The governed AI platform

Governed AI.
One platform.
Every domain.

AI projects keep rebuilding governance from scratch. LumiWright builds it once, as infrastructure: policy, human approval, provenance and proof. Domain packs carry the specialisation, so organisations deploy AI they can trust, faster.

Health first. Finance beside it. Then Education. Three domain packs, one governed gate, a core that never changed. Every decision writes signed evidence.

The problem

Every AI application rebuilds the same trust machinery.

That doesn't scale. Governance ends up duplicated, inconsistent and hard to prove, and every deployment waits longer for sign-off.

Duplicated workthe same approvals, policies and evidence, rebuilt for every project
Inconsistent controlseach application governs itself a little differently
Fragile proofevidence scattered across logs, tickets and screenshots
Slow sign-offevery new workflow starts its governance design again

The insight

AI should evolve the way operating systems did.

Don't rebuild governance inside every AI application. Build it once, and let every application inherit it. Three ideas from thirty years of infrastructure show why that works.

These three analogies explain how we think about our solution architecture. They describe our design approach, not technologies you need in order to use the platform.

Linux

One core, many roles.

Linux can become a file server, a firewall or a web server, without rewriting Linux.

Our core knows nothing about any application or industry. Because it knows nothing about them, it never has to be rewritten for them.

The layered network model

Each layer does one job.

Things plug in. A browser can specialise in doing one thing brilliantly, and nobody says the system is all over the place, because a firewall sits right beside it.

Focused applications on one platform, without it becoming a mess.

Microsoft Word

A platform, not an app.

A bank uses a word processor. A hospital uses a word processor. It isn't a banking app or a hospital app. It's a document platform.

A general capability that each industry uses in its own way. Ours is a governed AI platform.

What you gain

Accelerated deployment, with sign-off.

Start further aheadEvery new workflow inherits one authority path instead of a new governance project.
Change the rule, not the appChange a domain's policy and the same request is blocked or escalated. The application didn't change. The governance did.
Add domains without touching the coreHealth came first. Finance was added and the core didn't change. Education followed the same way.
Proof by designEvidence is produced as the work happens, and one record answers the regulator, the auditor and the person asking why.

How it works

One shared foundation. Specialised for your domain.

  1. Shared platformfactory, execution and governance runtimes
  2. Reusable capabilitiesintake, verification, approval, reporting, built once
  3. Domain packsthe rules, checks and approvers of your industry, declared
  4. Your configurationyour roles, providers, workflows and experience

The authority path and the six questions →

Proof

We didn't just design it. We measured it.

The same governed path ran against a real hardware security module and write-once storage for 24 continuous hours. We cut the network to the signing hardware mid-run, and the next decision was refused. We revoked write access, and the release was aborted before a single byte left. It fails closed.

854,100 governed decisions. 24 continuous hours. Zero failed batches. Real signing hardware in FIPS mode and real write-once storage.
854,100governed decisions in a continuous 24-hour run; 2,847 batches, zero failed
10regulated domains on one core; the last eight changed no existing file
2,031platform tests, including ten fault-injection drills
372 sto restore signing custody from an encrypted backup

From the paper Governed AI as Runtime Infrastructure (August 2026).

We don't compete on smarter AI.
We compete on trustworthy AI.

The story

Thirty years of infrastructure. The next chapter: governed AI.

From Novell and Microsoft through Cisco networking to cloud with AWS, every technology shift needed a safe, scalable foundation. How Ernest came to the solution →

Architecture session

Bring one consequential workflow.

Together we map its authority path, its evidence and its domain pack, and you leave with a one-page plan.

Book an architecture session

Not smarter AI.
Trustworthy action, with proof.