We've All Been There: What A&E Chaos Teaches Us About AI and Hospital Waiting Lists

Hour four in A&E
By hour four, I knew the vending machine almost as well as the staff. I knew it took cards, eventually. I knew the crisps in row C had probably been there for a while. And I knew the "current wait time" on the screen felt more like a rough estimate than a promise.
If you live in the UK, you've probably had your own version of this. A twisted ankle, a child with a stubborn temperature or a parent who "just doesn't seem right". You arrive at A&E, check in, and enter that strange waiting-room limbo where time seems to slow down.
From the waiting room, it looks like chaos. Phones ringing, nurses rushing past an d names being called, yet one thought kept coming back:
Nobody here is standing still, so why does everything feel stuck?
The honest answer is not that hospital staff are slow. We all know that they are drowning, and most of the water is paperwork.
The NHS waiting list problem is not what you think
The problem isn't a lack of data but It's more of a lack of capacity to process it.
Hospitals are not short of information either - every referral letter, triage note, test result, cancellation and theatre slot is recorded somewhere. The trouble is that it lives in different systems, inboxes and spreadsheets, and a human has to read it, interpret it and act on it, one case at a time.
Multiply that by the scale of today's NHS backlog. At the end of August 2026, an estimated 7.37 million treatments were waiting to be carried out in England, relating to 6.24 million patients, the highest level since 2025. That is not a queue, that is literally a small country.
Here is the bit that rarely makes the headlines. Much of what slows patients down is not the surgery or the scan itself but it’s everything around it: the referral that sits unread, the appointment letter that goes to an old address, the theatre slot that goes empty because nobody spotted the cancellation in time. Administrative work has been estimated at upwards of 30% of healthcare costs in the US, and anyone who has worked in or near the NHS will recognise the pattern.
So, the chaos I saw from my plastic chair was not people doing nothing. It was highly skilled people doing an enormous amount of the wrong kind of work.
Imagine an AI agent that never needs a tea break
This is where AI in healthcare gets interesting, and not in the robot surgeon way. The most useful AI for hospital waiting lists is quietly unglamorous: an AI agent that works through the admin mountain so clinicians do not have to.
It is already happening. In Canada, Manitoba researchers are using artificial intelligence to help patients get surgery sooner and reduce hospital backlogs. In the US, ARPA-H programme manager Dr Bon Ku has argued that today's models are already strong enough; the real challenge is getting AI into clinical workflows.
So, picture an AI agent sitting behind the scenes of a busy hospital. It could:
Prioritise waiting lists, continuously reordering patients by clinical urgency, time waited and risk, rather than relying on a list that was last tidied on a Tuesday.
Review referrals as they arrive, checking they are complete, flagging missing information and routing them to the right specialty first time.
Coordinate appointments, matching patients to cancelled slots and available clinicians, so an empty theatre becomes a shorter queue instead of a wasted morning.
Identify bottlenecks, spotting where patients pile up (diagnostics, discharge, a single overstretched clinic) before it turns into a crisis.
Escalate urgent cases, raising a flag to a human clinician the moment something in a referral or a test result suggests a patient cannot wait.
None of this replaces a doctor's judgement. It simply makes sure the right information reaches the right person at the right time, which, if you have ever spent four hours with a vending machine, sounds almost like science fiction.
More time with patients, less time chasing paperwork
What if staff spent more time helping patients and less time chasing paperwork?
Not fewer nurses, but nurs
es who spend their shift nursing. Not doctors replaced by algorithms, but doctors who walk into clinic with the referral already checked, the history already summarised and the urgent cases already at the top of the pile. Dr Ku himself pushed back on the idea that automation is mainly about cutting healthcare jobs.
Of course, healthcare is not a place to "move fast and break things". Patient data is among the most sensitive information there is, and any AI agent working with it needs to earn trust. In practice, that means:
Data stays put. The AI runs inside the organisation's own secure infrastructure, not off in someone else's cloud.
Humans stay in charge. The agent recommends and escalates; clinicians decide.
Every action is visible. Clear audit trails show what the agent did and why, so it can be checked, challenged and improved.
Compliance is built in, aligned with GDPR and recognised standards from day one, not bolted on after launch.
And crucially, start small. Rather than trying to fix the whole NHS backlog in one heroic leap, pick one high value process (referral triage, say), prove it works, measure the hours saved, then move on to the next.
So, next time you are in A&E...
Look past the chaos. The people rushing around are not the problem; they are the solution, buried under admin. The question for every healthcare leader is simple: how much of your team's day is spent caring, and how much is spent chasing?
AI will not make A&E waiting rooms disappear overnight. But it can take the paperwork off the people who should be with patients, and turn a mountain of unread data into decisions that happen on time.
At Neologik, we help organisations in complex, regulated environments put AI agents to work on exactly this kind of high value process, inside their own infrastructure, with governance built in. If your waiting list keeps growing faster than your team can read it, let's talk.
And if you happen to be reading this from an A&E waiting room: avoid row C.



Comments