templates that a customer installs. Sourcetoken.security/blog/introducing-enzo-by-token-security-the-ai-native-identity-security-application-builderread 2026-09-28 |
|
| Triggers & Channel Coverage | Full |
|
The platform triggers intelligent remediation workflows based on dynamic risk thresholds and contextual awareness, so detected risk starts the work without a person. Enzo adds workflow triggers and recurring identity operations. Sourcetoken.securityread 2026-09-28 |
|
| Model Flexibility & Routing | Not documented |
|
The model behind the Token AI Agent and Enzo is not named, and no choice of provider is offered. Customers can point their own assistant, such as Claude, ChatGPT, Gemini or Cursor, at the Token MCP Server, and that assistant then uses Token's tools on its own model. The AI platforms Token integrates with are estates it discovers. Sourcetoken.security/integrationsread 2026-09-28 |
|
| APIs, SDKs & MCP Extensibility | Partial |
|
The Token MCP Server lets Claude, ChatGPT, Gemini and Cursor query inventory, risk and blast radius, and initiate actions. Token publishes no endpoint, authentication scheme or tool list for it, and names no public REST API or SDK. Product documentation lives at docs.token.security. Sourcetoken.security/blog/introducing-the-first-nhi-mcp-server-ai-powered-smarter-driving-fast-remediationread 2026-09-28 |
|
| Testing, Debugging & Optimization | Not documented |
|
The Token AI Agent and Enzo applications have no published test harness with scored cases, preview or quality gate. Attack path analysis and compliance validation test the customer's estate, not Token's agent. Sourcetoken.security/productread 2026-09-28 |
|
| Browser & Computer Use | Not documented |
|
Integrations with clouds, identity providers and vaults run through APIs, and Token names no agent that operates a browser or desktop. Sourcetoken.security/integrationsread 2026-09-28 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Product updates
Rehearsal, a new mode in Token AI, takes a goal in plain words, such as finding which AI agents have no owner, and carries it out in the user's own tab. It plans a route through the platform, opens the pages, sets the filters, highlights the matching results and writes up the answer. Reading is automatic, but it asks before saving anything, never fills in a secret field, and stops the moment the user takes over with a click.
Bears on: Browser/computer use
View sourceToken Security customers can now install a Token plugin from the Grok Build marketplace, so developers and the Grok Build agent itself can query Token data from the editor, for example what a GitHub token can reach and who owns it. Token also reads the Grok Build Admin API to see which users are enabled and which MCP servers and tools each team has set up, and combines that with EDR data to find what each person actually runs.
Bears on: MCP / tool calling / API
View sourceToken Security introduced Enzo, an AI-native identity security application builder that allows users to create live applications, workflows, dashboards, and automations using natural language. The solution is built directly into the Token Security platform and requires no additional infrastructure or coding.
Bears on: Workflow orchestration
View sourcePricing
Not public; quoted through enterprise engagement with no self serve tier
enterprise subscription; basis not disclosed
What is public
No list pricing. Token Security sells through enterprise engagement, with no public rates and no self serve tier.
Billing mechanics
Rates and billing basis are not disclosed. Inference, not stated by the vendor: an enterprise subscription scoped to the identity estate, the environments covered and the modules enabled.
Cost watchouts
Inference, not stated by the vendor: broad coverage across cloud, on premises and CI/CD and larger identity estates may raise cost.
Variable cost rationale
Inference, not stated by the vendor: likely scoped to the number of non human identities, the environments covered and the modules enabled, so cost would grow with the size of the identity estate.
Additional watchouts
Confirm how pricing scales with the number of identities discovered and the environments in scope, and whether modules like threat detection and response are priced separately.
Sales call required
Yes, required for paid access
Free / trial
Demo on request; no public free tier
Key ambiguities
No public rate is disclosed, and the pricing basis, whether per identity, per environment, or platform tier, is not clear.
Related vendors
- 7AI — Agentic SOC from the Cybereason founders: domain specialized agents…
- Abnormal AI — Behavioral AI email security with three named autonomous agents: a…
- AirMDR — AI native MDR whose AI analyst investigates every alert and writes…
- Airrived — Agentic OS that unifies SOC, GRC, IAM, vulnerability management, IT,…
- Andesite — Human-AI SOC platform where customers build their own agents and…
- Anvilogic — Agentic security operations platform that works on top of existing…
Alternatives to Token Security
The closest documented capability profiles to Token Security among security and SOC agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Astelia8.0 / 14Fuller documented coverage on Human Oversight & Guardrails and Security, Identity & Governance
- Clutch Security6.0 / 14Adds documented Deployment & Data ResidencyToken Security vs Clutch Security →
- Equixly6.0 / 14Fuller documented coverage on Observability & Auditability
- Intezer7.0 / 14Adds documented Memory & State Persistence
- Pi6.0 / 14Adds documented Memory & State Persistence
- Quantro Security7.0 / 14Fuller documented coverage on Human Oversight & Guardrails and Prebuilt Agents, Templates & Packs
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded