The Trust Engine issues certificates. It never has to be reachable from the plant. The Orchestrator sits at every site, collects requests, and brings certificates back — even through links that have been down for hours.
Certificate governance solves half the problem. The other half is physical: how does a signed certificate get to a PLC that has no route to the CA, on a link that isn't always up?
A central certificate authority the plant can never see, and an Orchestrator that lives where the devices are.
Whatever a device can do, there is a door for it — and the Trust Engine only ever sees one request shape.
RFC 7030 over HTTPS. Modern switches, routers, and firewalls already speak it — turn it on and they enrol themselves.
A browser page. An engineer pastes a CSR for a device that can do nothing else — the fallback for the oldest hardware.
A single HTTPS call — wired into a vendor tool, a build pipeline, or a commissioning script.
No database. Three folders, and a signed executable that moves requests between them.
The request has arrived and is waiting to be sent onward.
Sent to the Trust Engine, awaiting the signed certificate.
The certificate has arrived and is ready for the device to collect.
Pull the power out mid-transfer, and nothing is lost — the folder a request sits in is its state.
A controller can hold a valid certificate without anything at its level ever reaching the internet.
The Orchestrator writes it to the queue and answers immediately with a ticket.
Nothing fails. Nothing retries in a loop. The request simply waits on disk.
The Orchestrator notices within a minute and uploads on its own.
Metadata is checked against the asset inventory before anything is issued.
It asks the same question it asked at nine — and this time the answer is yes.
The Orchestrator sits at Level 3.5 in the DMZ — or lower, if the site prefers — with zero inbound connections into OT.
If the Orchestrator is stolen, cloned, or fully compromised, the attacker gains a queue — not a certificate authority.
The Trust Engine issues under either anchor. Which one you use is a configuration decision, not a product decision.
Total sovereignty. Nothing depends on anyone outside the organisation.
No trust store to push to thousands of devices — they already trust the root.
A pilot needs one site, one config file, and no changes to your existing PKI.