Underwriting and Actuarial
Mickai runs the entire underwriting and actuarial function in-house, on hardware you own, fully offline. Risk scoring, loss curve modelling, pricing, claims triage and portfolio aggregation all execute behind your firewall, with every decision sealed to a post-quantum Open Audit Record. The applicant data, claims files and reserving assumptions that drive your book never leave the building.
Underwriting and actuarial work concentrates the most sensitive data an insurer holds: applicant and policyholder PII, medical and lifestyle disclosures, claims histories, reserving assumptions, loss triangles and the pricing logic that is itself a competitive secret. None of this can lawfully or safely sit in a shared multi-tenant cloud where it crosses tenant boundaries and jurisdictions. UK GDPR and the special category provisions govern health and biometric inputs, the FCA expects demonstrable control over pricing fairness and outcomes under Consumer Duty, the EU AI Act classifies risk assessment and pricing in life and health insurance as high-risk and demands documented governance, and DORA holds the firm accountable for the operational resilience of every system that touches the book including third-party AI. SR 11-7 model risk governance requires that every model be inventoried, validated and explainable on demand. Mickai keeps the models, the data and the decision record on hardware the firm owns, so the regulated boundary is enforced by architecture rather than by a vendor contract.
The 12 services in this pillar, each running on hardware you own and routing to the studio that delivers it. Every one names the regulated data it keeps in-house and the category of cloud tool it replaces.
Risk Scoring and Underwriting Decisioning
The offline brains read applicant submissions, medical disclosures, exposure data and prior loss history to produce a risk score and an approve, refer or decline recommendation entirely on hardware you own. Applicant PII and special category health data never traverse a network or reach a third party, and every decision is written to the Open Audit Record with the features that drove it. Underwriting authority stays inside the firm, sealed and reproducible on demand.
Actuarial Loss Curve Modelling
Mickai computes development triangles, IBNR reserves, exposure curves and liability run-off locally against your own claims and reserving data. The loss triangles, reserving assumptions and the actuarial judgement embedded in them stay air-gapped, so neither the data nor the proprietary curve logic is exposed to a shared environment. Reserving work remains fully under firm control and every figure is traceable to its inputs.
Quote-to-Bind Automation
The sovereign brains assemble quotes, apply underwriting rules and prepare binding documentation for complex commercial and corporate risks without any submission leaving the building. Broker submissions, schedules of values and the pricing terms attached to each quote are processed and stored on owned hardware, with the full quote-to-bind chain sealed to the audit record. The firm gains speed on placement while keeping the entire transaction sovereign.
Risk-Based Asset Pricing
Mickai computes premium rates and technical pricing from rating factors, exposure variables and loss cost models parsed locally, never in a vendor cloud. The rating tables and the pricing algorithm itself, both highly competitive assets, stay on hardware you own, and each rate is logged with the variables and weights that produced it for FCA fairness review. Pricing intellectual property and the data behind it remain fully sovereign.
FNOL Automation and Triage
As a First Notice of Loss arrives, the offline brains validate the claim against policy data, classify it and route it to the correct handler, all in real time on-premise. Claimant PII, incident details and any medical or accident reports are processed inside the firewall and never copied to an external service. Triage is faster and the claimant record stays sealed and sovereign from the moment it lands.
Claims Classification and Adjudication Support
Mickai categorises incoming claims by type and severity and cross-references coverage terms, deductibles and policy limits to support the adjudicator. Claim files, supporting documents and policy wordings are read on owned hardware so no claimant or medical data is exposed to a third party. Every adjudication recommendation is sealed to the audit record, keeping the decision defensible and the data in-house.
Claims Fraud Detection and Anomaly Scoring
The brains screen claims for fraud indicators, network links and statistical anomalies completely air-gapped, including AML and sanctions checks where required. Claimant identities, claim patterns and the fraud rules themselves stay inside the firewall, so investigative logic is never leaked and no PII crosses a tenant boundary. Suspicious claims are flagged with a sealed evidence trail that holds up under FCA and regulatory scrutiny.
Claims Severity and Payout Prediction
Mickai projects ultimate settlement values and likely legal payouts by modelling current claims against your historical loss and litigation data on-premise. The claims histories, injury and medical records and settlement outcomes used to train and run the model stay on hardware you own and never reach an external provider. Reserving and case management gain accurate forecasts while the underlying data remains sovereign.
Fairness, Bias Detection and Explainability Trails
Mickai tests underwriting and pricing models for disparate impact across protected characteristics and produces full explainability trails for each decision, sealed to the Open Audit Record. The protected attribute data and model internals required for this testing stay air-gapped, satisfying FCA Consumer Duty, EU AI Act high-risk obligations and SR 11-7 model governance without exposing anything to a vendor. Fairness evidence is demonstrable to a regulator on demand and never leaves the firm.
