MICKAI®
Mickai Bespoke Systems

Your own system, built on an operating system you own.

The studios cover work most organisations share. Some organisations need something that does not exist yet. In a bespoke engagement you take the Mickai sovereign operating system, your own private brains and the sealed audit record, and we build the applications on top to your requirements. It runs offline on hardware you own, and what we build for you is yours.

See how a build runs →
The gap

The work that makes you competitive is never in a catalogue.

A catalogue answers the problems organisations have in common. Payroll, email, documents, meetings, customer records. The work that actually decides whether an organisation is competitive is rarely in a catalogue, because it is specific to that organisation: the asset it monitors, the process only it runs, the judgement its people make a thousand times a week. The usual options are poor. Buy a cloud product and the workflow bends to fit the vendor while the data leaves the building. Commission a systems integrator and you get a bespoke application with no substrate underneath it, so the model, the governance and the audit trail belong to somebody else again. Build it in house and you are hiring a machine learning team before you have written a line of the thing you actually wanted. What is missing is the middle: a finished sovereign platform to build on, so the bespoke part is your application and not your infrastructure.

What you build on

The hard part is already built.

A bespoke system is only affordable if the substrate underneath it already exists. This is what your engagement starts from, on day one, before a line of your application is written.

01 / The substrate

A finished operating system, not a framework

The sovereign operating system, the studio runtime, the model orchestration and the desktop shell are already built and running. Your engagement starts from a working platform, which is why a bespoke build here is a fraction of the work it is anywhere else.

02 / The brains

Your data becomes your own model estate

We build the organisation a private brain on its own records and run inference against it on your hardware. The knowledge, the models and the questions asked of them never leave the building.

03 / The record

Every action sealed and checkable

Whatever we build for you inherits the Open Audit Record. Consequential actions are hash chained and signed, and a checkpoint held off box means alteration can be detected afterwards by someone who does not have to trust us.

04 / The boundary

Offline by construction

The system runs with the network unplugged and can be verified in that state. Nothing calls home, entitlement is bound to your hardware, and there is no vendor inside the trust boundary once it is handed over.

05 / The controls

Staged actions and permissioning

Actions that leave a record are staged for human clearance by default. Permissioning and sensitivity tiering are native to the platform, so a bespoke application inherits governance rather than having it retro fitted.

06 / The protection

Built on a filed patent estate

The substrate your system runs on sits behind 104 filed UK patent applications carrying 2,340 claims. You are building on protected engineering rather than on a stack anyone can assemble.

How a build runs

Five stages, specified before they are built.

A staged engagement with the success measures written down at the start, so the finished system is judged against something agreed rather than against a demonstration.

01

Scope

We sit with the people who do the work and map the process as it actually runs, not as the manual describes it. The output is a written specification with the success measures agreed before anything is built, so the result can be judged rather than admired.

02

Ground

The system is stood up on your hardware and your data becomes a private brain: your records, your documents, your telemetry, indexed and permissioned. Nothing is uploaded anywhere. If the estate needs work first, that is the AI Readiness engagement and we say so at scoping.

03

Build

We build your applications against the platform. Because the operating system, the inference runtime, the permissioning and the audit record already exist, the engineering effort goes into your problem rather than into rebuilding a substrate.

04

Prove

The system runs supervised alongside the existing process, with a person clearing every consequential action, and is measured against the criteria agreed at scope. It goes live when the evidence says it should, not when the plan says it should.

05

Hand over

You get the running system, the documentation and the operating knowledge to run it. Support and further development are available and are a choice, not a dependency built into the architecture.

What we can build

The shapes this fits.

These are patterns of work rather than a product list. If your problem looks like one of them, it is a good fit for a bespoke build on the sovereign platform.

Monitoring

Equipment and telemetry

Continuous monitoring of an estate of instruments or machinery for abnormal behaviour, where the readings are sensitive or contractual and cannot be streamed to a third party. The system watches, flags the exception and shows the evidence behind the flag.

