How TOPAS for Word and HEX 165 govern AI across UK policing, justice, government and regulated industries.
The following use cases are illustrative scenarios showing how TOPAS for Word, HEX 165 and the Sovereign Edge IDN (in progress) could be deployed. They do not represent existing contracts, endorsements, or client relationships.
Staff use TOPAS for Word to check a draft force procedure against approved reference documents. HEX 165 sends only what policy allows to an approved model, and the reviewer approves each suggested change before it enters the document.
A disclosure officer drafts a response to a freedom of information request with AI help in Word. Personal details in the source material are redacted before AI processing, a second named person approves the final wording, and the audit log shows what the AI received.
An officer summarises a case file while names, contact details and account numbers are tokenised or redacted before AI processing. Revealing a redacted value needs an approval and the access expires. Use on live case material falls under DPA 2018 Part 3 and follows the force's own DPIA.
Corporate teams draft and summarise papers for force committees and boards. HEX 165 applies the force's roles and labels, sends requests only to approved models within a set budget, and reports the cost of each task.
Real-time fraud detection inference on taxpayer submissions at the edge, designed to reduce response times from hours to milliseconds. HEX 165 governs Self Assessment, Corporation Tax, and VAT return processing with full audit logs for NAO scrutiny.
Edge inference for biometric matching, border systems, and immigration case processing at ports of entry. HEX 165 provides policy-driven case management with NPPV3 cleared personnel and sovereign data handling throughout.
Clinical decision support inference at hospital trust level with sub-10 ms latency for time-critical applications. Patient data sovereignty maintained throughout. Full GDPR compliance with governed approval workflows for clinical AI outputs.
The Compliance Engine scans the agentic AI estate against the EU AI Act, DORA, GDPR and NIST AI RMF in a single pass. 648 criteria, 15 agentic risk flags, and every finding traced to the article. The Governance Engine adds TokenOps cost control for FinOps and CFO reporting.
Real-time inference for benefits eligibility assessment, fraud detection, and citizen service optimisation. Governed workflows for Universal Credit processing with complete decision audit logs for tribunal readiness.
Tactical edge inference for logistics, predictive maintenance, and non-classified operational analytics. Deterministic, auditable AI governance for defence procurement and contract management workflows at OFFICIAL tier.
* Latency figures are design targets for the Sovereign Edge IDN, which is in progress.