Access AI in Healthcare Conference 2026

The first session of the day was a cross-sector panel chaired by our very own Laura Thompson, Marketing Director at Access Health​, with guest speakers:

• Professor Joe McDonald, Medical Director, former Consultant Psychiatrist & Chief Clinical Information Officer at CNTW NHS Trust​
• Dr Rowan Sil, GP & Chief Clinical Information Officer, Leicestershire Northants & Rutland ICB Cluster
• Steve McManus, recently appointed Interim Chief Executive of NHS England's College of Leadership and Management and former CEO of Berkshire Healthcare NHS Foundation Trust​

The panel engaged in a broad discussion about the current state of AI in healthcare, with a lot of talk about Ambient Voice Technology and the benefits it’s offering, but quickly pivoting to the challenges facing mental health services and the caution needed surrounding AI deployment in that sector.

Professor McDonald brought up an anecdote about 1986 and his judgement of a friend’s mobile phone. When we think of how far technology has come from those bulky brick-like devices to modern smartphones, where will AI be in the same time frame?

AI is already optimising healthcare, but the possibility of radical change in mental health is huge, and we must automate some of the clinical processes – beyond digitised paperwork – in order to resolve concerns within the Fonagy Report and surrounding the neurodivergency assessment backlog.

For Dr Sil however, AI’s first task must be to achieve better databases and data communication. The current processes are about integration and interoperability, but this can be clunky. In his words it’s why he joined Access – because they’re pursuing this approach, moving beyond integration protocols.

While there was some talk of human resistance to change, to new tech, to being led or dictated to, Steve McManus hit the nail on the head with his concluding remarks: we’ve got to put clinicians in the driving seat, we’ve got to release more clinical time, and crucially taxpayers want value for money, so we have to demonstrate that. The good news, says Steve, is that there have been many positives at the national and international level regarding coordination, development, guardrails and the deployment of AI, and that the UK has a prime opportunity to lead by example and be a main player in the future adoption and expansion of AI. As UK organisations, this will only benefit the work we’re doing.

To give people a chance to meet those around them, we held a brief roundtable discussion to gauge where organisations were currently with their usage of AI and their impressions of it as a solution.

Feedback from speakers was generally positive, but flagged concerns such as a risk of dependency on AI and a loss of critical thinking, ensuring clinical relevance rather than broad AI support, a desire to include clinicians at each step of the development and implementation process, and worries about the software infrastructure to embed AI – and the subsequent adoption rate issue that can be hindered by diminished capabilities.

That being said, people were excited by the prospect of providing a better patient experience and allowing AI to assist them so they can focus on the human interaction elements and deliver more personalised care, and given how much has changed in the last 10 years alone… the future’s exciting, if a little scary, but the consensus was that healthcare professionals and organisations alike are pulling in the same direction.

The next major panel discussion was presented by Hadleigh Stollar, CEO of Healthcare Innovation Consortium​ (HIC), with experts including:

• Kas Russell-Surtees, Digital Programme Manager, Cambridge Children's Hospital (part of Cambridge University Hospitals (CUH))​;
• Kate Hepton, Interim Chief Nursing Information Officer (CNIO) at University Hospitals of Northamptonshire, Co-Chair HIC EPR Network Paper-to-Digital Specialist Interest Group​; and
• Dheeraj Rathee, Chief Technology and Analytics Officer at Provide Digital​

After an initial look at what HIC do with their EPR Network, Data and Digital Genomics Network, and an AI Network coming in 2027, the discussion quickly moved from who’s using AI to who can demonstrably prove that AI has been beneficial – beyond anecdotal testimony.

This was a notable moment of the day with no hands raised. People are keen on AI solutions, they’re using them to enhance care via optimised workflows, but proving their impact is difficult and an area that needs further work.

Kas Russell-Surtees pointed out that without hyping AI up, there’s little to no money and thus no interest. We organisations have to be using AI to get the funds to make AI what we need it to be, but even then, some clinical settings or buildings are dated and cannot handle modern tech and electronics. We’re fighting a battle of limited time, limited funds, and in some cases having to use funds on estates rather than staff or solutions.

Dheeraj Rathee also chimed in on the cost front: saving 20 minutes is good, but is it cost saved or time and money spent elsewhere? This, in Kas’ eyes, is why we build carefully and infrastructure first. Know where tools can be implemented before committing.

Kate Hepton rounded things out with some wisdom on AVT and its popularity, cautioning that it fails to capture non-verbal comms like body language and facial expressions. Things are still missed, and this can have an impact in multidisciplinary care. She also flagged that healthcare professionals – especially management - need to know the evidence base for formulation. “I question people, so AI should be no different” she said.

Next up was a solo presentation from Dylan Robson, Data & Digital Coach at Everyturn Mental Health, who neatly followed on from the concerns of Kate Hepton by exploring bias in AI and the importance of properly challenging and appraising data capture and analysis.

His example showcased CognoSpeak; an AI identifying dementia based on language. The problem is that the solution was suggesting dementia in people who misspoke. Native English trained it, so local dialects and migrant variations or mannerisms weren’t understood and thus deemed a concern.

With the University of Sheffield, a study was performed on 1300 people – all confirmed of sound mental acuity and health – and a significant number were misidentified as a dementia risk.

