Identity, credentials and spend for AI agents
Agents get capabilities, not credentials.
HLOS lets an AI agent log in, call paid services and spend money on your behalf without ever holding your passwords, API keys or card numbers. It works inside a budget you set, and every action leaves a signed receipt.
01 · The problem
Handing an agent your keys is still the default.
Most agents get access the same way a person does: a password, an API key, a saved card. But agents keep screenshots, logs and traces, and one prompt injection can turn any of them into a leak. Once a key is in the agent's hands, everything the agent touches can reach it too.
An agent can't leak a secret it never had.
02 · What we're building
What we're building
- 01
Your keys stay out of the agent.
The agent asks for an action and gets back the result. The credentials that made it possible never enter its context.
- 02
Hard spending limits.
Set a budget before the agent starts. When it runs out, the agent stops.
- 03
A receipt for everything.
Every action is recorded with what was requested, what ran and what it cost, and signed so it can be checked later.
03 · Where we are
Where we are
HLOS is in private development and isn't open for sign-ups yet. We're building it for the people who have to sign off on it: security teams, identity engineers and the developers handing agents real access.
04 · Get involved
Giving agents real access? Let's talk.
We want to hear from you if you:
- are deciding what an agent may do with your credentials or your money
- are building agents that need to log in, call paid APIs or make purchases
- want to help build this