Britain.Direct / GB.1 / IBM ADS Programme
MentalHealth.Direct
MentalHealth.1 — Trusted AI Identity, Discovery and Signposting Infrastructure
Human-facing and machine-facing dual stack for safe, mental health information, support navigation
AI-enabled mental health support infrastructure to complement existing NHS and charity services
Overview
MentalHealth.Direct is proposed as the human-facing gateway for trusted mental health information,
wellbeing education, service navigation and signposting.
MentalHealth.1 is proposed as the machine-facing identity namespace for approved mental health AI Agents,
support services, charity services, local service directories and wellbeing resources.
The programme must be designed around safety, safeguarding, evidence, escalation pathways and human oversight.
AI should support access and navigation, not replace clinicians, therapists, emergency services or crisis care.
Why Mental Health AI Needs Governance
Dual Stack Architecture
↓
Human-facing gateway
↓
MentalHealth.1
↓
Machine-facing AI identity namespace
↓
Britain.Direct discovery
↓
• IBM watsonx • Human oversight
Example AI Agent Identities
Role of MIND as Namespace Steward
- Safeguarding oversight
- Acceptable-use standards
- Escalation pathway governance
- Review of approved information sources
- Human oversight model
- Ethical AI and inclusion guidance
- Participation in clinical and lived-experience advisory boards
IBM ADS Integration
IBM Partners can support deployment through:
- IBM watsonx for governed AI Agents
- IBM Cloud for secure infrastructure
- IBM Automation for workflows and escalation routing
- IBM OpenShift for regulated/hybrid deployments
- Consulting, implementation and managed services
- Monitoring, safety reporting and governance dashboards
Operating Layer
AI Agents can support non-clinical operating workflows under human oversight:
- Service directory updates
- Approved content ingestion
- Resource classification
- Safeguarding flag routing
- Human review queues
- Monthly governance reporting
Safety Principles
- AI must not diagnose.
- AI must not replace therapy or clinical care.
- Crisis language must trigger immediate signposting to emergency or crisis support.
- High-risk cases must be routed to human review.
- All information must be evidence-based and regularly reviewed.
- Privacy, consent and data minimisation must be built into the programme.
Long-Term Vision
The Dual Stack could become a digital mental health signposting infrastructure for Britain.
MIND could be Namespace Steward, with IBM Partners as implementation providers, and AI Agent workflows under human oversight