Care Connect 365: Leveraging Microsoft 365 AI Agents to Bridge Social Needs and Healthcare Services in the NHS

The UK’s National Health Service (NHS) faces a well-documented challenge: addressing the wider determinants of health—housing, employment, social isolation, financial insecurity, and access to community support—requires seamless coordination across a fragmented ecosystem of NHS services, local authorities, voluntary and community sectors (VCSE), and social care providers.

Integrated Care Systems (ICSs) were established to tackle this, bringing together partners for population health and reducing inequalities.

Social prescribing link workers already connect patients to non-clinical support, but manual processes, siloed data, and limited visibility into provider availability create bottlenecks, delays, and missed opportunities.

A digitally empowered “Care Connect” system—built using Microsoft 365 tools, particularly Copilot Studio, Power Platform, and related Azure services—could transform this by deploying intelligent AI agents that proactively identify needs, match them to services, coordinate care, and measure outcomes.

The Opportunity: Why AI Agents Now?

Recent NHS trials of Microsoft 365 Copilot demonstrate substantial productivity gains: an average of 43 minutes saved per staff member per day, potentially equating to hundreds of thousands of hours monthly across the system. Over 50,000 NHS staff already use it, with broader access expanding.

Microsoft’s ecosystem is particularly suited for healthcare:

• Copilot Studio enables low-code/no-code creation of custom AI agents.
• Power Platform (Power Apps, Power Automate, Power BI, Dataverse) supports rapid development of workflows, apps, and analytics.
• Integration with Microsoft Cloud for Healthcare, Azure AI, and existing NHS tools (including FHIR standards for interoperability) ensures compliance with data governance (NHS IG Toolkit, GDPR).

NHS partnerships with Microsoft, including five-year agreements and pilots with trusts like Manchester University NHS Foundation Trust and Oxford University Hospitals, provide a strong foundation.

Architecture of Care Connect AI Agents

Care Connect could be implemented as a suite of interconnected agents operating within the secure NHS Microsoft 365 environment (e.g., via NHSmail or tenant integrations).

1. Needs Identification Agent: Deployed in primary care, community teams, or self-referral portals. It analyzes structured (e.g., EHR notes) and unstructured data using natural language processing (via AI Builder or Azure AI) to flag social needs—loneliness, housing instability, benefits issues—against validated frameworks like the NHS Personalised Care model.

It integrates with existing systems (EMIS, SystmOne) and suggests prompts for clinicians or link workers. Patient-facing versions (via Power Pages portals) could enable secure self-reporting.

2. Matching and Referral Agent: The core “connect” engine. It maintains a dynamic, federated directory of local services (updated via APIs or Power Automate from VCSE providers). Using retrieval-augmented generation (RAG) and semantic search, it matches needs to available, suitable providers based on criteria like location, capacity, accessibility, cultural fit, and outcomes history.

It generates personalised referrals, handles consent, and sends secure notifications via Teams or Outlook. For complex cases, it escalates to human link workers with contextual summaries.

3. Coordination and Follow-Up Agent: Acts as an orchestrator across the ecosystem. It tracks referral status, sends automated reminders (with patient consent), surfaces updates in shared workspaces (e.g., Teams channels for ICS multidisciplinary teams), and flags risks like non-engagement. Power Automate handles workflows, such as escalating unmet needs back to NHS services.

4. Outcomes and Insights Agent: Leverages Power BI and Azure Synapse for analytics. It measures impact using standard metrics (e.g., ONS-4 wellbeing, reduced GP appointments, A&E attendances) and feeds into ICS population health dashboards. Predictive elements could identify at-risk cohorts for proactive intervention.

These agents build on proven patterns, such as Microsoft’s Healthcare Patient Support Agent architecture and NHS-specific pilots using Copilot Studio.

Benefits and Expected Impact

• For Patients: Faster, more personalised support; reduced “revolving door” of unmet social needs driving health decline; greater empowerment via digital access.
• For Clinicians and Link Workers: Dramatic reduction in administrative burden; more time for complex cases; better visibility and coordination.
• For the System: Cost savings from prevented escalations to acute care; improved population health metrics; stronger VCSE integration; data-driven commissioning.
• Equity: Targeted matching helps address inequalities in access for underserved communities.

Early Copilot data suggests significant time and cost returns; Care Connect could amplify this in the social domain.

Challenges and Mitigations

• Data Privacy and Security: Use Microsoft Purview, Entra ID, and record-level security in Dataverse. Adhere strictly to NHS standards and conduct regular audits.
• Interoperability: Build on FHIR and existing APIs; avoid rip-and-replace by layering agents on legacy systems.
• Adoption and Digital Divide: Hybrid human-AI model; user training; accessible patient interfaces.
• Regulation and Ethics: Align with MHRA/UK AI frameworks; ensure agents augment, not replace, professional judgment.
• Equity of Provision: Address varying digital maturity across ICSs through national templates and shared learning.

Conclusion: A Catalyst for Integrated Care

Care Connect represents more than technology—it embodies the shift to proactive, personalised, whole-person care envisioned in NHS Long Term Plan and ICS strategies. By harnessing Microsoft 365’s mature, secure, and increasingly agentic AI capabilities, the NHS can create an intelligent connective tissue across the social-healthcare ecosystem.

With strong governance, collaborative development involving clinicians, patients, and partners, and iterative deployment starting from proven pilots, Care Connect could become a national exemplar of how AI delivers tangible improvements in outcomes, efficiency, and equity. The building blocks exist today; the time to assemble them is now.