About

We started with one idea. You cannot control what you cannot see.

AI agents arrived inside organizations faster than any security practice could follow. House of Seal exists to give operators the visibility that every other control depends on.

Why we started

The same story, in every environment we looked at.

We came to this from research on AI security and from building the infrastructure that agents run on. Every time we looked at a real environment, the picture was the same. People used AI tools nobody had approved. Agents set up for a project kept running long after the project, and the person, had moved on. Credentials meant for a person were used by a process at three in the morning. Nobody was lying about it. Nobody knew.

Security teams had tools for users and tools for servers. An agent is neither, so it fell through. The gap was visibility, and everything else, policy, response, compliance, was waiting on it.

What we believe

Security starts with visibility.

Before a rule, before a block, before an audit, somebody has to see what is actually there. We build that first, and we build it from behavior, from what runs and what talks, because inventories made from memory are wrong the day they are written.

Then we attach evidence to every finding. Not a score, the actual process, connection or file. Operators make good decisions when they can see the facts, and bad ones when they have to trust a number.

The team

Researchers and engineers who have built the systems they now secure.

Based in Zurich and Berkeley. Backgrounds in AI security research, large-scale systems engineering and the operators' side of adopting new technology.

AI security

We study how AI systems get attacked and how agents fail, and we have built the infrastructure that compound AI systems run on. We know agents from the inside.

Research at UC Berkeley and ETH Zurich
Systems at scale

We have shipped systems software and developer tooling in environments where code runs at scale and a mistake costs real money. That is the standard we build to.

The operator's side

We have helped large organizations decide how to adopt new technology, and learned what their leadership needs to see before it says yes.

Foundations

Computer science with a focus on systems and security, and the habit of teaching it, which forces you to explain things plainly.

ETH Zurich · Princeton · German Academic Scholarship Foundation · Eduard Rhein Foundation

We run agents in research and in production. We have watched them do the wrong thing with the right credentials. That experience is what House of Seal is built from.

How we work

Four commitments to the people who use our product.

Evidence, always

Every finding carries what we saw. If we cannot show it, we do not raise it.

Plain language

A finding says what happened, why it matters and what to do next. Reading it should not require training.

Your decision

We show and we recommend. Approving, containing or removing an agent is a decision your team makes, and it is recorded.

Direct access

You talk to the engineers who built the product. Questions get answers from people who can change the code.

Talk to us

Bring a question about your environment.

We are happy to look at a specific situation before any demo. Write to us or schedule a call.