About

We make sure agents do the right thing.

kiss is the business-logic control plane for AI agents. Agents request actions; we verify each one against your own systems of record, then execute only when the values match.

Why we exist

The industry has secured agent access: least privilege, short-lived credentials, policy checks. But an authorized agent can still be tricked into an in-scope action with the wrong values, a refund for the wrong order, a deploy to production, a payment to the wrong account. Policy passes. The values are wrong. The damage is done.

What we do

Agents hold no credentials. They request an action, and kiss checks it against the business record that justifies it, the order, the ticket, the PR, the vendor master. Match, we execute through managed connectors. Mismatch, we block or escalate to a human. Credentials never leave our infrastructure.

Deterministic, not guesswork

We compare an action's values to the record that justifies it. Equality and membership checks, not another model second-guessing the first.

Honest about scope

We catch the consequential action, not every attack. Where there's no trusted record to check against, we escalate to a human instead of silently allowing.

Depth over breadth

Accuracy comes from integration depth and tuning, so we can be trusted to block money, data, and infrastructure, not just alert on them.