Six agents decide whether another agent's payment should settle, before a cent moves. Every decision is signed, reasoned, and replayable.
Revvyn sits in front of every autonomous-agent payment and returns one verdict, allow, block, or escalate, from a pipeline of agents that check identity, weigh policy and risk, and seal an immutable record. You can watch each decision happen, and verify it yourself.
A payment intent comes in one side; a governed verdict comes out the other. But the six aren't stages in a script, they're agents, each working its own domain, and one agent's finding changes what the next one decides.
This is the part most systems don't have: the agents aren't a checklist, they influence each other, and the influence is recorded on the chain, so you can trace exactly which signal changed which decision.
Every payment intent gets one verdict before settlement, and the reasoning behind it is captured in full, so a regulator, a partner, or you can ask why and get a real answer.
Agent verified, policy satisfied, risk within bounds. The payment proceeds to its rail.
A sanctions hit, an exposure ceiling, or a policy breach, the reason is logged, and no money leaves.
Borderline trust, or a dual-approval threshold. Settlement pauses; a human approves, rejects, or amends, and the step-up loop is recorded end to end.
Every decision on your payments, visible and verifiable, running on real engine data. Numbers are early because the network is early; every figure is the genuine current value.
Your payments governed, allowed, blocked, and escalated, with your live decision feed.
Re-open any of your past decisions and step through the signals and reasoning that produced it.
See your payment move through the six agents, agent by agent, in real time.
Integrate directly: submit a payment, get one verdict, pull the signed audit trail.
A governance layer earns trust with deterministic, explainable decisions first. So the capabilities that make Revvyn more than a decision engine are built and tested, and deliberately not live yet.
Agents resolving competing terms before a payment proceeds.
When two signals disagree, a real mechanism reconciles them.
A decision about an entity today, recalled when that entity transacts tomorrow.
The escalation contract is live and proven against a reference issuer. Plugging in a real bank's step-up is an integration, not a build.
We turned our most advanced capabilities off for launch on purpose, because determinism and auditability come before autonomy when the thing being governed is money. It's in the codebase. You can review it.
Revvyn is pre-production: the governance engine is real and usable today. Production rail integrations and formal certifications are in progress, not complete. We won't claim otherwise.
No slides, no recording. Fire a real payment through Revvyn's production pipeline right now and watch each agent decide, in order, with a signed and chain-verified trail.
Revvyn is in invite-only private preview. The site and this live proof are open to view; access is granted by request.
Payment mandates are cryptographically signed. Sign one yourself with a key that is not Revvyn's, claim to be Revvyn's issuer, and submit it. Revvyn verifies the signature against Revvyn's real published keys, catches the forgery in-line, and escalates it for a human to review. It is never silently approved.
Where this goes: every agentic transaction, governed, with a decision that's explainable, recorded, and yours to verify.
Today that's payments through a six-agent pipeline. The direction is the standard the rest of the stack can assume.