Lloyd Humphreys | Managing Director, Cogniss, Cambridge, UK
Citation: EMJ Innov. 2026; https://doi.org/10.33590/emjinnov/048UY9L1
You originally trained and worked as a clinical psychologist in the NHS before moving into digital health innovation. Looking back, what first made you feel that technology could meaningfully change the way healthcare is delivered?
My clinical training taught me two things. First, that behaviour change works. That’s what we trained in, and the evidence supports it. Second, access is often the limiting factor, not the approach. As a psychologist, I could spend an hour with a patient and see meaningful improvements, but what happened in the intervening period? And what about people who couldn’t access a psychologist but could benefit from support? It was clear there weren’t enough clinicians or enough capacity.
I had set up one of the Northwest’s first refugee and asylum seeker services, and people were often waiting 2 years to see me, sometimes being deported back to the country they had fled. I felt there had to be a better way, and this is what led me to leave the NHS. I set up my own business with another psychologist, but we faced the same capacity challenges. We were working with children’s services, supporting hundreds of children with complex needs, and we couldn’t see everyone.
We turned to technology to amplify our impact. We created one of the first triage and assessment services, where three psychologists could assess up to hundreds of children per month with just 2 days’ worth of work. Technology meant we could deploy teams where they were most needed.
That led us to ask what else we could do. From there, our work broadened into addiction services, where we were delivering training in psychosocial interventions. Because there weren’t enough psychologists, we were training non-professionals using huge manuals. We digitised this approach and created the first-ever digital therapeutic for addictions called ‘Breaking Free Online’. It was never about replacing clinicians. It was about amplifying impact, reaching more people with the same techniques and tools.
Across your career, you’ve worked at the intersection of psychology, behaviour change, and healthcare technology. How has your clinical background shaped the way you think about designing digital products and patient engagement?
My frontline experience helps me design technology from the perspective of both clinicians and the patients using these tools. I quickly learnt that information alone rarely changes behaviour. A psychologist never tells a patient what to do in therapy; we talk about guided discovery, and we use Socratic questions.
At every level, people don’t stop smoking or manage their diabetes just with information alone. It depends on a whole series of things like motivation, context, habits, barriers, relationships, and lived experience. Technology usually takes a one-size-fits-all approach. It treats everyone with depression in exactly the same way, as opposed to everyone’s unique experience.
In the past, everyone has had to build for the masses because of the cost of doing so and the ability to commercialise those technologies. Now, if we strip that away, we can actually build for the person at a micro level. And that’s the premise of Cogniss (Cambridge, UK), a no-code platform to allow healthcare experts, not software developers, to create experiences that reflect that clinical expertise and behavioural science, and to meet people where they are.
Being able to create hyper-personalised experiences for really small populations and niche communities can bridge the digital divide. We no longer need to take a one-size-fits-all approach. We can actually build for people rather than the population. And that’s been the key.
You’ve now seen several waves of digital health transformation from early digital therapeutics and patient-controlled records through to today’s AI-driven landscape. What has remained surprisingly consistent across those waves despite the pace of technological change?
What has remained consistent is the need to consider the totality of a person’s experience and the full care pathway, rather than isolated components.
In many health systems, including the NHS, there is a tendency to procure individual point solutions for specific parts of a pathway. As a result, the broader system context is often missed, and important elements of the pathway remain unaddressed. Even well-designed interventions can therefore underperform if they are not embedded within an end-to-end model of care.
It is also increasingly clear that this is no longer primarily a technology problem. The capability for non-technical people to easily build digital health tools, including AI-enabled solutions, is now well established. The greater challenge lies in adoption and implementation. Many organisations remain in cycles of repeated pilots, with limited transition to sustained, scaled deployment. The constraint is less innovation itself and more the system’s ability to operationalise it effectively.
Across successive waves of innovation, the pattern has been similar: significant promise followed by partial realisation. The issue is not a lack of capability, but difficulty embedding innovation within the realities of healthcare delivery.
Healthcare systems also face increasing complexity in navigating a crowded technology landscape. Multiple solutions exist across domains, but there is limited clarity on how they should be evaluated, integrated, and governed. This creates barriers around contracting, assurance, and governance, which in turn make it difficult to move beyond pilot phases into sustained adoption. Until these implementation and integration challenges are addressed, digital health innovation is unlikely to deliver meaningful system-level impact at scale.
Do you think we’re getting better at distinguishing between innovations that are meaningful compared to things that are trending?
There is always a hype cycle, and AI is no exception; however, these cycles often overlook the most important factor: the human element. Healthcare is fundamentally a relationship-based and trust-based system, and it cannot be addressed through technology alone. This is reflected in frameworks such as the Digital Technology Assessment Criteria (DTAC) in the UK and Digitale Gesundheitsanwendungen (DiGA) in Germany, which go some way towards providing structure and trustworthiness, but do not resolve the core challenge.
