JetStream Security is an AI governance company that helps enterprises see, approve and control the AI agents, models and tools their teams are adopting. It is built for security, risk and finance leaders who need to know which AI is running, who is accountable for it and what it costs.
Its JetStream Security-First AI Governance Platform gives those leaders a single record of AI use across the estate, which they can enforce while AI systems run.
At A Glance
| Vendor | JetStream Security |
|---|---|
| Category | AI governance, agentic AI identity |
| Identities covered | Human users, AI agents, non-human identities, service accounts |
| Deployment | Control plane governing AI systems at runtime |
| Agentic AI | Agentic Identity, Verified MCP Catalog |
| Certifications | FedRAMP High / Class D, SOC 2 Type I |
| Funding | Seed funded (March 2026) |
| Market classification | AI Governance & Compliance; also Agentic AI Identity |
What JetStream Security Does
Enterprises are taking on AI agents, models and workflows faster than they can track or approve them. JetStream’s mission is to speed up AI adoption by replacing that uncertainty with trust. Its approach treats each AI workflow like a governed asset with a known design, a named owner and a budget, and checks it continuously against that record. Because every stage shares the same record, a drifting agent can be traced straight to its owner and its cost, and security, risk and finance teams all work from the same view.
Products
AI Visibility
AI Visibility removes blind spots by continuously discovering AI across SaaS, endpoints, cloud, APIs and internal systems. The result is an inventory that stays current as teams try new tools, which is the starting point for every other control.
AI Design Control And JetStream AI Blueprints
AI Design Control turns that raw inventory into an approved design, kept as a versioned record. JetStream AI Blueprints apply this to agentic workflows: they list the agentic tasks running across any model, platform or team, record who authorised what and when, and enforce acceptable-use policies and compliance guardrails at the point of execution.
Runtime Governance
Runtime Governance compares live AI activity with the approved Blueprint and flags drift, such as a swapped model or an expanded permission. Each change leaves a record against the Blueprint, giving audit and risk teams a history of how a workflow evolved. Alongside it, AI FinOps Accountability gives finance and platform owners the economics of AI use, broken down by model, agent, workflow and owner.
JetStream AI Kill Switch
The Kill Switch stops a single AI agent quickly while the rest of the estate keeps running. Operators see the impact before they act, contain the agent without redeploying, and keep a full record of the stop and the restart. It covers cases such as a compromised agent, an agent that has exceeded its budget, or one that must be halted for policy or regulatory reasons.
How It Works
JetStream runs as a control plane that governs AI systems at runtime. Discovery draws on APIs, browser activity, endpoint signals, cloud logs and in-line observability. Telemetry sits in hot storage for 15 days by default, and retention can be extended to suit the customer.
| Component | What it does |
|---|---|
| AI Manifest | Holds the AI inventory. |
| AI Hub | The point through which governed AI requests pass. |
| Identity Broker and Key Broker | The components behind Agentic Identity, sitting between agents and AI providers. |
| AI Blueprints and AI Drift Detection | Hold the approved design and detect divergence from it. |
| Verified MCP | Supplies vetted MCP servers to agents. |
Identities And Systems Covered
| Human users | Accountability for AI workflows is assigned to people |
|---|---|
| AI agents | Every agent workflow has a provable owner |
| Non-human identities | Service accounts and provider secrets |
| Environments | SaaS, endpoints, cloud, APIs and internal systems |
| MCP servers | Tools that agents call |
Agentic AI Capabilities
Agentic Identity
Agentic Identity ties every AI workflow to an accountable owner, covering humans, agents and non-human identities together. Raw provider secrets are swapped for virtual, revocable keys issued by the Identity Broker and Key Broker. Because agents no longer share one provider secret, a single agent can be revoked without breaking the other workflows.
Verified MCP Catalog
Every MCP server in the catalogue is “ingested, scanned, hardened, and cryptographically attested before it ships”, and an agent can call it only after that process. Control works at the level of individual tools, so a read-only tool can be allowed while a destructive one on the same server is blocked, and all requests pass through one hub.
Company Background
| Launched | 3 March 2026 |
|---|---|
| Headquarters | Santa Clara, California, United States |
| Founders | Raj Rajamani (Founder and CEO), Jared Phipps (Founder and COO), Venu Vissamsetty (Founder and Chief Architect) and Jatheen (AJ) Anand (Founder and CTO) |
| Funding | A $34 million seed round led by Redpoint Ventures, with the CrowdStrike Falcon Fund taking part |
| Angel investors | George Kurtz, Assaf Rappaport and Frederic Kerrest |
| Public sector | The FedRAMP High posture extends to GovRAMP through reciprocity |
Note: This overview is compiled from JetStream Security’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.