Speed that stays compatible with institutional accountability.
DER's value depends on governance, so control is designed in: human approval, source-grounded AI, audit trails and release authority.
Nothing is published solely because an AI-generated draft exists.
Every release remains attributable to an authorised person and an authoritative source. AI may accelerate translation; it does not replace evidence validation, institutional review or release authority.
Five areas every deployment is designed around.
Control area
Minimum design requirement
Why it matters
Control area
Data protection
Minimum design requirement
Purpose limitation, minimisation, lawful basis, access control, retention and data-subject procedures.
Why it matters
Supports Kenya's Data Protection Act and analogous national regimes.
Control area
Statistical confidentiality
Minimum design requirement
Separate public outputs from confidential microdata. DER publication workflows are never used to expose protected unit records.
Why it matters
Protects trust and Principle 6 of the UN Fundamental Principles of Official Statistics.
Control area
AI governance
Minimum design requirement
Ground generation in approved sources; retain source links, human review, prompt and model logs and override history.
Why it matters
Designed to align with UNESCO human-oversight principles and NIST risk-management guidance.
Control area
Release authority
Minimum design requirement
Configurable review gates, segregation of duties, embargoes, version control and a withdrawal and correction workflow.
Why it matters
Prevents unauthorised or premature publication.
Control area
Security operations
Minimum design requirement
Encryption, MFA, logging, backup and disaster recovery, vulnerability management, incident response and supplier controls.
Why it matters
Turns a prototype workflow into dependable institutional infrastructure.
Each step has an accountable human role and a visible audit trail.
A typical approval route. Stages and roles are configured for each institution.
Draft product
Analyst verification
Directorate review
Communication clearance
CTO/ICT clearance
DG/CEO approval
Publication
Impact tracking
What is in place, and what is still to prove in production.
The core workflow is visible today. Security, connector reliability, AI controls and impact methodology are validated in production before enterprise claims are made.
Capability
Status
Capability
Role-based access and permissions
Status
Demonstrated
Capability
Audit log of user actions
Status
Demonstrated
Capability
Draft, review, approval and publication workflow
Status
Demonstrated
Capability
AI assistance that stays a draft until a person approves it
Status
Demonstrated
Capability
Configurable review stages and release authority
Status
Configurable
Capability
Data residency options: cloud, sovereign or on-premise
Status
Configurable
Capability
Source-grounded generation with prompt and model logs
Status
Planned
Capability
Embargoes, legal and privacy gates and release calendars
Status
Planned
Capability
Single sign-on and multi-factor authentication
Status
Planned
Capability
Encryption, backup and disaster recovery, security monitoring
Status
Planned
Capability
Data protection impact assessment and security testing before go-live
Status
Planned
Bring your security and data-governance teams early.
We will walk through the controls, the deployment options and what is validated in production.