JetStream Security

AI is ready. Trust isn’t.

Today, we’re making something official.

JetStream Security has raised $34M, led by @Redpoint Ventures with participation from the @CrowdStrike Falcon Fund and industry leaders, to solve a critical enterprise problem: how to trust the AI systems already operating in production.

AI now calls APIs, accesses data, inherits permissions, and consumes budget. But when boards ask simple questions like where it is running, what it is doing, who owns it, and what it costs, many organizations cannot answer with confidence.

The gap isn’t model performance. It’s governance.

AI has moved from experimentation to infrastructure. It acts under credentials and makes decisions with real financial and operational impact. Without visibility and accountability at runtime, trust breaks down. When trust breaks down, adoption slows.

We built JetStream around a simple belief:

Governance enables trust. Trust unlocks advantage.

Governance is the control layer that makes AI visible, attributable, and defensible in production.

Our founding team—Raj Rajamani (CEO), Jared Phipps (COO), Jatheen Anand (CTO), and Venu Vissamsetty (Chief Architect)—has led product, engineering, and go-to-market at some of the most influential security companies of the last decade. We’ve built platforms that protect the world’s largest enterprises, scaled organizations through hypergrowth, and navigated the security challenges that emerge with every new computing paradigm.

AI is ready for takeoff.

Now it needs governance-grade trust.

Read the release: bit.ly/46X2urt

Learn more: bit.ly/4u7cTup

Agents are already running in your environment. Some you approved. Some you have never seen. Governing them comes down to five things, which we call the AI Trust Gaps:

You can't control what you can't see.
You can't approve what you can't describe.
You can't enforce what you can't verify.
You can't trust what you can't attribute.
You can't scale what you can't afford.

On November 10th at 1pm PT, our solutions engineers will show you how to close all five in a live walkthrough of the JetStream SAIG Platform™. The JetStream Pre-Flight Demonstration covers:

1.) The five AI Trust Gaps, and why enterprise AI stalls before production
2.) A live product demo
3.) A first look at everything released since the last session, including what is not on our website yet

Your questions, answered live. Come hear what your peers are asking, and leave knowing whether JetStream meets your needs.

ow.ly/VbKI5105eF9

#AIGovernance #AISecurity #AIAgents

Nothing else matters… when you’re with the right team. 🤘

The JetStream team traded dashboards for devil horns at Metallica at the Sphere in Las Vegas. Hearing the crowd, and our team, singing as one was a reminder of why we do this: great things happen when people come together.

Security is a team sport. The best moments aren’t always in the boardroom. Sometimes they’re a few rows up, arms in the air, next to the people you trust to build something better.

A huge thank you to our partners at Vega for joining us and helping make this a night to remember, and to CISO Experiences for hosting an unforgettable experience.

To the JetStream crew: thanks for bringing the energy. 🔥

#JetStream #TeamBuilding #Metallica #TheSphere #CISOExperiences #CyberSecurity #CompanyCulture

Agents are already running in your environment. Some you approved. Some you have never seen.

On November 10th at 1pm PT, our solutions engineers will walk through the JetStream SAIG Platform™ live at the JetStream Pre-Flight Demonstration. You will see the platform discover every AI asset, sanctioned or not, and enforce guardrails on browsers, endpoints, and coding agents.

Bring your questions. Leave knowing whether JetStream meets your needs.

ow.ly/Oovf5103xaV

#AIGovernance #AISecurity

Keep a human in the loop" is the right answer. It is also where most security teams are about to get it wrong.

Joe Poyner makes the case in his latest piece. An approval queue a person clears 400 times a day is not a control. It is automation with a signature on it.

The research backs him up. Run a monitoring task with automation and people miss 41% of the events it never flags (Skitka, Mosier, and Burdick, 1999). Take the automation away and the same people miss 3%.

So the loop has to be built to survive human attention, not assume it. Approvals rare enough to take seriously. Rich enough in evidence to verify independently, the actual diff and the actual target. Signed with a name that owns the outcome.

Read it: jetstream.security/insights/ai-ag…