By Micky Irons, founder of Mickai.
The ground the cloud cannot cross
Mickai is a sovereign intelligence operating system that regulated businesses own and run inside their own walls. That single sentence is also a boundary line. On one side sits the public cloud, where every model call is a data transfer to someone else's estate. On the other side sits the regulated enterprise, which by law, by regulator expectation, or by contract cannot make that transfer. We built Mickai to live entirely on the second side, and we built it to stay there.
We run on the customer's own hardware, on premises and air gapped, with zero data egress and no public cloud round trip. There is no telemetry leaving the building, no vendor endpoint quietly holding a copy, no shared tenancy to reason about in a risk committee. The intelligence comes to the data, and the data never moves. That is the whole design, and it is not a roadmap item. It is built and live today.
This piece sets out why that architecture is difficult to copy, why the audit trail matters more than the model, and why the market we serve is one the largest platforms in the world cannot lawfully reach. The through line is a moat made of prior art: 104 filed UK patent applications that stake priority across the exact mechanisms a sovereign AI system needs.
What we actually run
We run about fifty specialist models, twenty five domain and twenty five operational, with cross-model routing under a deterministic arbiter. The arbiter matters as much as the models. Because routing is deterministic, the same inputs produce the same path and the same outputs, which is the property a regulated buyer needs before an AI system is allowed anywhere near a decision that a regulator can later question. Reproducibility is not a nice-to-have in this market. It is the entry ticket.
On top of that substrate we run a set of studios. Each carries a Greek name and a serious function. Nemesis handles fraud and anti money laundering. Plutus covers finance and FP&A. Tyche runs underwriting. Prometheus does forecasting. Iris takes customer service. Nomos owns compliance, Astraea owns legal, Panacea covers clinical work, Pythia is business intelligence, and Aletheia runs audit. Vinis handles voice. The Agentic Marketing Team runs the outward-facing marketing function. Trust Agent is the perimeter. And OAR-as-a-Service exposes the audit ledger as a standalone capability for teams that want the record even before they adopt the rest.
None of these are wrappers on a distant API. They are our own sovereign models, running inside the customer estate, arbitrated locally, producing outputs the customer can defend line by line.
The record is the product
In regulated work, the answer is only half the job. The other half is proving, later and to a hostile reader, that the answer was produced correctly and has not been altered since. This is where we put our deepest engineering, in the Open Audit Record.
Every consequential action is signed under post-quantum cryptography, using FIPS 204 ML-DSA-65 with ML-KEM-768, and hash-chained into a tamper-evident, append-only ledger. Anyone can verify that ledger offline, for decades, without trusting us. Read that last clause again, because it inverts the usual vendor relationship. The buyer does not have to take our word for anything. They hold a cryptographic record they can check on their own hardware, long after any commercial relationship has ended, and long after the cryptographic assumptions of today have been stress-tested by tomorrow's machines.
That is why we describe the record, not the model, as the product. Models will improve and be replaced. The obligation to prove what happened, and to prove it survives quantum-era scrutiny, does not go away. We built for the obligation.
Beyond a single site, Pantheon extends the same principle across many. It is a post-quantum Layer 1, currently on testnet, that gives multi-node attestation across fielded units with no central server. Each unit attests to the others. There is no coordinating authority to compromise, and no single point whose failure quietly rewrites the shared history.
Why this is hard to copy: the prior-art moat
A determined competitor can assemble on-premises inference. What they cannot easily assemble is the specific combination of sovereign routing, deterministic arbitration, offline-verifiable post-quantum audit, and multi-node attestation, without walking into ground we have already staked.
We hold 104 filed UK patent applications, comprising roughly 2,340 claims across 13 invention families, owned by Mickai LTD, with named inventor Mickarle Sean Junior Wagstaff-Irons. These are filed, not granted. We are precise about that, because filing is what matters for the moat. Filing establishes priority. It sets the date against which everyone else's later work is measured, and it converts our architecture into prior art that a competitor has to design around rather than reimplement. The claims track the mechanisms above, which is the point. The patents are not a wall around a logo. They are a wall around the way sovereign intelligence is made verifiable.