Policy Lifecycle Risk Recalibration
As endorsements, renewals and fresh claims data register, the offline brains recalibrate the risk model on each active policy without an external feed. The in-force policy data, updated exposure information and emerging loss signals are processed locally and every recalibration is timestamped to the audit record. The book stays accurately priced through the lifecycle while all policyholder data remains sovereign.
Portfolio Risk Aggregation
Mickai totals exposures across lines of business, geographies and perils to compute accumulation, concentration and stress test limits entirely in-house. The full book of exposure data, treaty terms and catastrophe scenarios stays on owned hardware, so capital and solvency positions are modelled without exposing the portfolio to a shared cloud. Aggregation results are sealed and reproducible for regulatory capital reporting.
Reinsurance Optimisation Inputs
The brains model retention levels, layer structures and treaty parameters to recommend how excess risk should be ceded, working only from your own loss and exposure data. The reserving assumptions, ceded loss histories and commercial reinsurance terms involved stay behind the firewall, so the firm's risk transfer strategy is never visible to an outside platform. The cession recommendations are auditable and the underlying book remains fully sovereign.
Each service runs in a purpose-built studio. See the whole application surface on all studios, and the eighteen enterprise studios on the sovereign services catalogue.
The advantages hold across every department, and they are architectural, not promotional. The third-party cloud-exposure vector is removed; your own physical, insider and compliance controls remain yours.
The data never leaves your hardware, so no third party and no cloud-provider employee ever sees it. What happens in the server room stays in the server room.
You own the compute and the capability, so the system runs independent of the internet and of any cloud vendor's pricing, terms, or availability.
The data never crosses a geographical or digital border because it never leaves the building, which removes the cross-border-transfer and third-party-processing friction of UK GDPR, Schrems II, and the sector rules. You keep your own obligations.
Fine-tune and run retrieval on your deepest archives to build a hyper-customised co-pilot, with no risk of your proprietary edge training a public model or leaking.
After the hardware and licence, queries cost essentially electricity. A capital asset you own and depreciate, instead of volatile per-token cloud bills.
There is no third-party cloud path, so no competitor and no vendor insider can scrape, intercept, or subpoena your prompts or your fine-tuned weights from the internet. The trust vault is closed by architecture.
You own the software snapshot on your own hardware, so a change to a cloud vendor's terms, a model deprecation, or an outage cannot reach you. The system stays predictable and auditable on-premise as the rules evolve.
Can underwriting and actuarial AI run fully on-premise and air-gapped?
Yes. Mickai runs the entire underwriting and actuarial function, including risk scoring, loss curve modelling, pricing, claims triage and portfolio aggregation, on hardware the firm owns, fully offline. No applicant PII, claims data or reserving assumption ever leaves the building, and every decision is sealed to a post-quantum Open Audit Record.
How does Mickai satisfy FCA, EU AI Act and SR 11-7 requirements for pricing and risk models?
Every model decision is written to a sealed, tamper-evident audit record with the features and weights that produced it, so the firm can demonstrate fairness under FCA Consumer Duty, governance under SR 11-7 and the documentation required for EU AI Act high-risk insurance systems. Dedicated fairness, bias detection and explainability trails run on-premise so protected attribute data is tested without ever being exposed. Because the models and data stay on owned hardware, the firm retains full accountability under DORA rather than depending on a third-party vendor.
Where do applicant and claimant data sit when Mickai runs underwriting?
All applicant PII, special category health disclosures, claims files and loss histories remain on hardware the firm owns, behind its own firewall, with zero data egress. Nothing is copied to a multi-tenant cloud or shared with an external provider at any stage of underwriting, pricing or claims handling. This keeps the data inside the regulated boundary that UK GDPR and insurance regulators require.
Is Mickai a cloud service or SaaS for insurers?
No. Mickai is a Sovereign Intelligence Operating System, a complete AI operating system that runs entirely on hardware the customer owns, on-premise and offline. It is acquired as an owned asset rather than a metered subscription, with no per-seat or per-token charging and no dependence on an external cloud.
What models power the underwriting work, and is the pricing logic exposed?
Mickai runs its own sovereign offline brains, 50 in total across 25 domain and 25 operational roles, entirely within the firm's environment. Proprietary rating tables, reserving assumptions and pricing algorithms are processed and stored locally, so the firm's pricing intellectual property is never visible to any outside platform. The competitive logic that drives the book stays sovereign by design.
Does keeping underwriting in-house mean losing the cloud entirely?
No. The public cloud remains valuable for non-regulated workloads, and Mickai is the answer specifically for the regulated-data boundary where applicant PII, claims and pricing logic live. The distinction is architectural: regulated risk and actuarial data runs sovereign on owned hardware while other work can stay wherever the firm chooses.
Bring the underwriting and actuarial pack in-house.
Briefings are for organisations weighing a sovereign, on-premise deployment. Tell us about your estate and we will walk the services, the regulatory crosswalk and the deployment that fits.