Scaling AI in healthcare

Across the NHS, there is growing recognition of artificial intelligence’s (AI’s) potential to boost productivity and improve care, yet many organisations remain stuck in ‘pilot mode’. The challenge often isn’t whether AI works, it’s how to scale it effectively.

With financial pressures mounting and demand continuing to rise, the current toolkit is no longer sufficient. Instead, there is an opportunity to enable NHS professionals with AI and intelligent automation to boost productivity, but crucially to do so in a way that improves patient and workforce experience. 

This is a new path but is not unproven: Accenture research across industries shows that the number of organisations with fully modernised, AI-led processes nearly doubled from 9 per cent per cent in 2023 to 16 per cent in 2024, and leaders in these organisations are seeing 2.4x greater productivity. Whilst healthcare is not always comparable with other industries, there are similar productivity gains to be realised. For example, research from the King’s Fund finds that nearly two in every three patients have experienced a problem with NHS admin or poor communication.

The barriers and the path forward

But scaling AI in NHS organisations is complex. Data remains fragmented and siloed. Legacy systems and infrastructure gaps slow down integration. Talent shortages, operational capacity for change, and workforce readiness add further friction. Finally, there are safety concerns along with regulatory and compliance challenges.

Based on experience delivering over 2,000 generative AI projects globally, Accenture has identified five imperatives to help leaders scale AI with confidence, which are relevant to the NHS.

1. Focus on delivery for patients

It may require bold thinking, but moving beyond isolated pilots and automation of single workflows can help reimagine entire organisational processes. For example, AI can change the way that end-to-end patient journeys are experienced, from referral and triage to discharge and follow-up, improving outcomes and reducing pressure on overstretched services. 

Whilst the vision and roadmap need to be bold, AI-enabled process change will need to be deployed and proven incrementally due to the complexity involved. Solutions such as AI-enabled triage, EPR copilots, or AI-generated discharge letters are likely to involve multiple regulated medical devices.

2. Build a secure, interoperable digital core

The NHS holds vast, rich datasets, but they are often locked in disconnected systems. Much of the data is deeply sensitive and must be treated as such. Integrating and securing this data is essential to power AI that is effective and trusted. This is difficult and needs due attention and care. There is an opportunity to enable and accelerate this agenda with some of the NHS critical initiatives already in flight such as the Federated Data Programme (FDP) and the Single Patient Record (SPR).

3. Reinvent talent and ways of working

As one of the largest employers in the world, there is tremendous potential for the NHS to proactively re-design the nature of work for an AI-enabled workforce. Using automation to tackle backlogs of administrative tasks frees up human capacity for more specialist work and equipping the workforce with AI assistants can significantly boost productivity. For employees this means a reduction in emphasis on skills like documentation, towards skills like critical thinking, data literacy and digital ethics. 

4. Close the gap on responsible AI

Patient and workforce trust must be the top priority. Data must be used responsibly, and AI must be explainable, fair, and secure, especially when it’s used to support diagnoses or treatment decisions. Clear governance and transparency are essential to maintain public confidence, as well as regulatory and policy adherence and navigation of standards for software as a medical device. We know that 76 per cent of organisations across industries have fully operationalised their governance model for AI; a dramatic increase from just 31 per cent two years ago. However, far fewer organisations have implemented measures such as a systematic risk-identification process, risk testing and mitigation measures, or fully operationalised AI monitoring and control processes. AI must not be used for direct patient care, or scaled for any use, without these measures in place.

5. Drive continuous reinvention

AI-driven transformation is not a one-off project. NHS organisations must embed continuous innovation into transformation programmes: scaling what works, learning from what doesn’t, and adapting as technology evolves.

The opportunity

The future of AI in the NHS is about embedding it into the fabric of how services are delivered, which will involve rethinking organisational processes and ways of working and engaging staff and patients in design. By doing this responsibly and strategically, leaders can empower their workforce, improve access, experience and outcomes for citizens, and build more resilient systems. 

Catherine Inness is managing director of data and AI at Accenture. You can follow Catherine on LinkedIn.