Governance decides. Someone has to enforce. KPATH exists because in most agent estates nobody is in a position to.
The gap
Identity platforms issue agent identities and decide what should be allowed. That is the right place for the decision. But a decision is not an enforcement. Something has to be sitting there at the moment the agent acts, able to say no, and most estates have nothing in that position. The spending, the sprawl and the unowned agents go unnoticed until someone audits them.
Three problems tend to arrive together. Token spend climbs with no per-agent attribution. Agents get built independently across teams with no shared inventory and no named owners. And an agent that is useful holds real access, so when one behaves badly there is no way to stop it mid-action and no record that survives an audit. They are not separate problems; they come from the same gap.
Your identity provider decides. KPATH enforces. KPATH is the Policy Enforcement Point that sits on the path between agents and services, so it can allow or refuse the call while it is happening, and log it.
Why security people built it
The founding team built and scaled HYDN Security, a cybersecurity firm whose clients include Rapid7, a16z, Consensys and MetaMask. The founders bring more than 20 years in cybersecurity and network security, more than 25 years in deep tech and transformation, and more than 15 years in security product and consulting.
That background shapes one design rule. Governance inside the agent, or inside the framework, puts the control inside the thing you are trying to control. A prompt injection captures the agent’s own code first, and a hijacked agent can decide anything it likes. KPATH sits outside the agent, on the path its calls have to travel, so the decision is not the agent’s to make.
The two lanes
KPATH earns its place when three conditions hold together: many services with a growing agent population, different teams on different frameworks, and agent actions that touch regulated data or money. Two lanes meet all three.
The first is regulated financial services, where a payments agent moving money needs identity, budget, human approval and a record that answers which agent did this and on whose authority.
The second lane is defence, where the constraint is fixed: KPATH enforces human control; it is not autonomous decisioning.
Neutral infrastructure
- Any framework. Agents make a small change to send their calls through KPATH and keep their framework, model and prompts. Nothing is rewritten.
- Any cloud or none. Self-hosted in AWS, Microsoft Azure or Google Cloud, on premises, air-gapped, or managed by KPATH.
- Your vendors, orchestrated. KPATH consumes the identities your identity provider already issues, including workload identities. It invokes the guardrails engine you already run, whether you bought it or wrote it. It sends evidence to the SIEM you already watch. KPATH replaces nothing, and adds the part the rest of your stack was never built to do.
Where KPATH stands today
A working, demonstrated platform with a live demonstration agent fleet. The evidence KPATH produces can be verified without trusting KPATH, which is the standard we would like to be held to.