A persistent issue remains: how clinicians on the ground are expected to choose between a growing number of available solutions. This creates ongoing complexity in adoption and decision-making. What is needed is a different approach, with a stronger focus on trust, assurance, and the ability to evaluate and scale solutions at the level of portfolios or care pathways, rather than individual tools.
We often talk about technology transformation in terms of products and innovation, but what role do clinicians play in making those innovations work in practice?
Often, technology is implemented to clinicians rather than with them. At the executive level, a product may be procured and introduced to clinical teams with an expectation of immediate use. However, clinicians may not feel it fits naturally into their workflows or understand how it improves day-to-day practice. As a result, if they have not been meaningfully involved in its development or implementation, there can be reluctance to change established ways of working.
A common issue is that technology is layered on top of existing processes rather than replacing or reshaping them. This reduces its impact. During COVID-19, by contrast, traditional pathways were disrupted, and technology had to be integrated into workflows rather than added on. In that context, adoption was more effective because it replaced parts of the existing system rather than duplicating them.
Since then, many systems have reverted to pre-pandemic models, with technology again often layered rather than embedded. This highlights the importance of clinician involvement. Co-design and collaboration are essential, ensuring solutions are developed with clinicians rather than imposed on them. When done well, this can enable hyper-local solutions that are tailored to specific services, while still being scalable across broader systems.
Could you explain what the digital health publisher model looks like in practice, and why you believe more traditional approaches to digital health deployment have struggled?
We have seen many strong digital health innovations reach what feels like a summit. The traditional journey has often focused on getting a product built, overcoming technical development, cost, regulatory, and compliance challenges. Once these are addressed, teams often feel they have ‘arrived’ at success, typically marked by a successful pilot.
However, this is only one peak. There is a second, often larger challenge: adoption. This is where procurement, integration, and system embedding become critical. Many current models are not designed to bridge this gap effectively. We have tended to assume that every successful digital health innovation should become a standalone company, responsible for navigating this entire journey alone.
A more effective analogy is publishing. An academic does not print and distribute their own textbook or build the entire commercial distribution chain. Instead, they work with a publisher, who manages the end-to-end process of production, distribution, and access. The value is in aggregation and trusted dissemination, not isolated effort.
A digital health publisher model applies this principle to healthcare innovation. It aggregates trusted solutions built on common infrastructure and architecture, providing a single point of access for health systems. From the system perspective, this means one contracting mechanism, one governance and due diligence process, and a more coherent route from innovation to deployment. It builds a structured bridge between innovators and health systems, aligning incentives across both sides.
Because publishers do not rely on individual companies carrying the full burden of commercialisation and scale, solutions can be offered at more flexible and sustainable commercial rates. This makes adoption more viable for systems such as the NHS. In this context, we are also launching a separate digital publishing company designed as a crowdsourced, citizen-owned, and crowdfunded initiative intended to support this shift away from traditional standalone commercial models.
Where do you think optimism around AI in healthcare is justified, and where do we risk repeating old mistakes in a new technological form?
The ‘last mile’ in healthcare is always people and local context. Large-scale solutions often fail to recognise that every system differs: integration requirements vary, pathways differ, and local configurations are unique. This dimension of localisation has historically been underestimated.
There is genuine cause for optimism with AI. When combined with approaches such as digital publishing models and low-code or no-code infrastructure, it has the potential to significantly improve productivity, enhance personalisation, and reduce administrative burden. Emerging technologies such as ambient voice tools already demonstrate this direction of travel. Public-facing behaviours are also shifting, from regular search engines to AI chatbots, reflecting increasing accessibility of AI-driven support.
However, the core challenge remains adoption and implementation. The pace of AI development is extremely rapid, making it difficult for governance, workflows, and trust frameworks to keep pace. As a result, while innovation is accelerating, scaling remains uneven. Without addressing implementation barriers, there is a risk of repeating familiar patterns: high levels of innovation and investment, but limited system-wide adoption.
For clinicians, researchers, or entrepreneurs entering digital health today, what assumptions about healthcare innovation would you most challenge them to rethink?
When mentoring innovators, a common pattern is that they arrive with a predefined solution and then attempt to retrofit a problem around it. A necessary reset is to begin with the problem itself, and only then determine whether technology is the appropriate solution.
In many cases, the solution may not be technological. It may involve pathway redesign, additional workforce capacity, or changes to workflow and process. Technology can play a role, but it is not always the primary answer. Without starting from the problem, it is difficult to identify the correct intervention.
A further shift in thinking is moving from isolated solutions to integrated portfolios of technologies. When solutions are designed as part of a wider system, it becomes possible to focus more deeply on specific parts of a pathway, while relying on the broader ecosystem to support the rest. This enables more targeted design for small or underserved populations.
Ultimately, the goal should not be whether a solution can be built, but whether it can become part of routine care within a real-world system. Whether one is an innovator or a health system stakeholder, the underlying goal is the same: delivering impact, value, and improved outcomes for patients.
