Sovereign AI for Energy and Water: Forecasting and Control Without Data Egress
How critical-infrastructure operators run forecasting and anomaly detection on air-gapped networks, meeting the NIS Regulations without sending a single byte to a public-cloud model.
By Micky Irons, founder and CEO of Mickai
The problem no public-cloud demo solves
An energy or water operator does not have a data problem. It has a data egress problem. Grid telemetry, SCADA logs, demand curves, treatment-plant sensor streams and outage records are some of the most operationally sensitive data a nation holds. The Network and Information Systems Regulations treat the firms that run this infrastructure as Operators of Essential Services, with duties to manage security risk, protect networks, detect incidents and report them. None of that sits comfortably with piping live operational data to a model endpoint you do not own, in a region you do not control, under a foreign legal regime.
So the typical AI pitch fails at the first meeting. The forecasting demo looks impressive. Then the security team asks where the data goes, and the answer ends the conversation. The operator is left choosing between modern AI and regulatory defensibility, and it rightly chooses defensibility.
Mickai removes the choice. It is a sovereign AI operating system, a SIOS: AI that a regulated business owns and runs inside its own walls, on-prem and air-gapped, with every action written to a tamper-evident, post-quantum-signed audit record we call the OAR. This is built and live, not a concept. The model runs where the data already is. Nothing leaves the boundary.
Prometheus: forecasting that never leaves the boundary
Prometheus is the forecasting Studio. For an energy operator that means day-ahead and intraday demand forecasting, renewable-generation prediction tied to local weather feeds, load balancing, maintenance-window planning and reserve sizing. For a water operator it means demand and flow forecasting, pressure modelling, leakage prediction, pump-energy optimisation and reservoir-level planning against rainfall data.
The point that matters for a critical-infrastructure buyer is not a headline accuracy figure. It is that the entire forecasting loop runs inside the operational technology boundary. Historical load data, SCADA history and live sensor feeds are read in place. The model trains and infers on the operator's own hardware. Forecasts come out. No telemetry transits a public network, no data is held by a third party, and there is no foreign-jurisdiction exposure under instruments like the US CLOUD Act. An air-gapped deployment is a first-class mode for Mickai, not a downgraded one.
Pythia: anomaly detection and operational intelligence
Pythia is the business-intelligence and analytics Studio, and on a critical-infrastructure network its most valuable job is detecting the abnormal. Sensor drift, an unexpected pressure transient, a load pattern that does not match weather or season, a substation reading that breaks from its neighbours, a flow signature that suggests a developing leak or an intrusion attempt against the control network. Pythia surfaces these against learned baselines and lets operators interrogate them in plain language, on data that never leaves the plant.
This maps directly onto NIS duties. The Regulations require an operator to detect incidents and to understand the security state of its essential service. Anomaly detection that runs on-prem, that explains its reasoning, and that writes every query and finding to the OAR gives the operator both the detection capability and the evidence trail that a Competent Authority expects to see. When the regulator or auditor asks what the system saw and when, the answer is a signed, tamper-evident record, not a screenshot.
Why air-gapped is the architecture, not a setting
Most AI vendors treat on-prem as an enterprise tier bolted onto a cloud product. Mickai was built the other way around. Sovereignty is the substrate. The operating system, the Studio modules, the audit layer and the model runtime are designed to run with no outbound connectivity at all. That is the only honest architecture for an Operator of Essential Services, because a security control that depends on a working internet link is not a control you can rely on during an incident.
Every action across the SIOS is written to the OAR, post-quantum-signed so the record stays trustworthy even as cryptographic threats evolve. For a sector where an audit trail may need to hold its integrity for decades, that horizon matters. The operator owns the model, the data, the logs and the keys. There is no shared-responsibility ambiguity about who holds what.
This work sits on a deep IP position: 104 filed UK patent applications with roughly 2,340 claims, owned by Mickai LTD, covering the sovereign operating model, the audit architecture and the Studio approach. Filed rather than granted, these establish priority and a prior-art moat around running regulated AI inside the wall.
Where Mickai fits in your stack
Mickai is an ally to your existing systems, not a rip-and-replace. It sits alongside your SCADA, historian, GIS and asset-management platforms as the AI layer, reading the data they already hold and writing intelligence back into the workflows your engineers use. Prometheus and Pythia are two Studios in a wider set that includes Nomos for compliance, Astraea for legal, Aletheia for audit and Nemesis for fraud and anomaly work, so the same sovereign substrate extends across the regulated functions a utility runs.
As a third-party momentum signal, in June 2026 Crunchbase ranked me, Micky Irons, at number four on its CB Rank for people, with Mickai placing in the top one to two percent of companies globally. We are a UK company with Birmingham manufacturing secured, and we are building to scale. I share that rank as a dated June 2026 snapshot of where the market is looking, not as a permanent claim.
A window for selected partners
The buyers here are twofold. Energy and water operators who need forecasting and detection they can legally run. And the regulators, integrators and infrastructure owners who need sovereign AI to be demonstrably defensible. Mickai serves both.
A pre-seed window is open to selected partners as Mickai scales. This is an invitation to get involved early in sovereign AI for critical national infrastructure, not a sign of need. If you operate essential services, integrate for them, or want to back the sovereign-AI thesis at the substrate layer, I would like to talk.
Contact me directly at micky@mickai.co.uk.
Frequently asked questions
Does Mickai send any operational data to a public cloud?
No. Mickai is a sovereign AI operating system designed to run on-prem and air-gapped. For energy and water operators, the forecasting and anomaly-detection loops run entirely inside the operational technology boundary, on the operator's own hardware. No telemetry, SCADA history or sensor data transits a public network, which removes foreign-jurisdiction exposure under instruments like the US CLOUD Act.
How does Mickai help with the NIS Regulations?
The NIS Regulations require Operators of Essential Services to manage security risk, protect networks and detect and report incidents. Pythia provides on-prem anomaly detection against learned baselines, and every action across the system is written to the OAR, a tamper-evident, post-quantum-signed audit record. That gives operators both the detection capability and the signed evidence trail a Competent Authority expects to see.
Which Studios apply to energy and water operators?
Prometheus handles forecasting, including demand, generation, flow and reserve planning. Pythia handles business intelligence and anomaly detection. Both run alongside wider Studios such as Nomos for compliance, Astraea for legal, Aletheia for audit and Nemesis for fraud and anomaly work, all on the same sovereign substrate inside the operator's boundary.
Is Mickai built and available now?
Yes. Mickai is built and live, not a concept. It is a UK company with Birmingham manufacturing secured, building to scale. The capability described here, sovereign on-prem and air-gapped AI with a post-quantum-signed audit record, is the substrate the product is built on.
What is the patent position?
Mickai LTD holds 104 filed UK patent applications with roughly 2,340 claims, inventor Micky Irons, covering the sovereign operating model, the audit architecture and the Studio approach. These are filed rather than granted, which establishes priority and a prior-art moat around running regulated AI inside the wall.
How do I get involved as a partner or investor?
A pre-seed window is open to selected partners as Mickai scales. It is an invitation to get involved early in sovereign AI for critical national infrastructure. Operators, integrators and backers of the sovereign-AI thesis can contact Micky Irons directly at micky@mickai.co.uk.