Moving AI from potential to practice in health and social care

Guest blog by Laura Thompson, Director of Marketing for Health, Social Care and Technology at The Access Group

Last week, I had the privilege of speaking on an AI panel at the Health and Care Summit alongside Rob Reid from MHRA and colleagues from across the sector. The conversation centred on a question that matters to all of us working in health and social care: how can AI actually reduce costs and increase productivity, rather than just promise to do so?

Here are the insights I shared, drawn from our work deploying AI solutions across NHS trusts, local authorities, and care organisations.

Where AI delivers real productivity gains today

The productivity crisis in health and social care is well documented. But AI is already creating measurable capacity gains in three distinct areas.

Clinical documentation is eating precious time. The Health Innovation Network reports 1.75 million daily patient appointments across the NHS. Ambient voice technology captures consultations and generates documentation automatically. When you save 5-10 minutes per appointment on note-taking, you create capacity equivalent to adding hundreds of professionals to the system. This frees people to focus on the person in front of them rather than the screen beside them.

Fragmented systems waste hours every day. How many times do your teams open four different systems to piece together a patient's story? We can now bring data from multiple clinical systems into a single dashboard. AI summarises information from separate EPRs, care management systems, and clinical notes, with references back to source records. This saves 10-15 minutes per patient review. Solutions like this are being rolled out at Shropshire Community Health NHS Trust and Midlands Partnership NHS Foundation Trust, demonstrating how AI can reduce friction in care coordination.

Prevention beats crisis response every time. In Sutton, technology enabled care uses AI to analyse sensor data from vulnerable residents' homes. It spots changes in daily routines before they become crises. The system generates 25 referrals per month to urgent community response services. These are people who would otherwise have ended up in A&E or admitted to hospital. The solution has now scaled to 500+ residents across 22 schemes, and DSIT highlighted it as a model for replication.

Why strong governance enables faster adoption

There is a common assumption that governance and speed exist in tension. My experience suggests the opposite.

Access Group holds ISO 42001 certification for responsible and ethical AI governance. This international standard forced us to build better products. We identified risks earlier, built explainability into our systems from the start, and created audit trails that satisfy both regulators and professionals.

The result? Faster adoption. NHS trusts and local authorities can move through their internal approval processes more quickly because the assurance work is already done.

We are also the only technology enabled care provider featured in the government's Algorithmic Transparency Recording Standard. AI systems are subject to public scrutiny about how they work, what data they use, and how decisions are made. That transparency builds trust with professionals and the public.

When we developed intelligent care platforms to summarise patient information across multiple systems, ISO 42001 required us to answer difficult questions from the start. How does the AI handle conflicting information from different sources? How do we ensure it does not miss critical details? How do we audit what the AI included or excluded? Those questions led to better design decisions that make platforms safer and more trustworthy.

Building professional confidence through explainability

Professionals do not trust black boxes, and they should not have to. Trust comes from five practical commitments:

Explainability is non-negotiable. Ambient voice technology shows people which parts of the conversation informed each section of the note. Intelligent care platforms provide references back to source records so people can see where information came from. The AI assists. The person decides.

Validation happens in real workflows. Tools developed with frontline staff from the start work the way they work. They shape the interface, the terminology, and the review process.

Performance transparency builds trust. We publish accuracy metrics for our tools. We are honest about limitations. That honesty builds trust more effectively than overpromising.

Humans stay in the loop. Technology enabled care systems generate alerts when behaviour patterns change. But a human reviews every alert and decides the response. The AI spots the pattern. The professional interprets the context and takes action.

Ongoing engagement matters. We run user groups with people who use our tools. They tell us what works, what frustrates them, what they need. People trust tools that respond to their input.

Why pilots fail to scale (and how to fix it)

At last year's Labour Party conference, Health Secretary Wes Streeting remarked that "NHS has more pilots than the RAF." He was right. The graveyard of pilots is real. But the barriers to scaling have little to do with whether the AI works.

Integration is underestimated. Pilots often run in isolation with manual workarounds. Scaling requires full integration with existing systems. Intelligent care platforms need to connect with multiple clinical systems, each with different data structures and APIs. That upfront integration effort is why successful solutions scale. The Sutton technology enabled care deployment succeeded because integration with care planning systems was built from the start. Carers see alerts in the same mobile app they use for visit records.

Funding models are misaligned. Pilots get innovation funding. Scaling requires recurrent operational budgets. AI tools deliver productivity gains, but those gains often appear in different budget lines from where the costs sit. A tool that prevents hospital admissions saves the NHS money, but the local authority pays for the technology. We need funding mechanisms that recognise whole-system benefits.

Change management is underfunded. Scaling means changing how hundreds of staff work, not just how a pilot team operates. Many organisations fund the technology but not the transformation. Successful scaling requires executive sponsorship, cross-departmental collaboration, and realistic change management.

The graveyard of pilots happens when we treat AI as a technology project rather than a service transformation.

Measuring what matters: Beyond traditional ROI

Traditional ROI calculations struggle with AI in health and care. We need to measure four types of impact:

Time saved creates capacity. When people save 5-10 minutes per consultation or 10-15 minutes per review, that capacity means more people seen, reduced waiting times, or time for proper breaks. Time is the scarcest resource in health and care.

Crises prevented save lives and money. Each of the 25 monthly referrals from Sutton technology enabled care is someone who might otherwise have fallen, missed medication, or deteriorated at home. Those prevented crises avoid ambulance callouts and hospital admissions. The financial value is significant. The human value is greater.

Capacity created improves quality. When AI handles routine monitoring or administrative tasks, staff can focus on complex cases that need human judgement. That capacity creation shows up in improved care quality and staff satisfaction.

System-level savings require system-level metrics. AI prevents hospital admissions, reduces length of stay, or speeds up discharge. Those savings appear across multiple organisations and budget lines. We need ROI frameworks that capture whole-system value.

For professionals, the ultimate ROI measure is simple: does the tool make their job better? If people choose to use the tool when they have alternatives, it delivers value.

What health and care leaders should focus on

If you are leading digital transformation in health or social care, here is what matters:

Build integration from day one, not as an afterthought. Your AI solution needs to work with the systems your teams already use.

Invest in governance frameworks that build trust. ISO 42001 certification and transparency standards like ATRS are not bureaucratic hurdles. They are enablers of faster, more confident adoption.

Involve frontline staff in design and validation. The tools that succeed are built with the people who will use them, not for them.

Plan for service transformation, not just technology deployment. Change management, training, and workflow redesign need the same attention as the technology itself.

Measure what matters to your teams. Time saved, crises prevented, and capacity created are as important as cost reductions.

The path forward

AI in health and social care is moving from potential to practice. We have proven solutions delivering measurable productivity gains. We have governance frameworks that build trust. We understand the barriers to scaling and how to overcome them.

The question is no longer whether AI can help. The question is how quickly we can scale what works.

Health and Social Care Programme activities

techUK is helping its members navigate the complex space of digital health in the UK to ensure our NHS and social care sector is prepared for the challenges of the future. We help validate new ideas and build impactful strategies, ultimately ensuring that members are market-ready. Visit the programme page here.