The worry, said Dylan, is that aside from improper care, there’s also a loss of trust and engagement that can happen with a solution or worse – a provider. Errors are also at scale. A doctor is 1-1 with a patient, but an AI is 1-many, potentially.

This is why rigorous testing exists, and these issues aren’t always malicious, or failed data interrogation. Sometimes it’s just an absence of information, but there’s a duty of care to pursue these gaps and fill them in to complete the picture and deliver appropriate care.

Before lunch it was the turn of Katherine Tattershall to interview Jordan Ashenden, Regional Healthcare Manager at Microsoft UK, about the future of work.

In Jordan’s view, AI is democratising access to information and is suitable as a personal assistant for our daily working lives. We’re already experiencing AI through smartphones helping with searches, or enhancing functionality in dating apps, as examples, but beyond that is where risk is incurred – especially within healthcare.

For Jordan, delegation to AI is extremely powerful but needs equally powerful and deliberate guardrails, with an ethos of AI done for you but not to you. This, he says, should be built with clinical expertise alongside developers, but shouldn’t lead to the replacement of clinicians by agentic AI – something agreed upon during the subsequent Q+A session. One attendee cited presenter and scientist Hannah Fry’s recent experiment where AI stole money from a bank account to achieve its instructions, and we’ve also seen issues with AI agents collaborating to achieve a goal by altering its environment.

Rogue behaviour is a risk, AI shepherding and the loss of human intuition and relevance is a risk, but the advent of AI does offer more opportunities for work - potentially in the same way machine operation expanded upon human labour.

In his closing remarks, Jordan suggested that while we can bolt AI on to current healthcare solutions and systems, the real benefits will come with redesigns to include AI properly. The question then is, what remains human?

After lunch it was time for an Access presentation, brought to us by ​Mark Jubb, Head of Customer Engagement at Access Health​, and Tom Smullen, Head of Product Marketing at Access Health​.

Their demonstration started by exploring how the cost of problems in clinical workflow is felt by people, not just systems, before moving on to how we remedy this.

Access Evo Builder was showcased on stage, citing a proposed solution from Professor Joe McDonald about an AI that could prompt patients with social prescribing opportunities whilst on waiting lists for GP and/or secondary care. The application, literally made during the lunch break, was rudimentary but functional and showcased how it could connect to records and other solutions to coordinate recommendations.

Also on show were the Patient Quick Contact App for skimming an EPR to engage a patient whilst they wait on a clinician, the Evo HCP Assistant Agent, and the Evo Rio Guardian + Agent app, which handles security against unauthorised access. This is pertinent given the news of improper record access lately. The solution automatically reports breaches to HR and loops line managers in.

Their conclusion was simple: use AI where it makes sense. You don’t need hundreds of apps, but you also shouldn’t miss out on a good thing. You can pick and choose and build your own ecosystem.

The final roundtable of the day was a follow-up to the morning’s chat. That discussed where we’re at with AI, while this time it was about where we’d like to be.

All tables unsurprisingly echoed the same points: people were happy to use AI tools, either standalone or embedded in existing solutions, but they lacked confidence. They wanted risk assurances, to be taught and confident in usage, for functionality to be easy to pick up, and to know that the work they were doing with the AI was worthwhile.

This last part is particularly interesting, because it brought up the HIC talking point about proving the AI benefits beyond the anecdotal “well it saved me 20 minutes of work today” but also challenged the real-world issues around AI and data centres and the criticisms of noise pollution and water usage for cooling.

The final session of the day was a longer sit-down discussion between Dr Rowan Sil, GP & Chief Clinical Information Office at Leicestershire Northants & Rutland ICB Cluster​, and Vicky Mudd, Clinical Director at Access Health​.

Guardrails was pretty much the keyword of the day, if one had to be picked, so it was apt that this was our finale. For Dr Sil though, he wonders where in the process do the guardrails go? For him, AI is creating data, not just capturing it and analysing it.

Vicky posed the issue of moving from traditional healthcare and efficiencies to greater involvement with mental health records and clinical suggestions – something she described as “a high clinical risk”. In her mind, mental health care is a very difficult area to assess the impact (positive or negative) of AI.

One way round that, said Rowan, is to alleviate the mental toll on clinicians so that they can do more without anguish. Clinicians deal with a lot of trauma, so is there a way that AI can reduce exposure?

Ambient Voice Technology was mentioned here, and Vicky again flagged how important the human in the loop is with these types of tools; ensuring no laziness, no errors, and accountability.

This then led to a final point about the DCB0160; the NHS clinical safety standard for healthcare organisations that “deploy, use, maintain, or decommission health IT systems in England”. As procurement changes and needs change too, Vicky and Rowan praised these standards and said how crucial they are, along with upcoming CQC standards, for validating the safety of tech and those who develop it for an arguably sceptical customer base, ultimately making accredited solutions that are more appealing to buy.

With this, the day’s conference was brought to a close, with some final remarks from Laura Thompson and Steve Wightman respectively to thank everyone for their time and their input. It was a great success, connecting people from across the UK healthcare industry to properly air out the hopes and dreams for healthcare services, as well as what providers like Access can do to support them. We can’t wait to do it all again next year, and we hope you’ll be there with us.