Review

Documents at volume

Reading, comparing and checking a large body of contracts, certificates, submissions or case files against rules specific to the organisation, producing an evidenced answer rather than a summary.

Casework

Process and case management

A workflow the organisation runs at scale that no product models properly, where the value sits in the judgement applied at each step and the requirement is that the judgement is recorded and reviewable.

Intake

Triage and routing

Sorting a high volume of inbound material, correspondence, claims, applications or reports, deciding what matters and routing it with the reasoning attached, so the decision can be examined later.

Assurance

Continuous compliance

Checking an operation against its own control framework as it runs rather than at audit, and assembling the evidence pack as a by product of the work instead of as a project.

Interface

Voice and hands free operation

Environments where staff cannot be at a keyboard, so the system is driven and interrogated by voice, on the same sovereign substrate, with the same record of what was asked and done.

What you end up with

A system you own, not a subscription you renew.

  • A working system built to your specification, running on your own hardware, that does the thing you actually asked for.
  • A private brain built on your own records, so the system understands your organisation rather than the internet's average of it.
  • A tamper evident audit record covering every consequential action, verifiable offline by someone who does not have to trust Mickai.
  • Ownership of the system built for you, with the documentation and operating knowledge to run it.
  • No data leaving the building at any point, including during the build itself.
  • A platform underneath that keeps developing, so your system inherits improvements to the substrate without being rebuilt.
Said plainly

What we are

A software company with a finished sovereign platform. A bespoke engagement is engineering work on top of it, scoped, quoted and delivered as engineering work.

Where a catalogue is better

If a studio already does what you need, take the studio. It is faster and it costs less, and we will say so at the scoping call rather than after it.

Regulated and safety critical use

Where an intended use falls under a device, clinical or safety regime, that classification and its approvals sit with the operator and are not something Mickai holds today. We build the system and evidence its behaviour; we do not claim clearance we do not have.

Certifications

Cyber Essentials Plus, ISO 27001 and SOC 2 Type II are targets this round is raised to pursue. None is held today. The security position rests on published architecture and a checkable audit record.

Questions

What people ask before a scoping call.

Can Mickai build a system specific to our organisation rather than selling us studios?

Yes. That is a bespoke engagement. You take the Mickai sovereign operating system, a private brain built on your own data and the sealed audit record, and we build the applications on top to your specification. The studios remain available but you are not obliged to take any of them.

Do we own what you build?

You own the system built for you and the private brain built on your data. The underlying sovereign operating system and its patented substrate remain Mickai intellectual property and are licensed to you. That split is written into the engagement rather than left to interpretation.

Does our data ever leave our building?

No. The platform runs entirely on hardware you own, offline. Nothing is uploaded for training, inference or support. The system can be verified running with the network physically disconnected, and we invite that test.

How is this different from hiring a systems integrator?

An integrator builds you an application and then has to source a model, a runtime, governance and an audit trail from elsewhere, usually a cloud. Mickai has already built all of that, so the bespoke work is your application on a finished sovereign platform rather than a bespoke stack assembled from other people's services.

What if our data is not ready?

Then we say so at scoping. Organising, digitising, cleaning and classifying a scattered estate is the AI Readiness engagement, and it runs before a bespoke build rather than being quietly absorbed into one.

Can it run somewhere with no internet at all?

Yes. Offline operation is the design centre rather than a mode. The system runs air gapped on your own hardware, and the audit record can be verified in that state without contacting Mickai.

What sectors does this suit?

Regulated and sensitive operations where the data cannot go to a third party: healthcare and life sciences, financial services, legal, defence and its supply chain, critical infrastructure, government and manufacturing. The common factor is a specific process and a hard constraint on where the data may sit.

How do we start?

A scoping call, then a written specification with success measures agreed before any build begins. Pricing is discussed at briefing rather than published, because a bespoke scope has no list price.

Tell us what it has to do.

A scoping call is a conversation about the process, not a demonstration. If a studio already covers it we will tell you. If it does not, we will tell you what building it looks like.

Not sure your data is ready? →