Mickai LTD is a UK company, Companies House number 17166618, with Birmingham manufacturing secured. Micky Irons is founder and CEO. The estate sits inside a real operating company with a real production path, not a holding vehicle.
A market the cloud cannot lawfully serve
The sovereign AI market is roughly USD 40 billion in 2025, rising to about USD 148 billion by 2032. Those numbers are large, but the more important number is who is locked out of the public-cloud version of it. Around 0.85 million UK businesses, about 15 percent of the total, and roughly 5 million across the EU, legally cannot send their data to public cloud AI. That is not a preference. It is a legal boundary.
The drivers are named and specific. PRA model-risk expectations under SS1/23. UK GDPR special category data. The NHS Data Security and Protection Toolkit. The EU AI Act high-risk classification. ITAR and EAR export controls. The NIS Regulations. The US CLOUD Act, which is precisely why a US-headquartered cloud cannot credibly promise a European regulator that data will never be reachable under foreign legal process. Each of these turns a general preference for privacy into a hard constraint that a public-cloud deployment cannot satisfy. We are built to sit exactly inside that constraint.
This is the wedge. We do not have to win a general contest for AI mindshare. We have to serve the customers the incumbent architecture cannot lawfully touch, and we are the architecture that touches them.
The dual-buyer thesis
Our strategy has two motions. First, we sell sovereign AI directly to regulated firms that the public cloud cannot lawfully reach. Second, we license the patented stack to the platforms that want to reach those firms and currently cannot.
Our internal analysis maps 196 companies and 311 patent-company pairs as potential licensees, including names such as Microsoft, AWS, NVIDIA, Google, Adobe and IBM. We are careful about what that is and is not. It is potential-licensee sizing, a structured view of where our mechanisms intersect with others' product surfaces. It is not a signed book of business, and it is not an infringement claim against anyone. We publish it as a map of opportunity, not as a threat.
The framing matters. We are an ally to the AI majors, not an OpenAI killer. A platform that adds a sovereign, verifiable layer instantly gains the regulated market it cannot serve today, and it gains it with priority already established rather than contested. That is a constructive position for everyone in the picture, which is exactly why we hold it.
Where this leaves us
We have built a system that keeps data inside the customer's walls, produces reproducible outputs from about fifty arbitrated specialist models, and signs every consequential action into a post-quantum record that anyone can verify offline for decades. We have filed 104 patent applications across 13 families to stake that ground. And we point all of it at a market defined by law rather than by taste, a market the largest incumbents cannot lawfully enter on their current terms.
The moat, in the end, is not any single feature. It is the compound: a live product, a verifiable record, a filed patent estate, and a customer base the cloud cannot reach. Each part makes the others harder to copy. Our pre-seed round is opening soon, and we welcome inquiries from interested partners by email at micky@mickai.co.uk or on LinkedIn.
Are the patents granted or filed?
They are filed, not granted. We hold 104 filed UK patent applications, roughly 2,340 claims across 13 invention families, owned by Mickai LTD. Filing establishes priority and creates a prior-art position that competitors must design around. We do not describe them as granted, because that would be inaccurate and unnecessary. Priority is what does the work.
Does Mickai run in the public cloud?
No. We run entirely on the customer's own hardware, on premises and air gapped, with zero data egress and no public cloud round trip. That is the point of the design. It is what allows regulated firms, who legally cannot send data to public cloud AI, to use advanced intelligence without breaching the constraints they operate under.
How can a buyer trust the audit record without trusting the vendor?
The Open Audit Record is hash-chained and signed under post-quantum cryptography (FIPS 204 ML-DSA-65 with ML-KEM-768) into an append-only ledger. Verification is offline and independent of us. A buyer checks the record on their own hardware, with no call back to any Mickai endpoint, and that property holds for decades. Trust in the vendor is designed out of the verification path entirely.
Written by Micky Irons, founder of Mickai. Originally published at https://mickai.co.uk/articles/prior-art-moat. More from Mickai at mickai.co.uk.
Top comments (0)