The claims on this site are meant to be checked
Every vendor in this market says their platform is reliable. The useful question is what any of them will let you look at. This page is what we can show you, what we cannot show you yet, and how to ask.
Evidence that already exists
Each of these is a thing that happened, not a design intention.
- A release gate that has refused a real release. The life-safety assurance suite runs on every promotion, and a failure rolls the release back automatically. It is not advisory and it has already stopped a release that a person wanted to ship. A gate that has never said no is a gate nobody has tested.
- A completion figure that has been revised down. Our completion figure has been revised down twice, held once when we could have raised it, and revised up once. All four times because somebody actually checked. Ask to see the arithmetic and we will walk you through it.
- Database failover proven by drill. Not a design claim about redundancy. A drill, run, with the result recorded. A standby database is mandatory at every scale we deploy, because a single-instance database under a duress workload is a configuration we refuse to deploy.
- Backups that have been restored. Encrypted, continuous, and automatically restore-tested. A backup that has never been restored is a hope, not a backup.
- Normalisation proven across two independent vendor dialects. The adapter seam is not a diagram. Two different vendors' output, normalised to one canonical fix, is the test that tells you whether vendor-agnostic means anything.
- A tamper-evident audit chain. Hash-chained with signed checkpoints, so an alteration after the fact is detectable rather than a matter of trust.
What an evaluation actually gets you
These are the four things worth asking for, of us and of anybody else you are looking at.
Break it in front of you
Pull the network mid-escalation. Restart a server while a timer is running. Take a locator offline. The interesting demonstration is not the happy path, it is what the screens say when something is wrong.
The safety documentation
The hazard register and the clinical safety assessment, for your own governance people to read. Signing that assessment is a named human's job here, not a checkbox in a pipeline.
The arithmetic behind the figures
Any number we put in front of you comes with how it was counted and what it excludes. If a figure cannot survive that conversation, we would rather not quote it.
A proving deployment
One team, real work, running alongside whatever you have today rather than instead of it. That produces evidence rather than a demonstration.
What we will not do
Every capability in a pilot is demonstrated before you rely on it, and anything not yet in place is told to you as such, in writing, before go-live.
That is the sentence the rest of this site is written to. It is also the one worth holding us to.
We do not quote uptime percentages, accuracy figures or response times as marketing numbers. Where a measured figure exists it belongs in a document with its method attached, in front of somebody evaluating us, not on a web page where it has no context and invites more scrutiny than it earns. Security and assurance posture →
Ask us for the evidence
Name the claim you want tested and we will tell you what we can show, what we cannot show yet, and which it is.