The value in taking AI strategy beyond the consultation room

Healthcare has adopted AI faster than some people expected. A survey of nearly 600 UK GPs working in NHS and/or private practice, published in May 2026, found that around 40% are currently using an AI scribe, while a further 23% have used one at some time. Among GPs doing regular private work, adoption was nearly three times higher.

The sector isn't holding back from AI adoption. As in most industries, it has focused on applying AI to the most visible pressure points. In this case, that's the documentation burden carried by clinicians. More than 70% of the GPs reported that thanks to these tools, they spend less time writing notes and doing admin, potentially freeing more face-to-face patient time.

Gains from AI healthcare tools are already tangible

The measured gains are impressive as well. In a 2026 survey of UK clinicians, 42% of those who used AI tools said they saved at least 132 hours (around three working weeks) a year. More than a third said it meant they were seeing an average of seven more patients a week. Across an entire provider, that's a substantial amount of extra capacity.

So the case for clinical AI is well established. The question is, what happens to the capacity that it releases?

Released capacity only becomes value if the organisation can convert it. Seven more patients a week per clinician means many more sets of referrals, results, follow-ups and invoices, which the existing operational team has to handle.

The same UK GP survey found that more than a fifth of those who had adopted a scribe then stopped using it. This was more often because the practice they work in didn't have the capacity or processes to support it optimally, rather than because the tool didn't work. There seems to be a disparity between individual willingness and organisational readiness. Even though clinicians often rate these tools highly, some still stop using them. A tool can succeed at task level but fail at the level of the pathway.

Most AI strategies have focused on a single layer

Almost all the AI deployed in healthcare so far does one kind of job: helping the clinician during or immediately around the consultation. We're talking about scribes, ambient documentation and decision support. All these improve the time when clinician and patient are in the same room.

As more providers adopt these tools, the advantage they confer narrows. Consultations run well, clinicians can focus on the person in front of them and the interaction is well documented and supported with relevant, real-time information. But it's what happens beyond that makes the improvement sustainable and exponential.

Unclaimed value resides in a second AI layer

AI is capable of taking that efficiency beyond the consultation room – working out what needs to happen next, then making sure it happens. That can mean flagging a patient who didn't attend and was never rebooked, or an episode of care that was completed but not billed. It might pick up a missing pre-appointment questionnaire or provide information that the next team needs before they see the patient.  

The clinician tools are valuable at a single – and crucial – point in the patient journey. But when they stand alone, healthcare organisations are not optimising their potential. A provider can buy the best clinical AI available and still find that patient needs are not met, because there's no change in the coordination between one step and the next.

The impact on workload for the clinician and beyond

Consider a patient seen privately about a changing mole. The consultation, supported by clinical AI, is efficient: the dermatologist takes a biopsy and explains what happens next.  

Outside the consultation room, the pathway continues. The sample is sent to an external lab. The results come back – or if they don't, they need chasing. Then, a specialist has to recognise what the histology means and act on it. If the next step is an excision, the patient needs to be informed and prepared. If the patient is insured, the excision needs its own authorisation, separate from the one that covered the biopsy. Someone must book surgery time and space and confirm them with the patient. If there's other advice or no further action, someone needs to tell the patient.
‍

These steps must happen sequentially and in good time. Insurance approvals expire and delays can lead to repeat work. A result – such as a biopsy – that sits unactioned, because nobody owns the handoff, can carry real clinical risk for the patient.

Around a third of a community-based clinician's time is typically spent on admin and coordination. The rest falls on referral coordinators, schedulers, practice managers and billing teams. Seeing more patients in clinic generates more downstream actions for the same number of people to carry.

‍

Task-level returns can hide pathway-level costs

Most AI investment is assessed at the level of the clinical task it improves, such as minutes saved per note or hours returned per clinician per week. Those numbers rarely add up to anything visible in the organisation's operating position. That's because the impact is not yet being amplified by carrying AI through into the administrative and operational layer.

McKinsey's 2026 survey of healthcare leaders identified administrative efficiency as the area with the greatest potential for AI. Around half of the responding organisations were already exploring agentic approaches. They understand where the value lies: now it's time to champion the investment.

Reviewing strategy means looking at both AI layers

Measuring return along the whole pathway is far more significant than at the level of the individual task. Key questions include:

• What AI tools or applications are individuals and teams using?
• Are these tools in the clinical layer or the administrative layer?
• Which parts of the patient pathway have measurably changed as a result?
• How much operational work exists because systems don't trigger the next action?
• Is benefit and ROI measured at task level or pathway level?

The last question may be the most important. At task level, you're only seeing the performance of one or more AI tools, with no visibility of the downstream impact.

Care orchestration is the term emerging for this second layer, which connects tasks, people and communications across an entire pathway, maintaining progress without manual intervention at every step. This is the space Semble operates in, helping providers bring those parts together so care moves reliably from one stage to the next.

Clinical AI releases capacity in a meaningful and measurable way. What happens to that capacity is harder to see. It could convert into completed care pathways, or it could merely disperse into uncosted coordination work that neutralises the clinical advantage. Taking a care orchestration perspective improves your AI strategy both operationally and commercially.