Skyrelis is an AI security company that controls what AI agents are allowed to do while they run. It serves security leaders and SaaS product teams who are putting agents into production and need to stop an agent from calling the wrong tool, touching sensitive data or acting outside its purpose.
Its Agentic Security Platform (ASP) sits inline between agents and the models, tools, APIs and data they reach. Every action is checked against policy before it executes, and policies are written in plain English rather than code.
At A Glance
| Vendor | Skyrelis |
|---|---|
| Category | Agentic AI identity, runtime policy enforcement for AI agents |
| Identities covered | AI agents built in house, embedded in products or supplied by vendors |
| Deployment | Inline gateway, no SDKs or code changes in the agent |
| Agentic AI | Per agent runtime policy, human approval gates, prompt and output inspection |
| Customers | Persona X |
| Market classification | Agentic AI Identity |
What Skyrelis Does
Identity and access management establishes who an agent is. It does not decide, at machine speed, whether a particular action by that agent should go ahead. Skyrelis fills that second gap. The platform keeps an inventory of enterprise agents, governs them and controls each action at runtime, producing audit-ready compliance evidence as it goes.
Rules are context aware. A decision takes account of the agent’s mission, the tools it is scoped to and how sensitive the data is, and the outcome is to allow, block or escalate. Because policy can be set “per agent, per customer, per action”, one agent can behave differently for different customers without a change to the application.
Products
Agent Security Gateway
The Agent Security Gateway is the core of the platform. Teams running internal agents, for example a Sales Ops agent updating the CRM or a Finance agent processing expenses, use it to keep each agent inside its remit. SaaS providers that ship agents inside their products, such as a booking assistant or a support copilot, use it to meet the differing security demands of their customers, including healthcare clients that need HIPAA-compliant behaviour. Its capabilities are:
- Enforcement on every agent tool call, API request and data access before execution
- Policies defined in plain English
- Prompt and output inspection
- Tool and API access control
- Sensitive data protection
- Human approval gates for escalated actions
- Audit evidence for compliance
How It Works
The gateway works as an inline proxy placed between agents and everything they call. Agent traffic passes through it on the way to LLMs, tools, APIs and data sources, and the policy engine evaluates each request before it is forwarded. Actions marked for escalation are routed to a human approver, and every decision, whether allowed, blocked or escalated, is recorded as audit evidence for compliance teams. Skyrelis says a deployment takes a reported 30 minutes, and the agent itself needs no SDK or code change. The gateway works with agents built on LangChain, CrewAI or custom code, and with the Model Context Protocol (MCP) and the Agent-to-Agent Protocol, so tool calls and agent-to-agent traffic pass through the same control point.
Identities And Systems Covered
| Internal agents | Agents built in house for functions such as Sales Ops, Finance and HR. |
|---|---|
| Product embedded agents | Agents shipped inside SaaS products, such as travel booking assistants, support copilots and financial advisory tools. |
| Vendor built agents | Agents supplied by enterprise software vendors. |
| Governed resources | LLM models, APIs and tools, and the data sources agents reach. |
Company Background
| Headquarters | San Mateo, California, United States. |
|---|---|
| Founder | Jaz Lin. |
| Stage | Pre-Series A, working with design partners and early production deployments. |
| Recognition | Included in the OWASP Gen AI Security Project solutions directory for runtime policy enforcement and agent behaviour monitoring in multi-agent systems. |
Note: This overview is compiled from Skyrelis’s publicly available information as of 3 November 2026. It is provided for information only. NHI Mgmt Group does not endorse any vendor, and nothing on this page represents the opinions of NHI Mgmt Group.