Why my power outage proved the future of AI in healthcare and why it could finally end the 8am scramble

Earlier today I had a power outage at home. Nothing dramatic, but enough to disrupt the day and send me straight to Facebook Messenger to contact the electricity provider. Within seconds, I knew what I was dealing with: a bot. A limited one. Scripted responses, minimal options, no understanding of nuance. It couldn’t interpret my postcode, it couldn’t tell me whether this was a local issue or a wider outage, and every question I asked triggered the same generic flowchart.

So I did what most people do in this situation, I gave up.

But shortly afterwards, I received a message from a human. A real person, clearly working in a call centre, who asked the right questions, checked the right databases, and solved the problem within minutes. No frustration. No repeated explanations. No guessing. Just clarity.

That small experience perfectly sums up where we are with customer service and AI today: most organisations are deploying old-style chatbots that are fundamentally unfit for purpose. They can’t understand context, they can’t interpret natural language, and they certainly can’t handle even mildly unusual queries like a sudden power outage.

But the crucial thing is this: the technology exists to do this brilliantly, it’s just not being implemented yet. And nowhere is this more relevant than in healthcare, where outdated systems are still causing daily frustration for patients and staff.

AI can handle these problems - if we let it

If my outage query had been handled by an advanced AI system, it wouldn’t have failed. In fact, it would have performed better than a human in many ways.

Modern AI can understand free text in the same way a trained call-handler does. If I’d typed:

“My electricity has gone off in LE14 - it happened around 2pm - is it part of a wider outage?”

AI could have instantly:

• Read the intent (“verify outage”)

• Identified critical data (“LE14”, “2pm”)

• Cross-referenced live outage data

• Checked the fault status for my exact location

• Provided accurate updates

• Escalated to a human only if genuinely needed

Not through a flowchart. Not through scripted options. But through intelligent reasoning.

The issue isn’t the capability of AI, it’s the implementation of AI. Companies are still deploying the equivalent of digital switchboards when they could be installing digital problem-solvers.

And this brings us directly to the GP access crisis. Because if AI can handle customer service triage, it can absolutely handle healthcare triage.

The 8am scramble isn’t a clinical problem, It’s a system problem

Everyone in the NHS knows the famous 8am rush. The phone lines open, hundreds of patients compete for a finite number of appointments, receptionists are overwhelmed, and by 8:07 every same-day slot has vanished. Patients hang up frustrated, staff feel attacked, and practices are forced into firefighting mode before the day has even begun.

Here’s the truth: This is not a medical crisis, it’s an access crisis.

The clinical issues are the same every day. The administrative system is the part that’s broken. And AI can fix it in ways no human-resourced system ever could.

How AI could end the 8am scramble for good

Imagine this. Instead of waiting for 8am, a patient can contact their GP practice at any time using natural language: “I’ve had chest tightness since last night. It’s worse when I climb stairs.”

Or:

“I think I need a fit note extended.”

Or:

“My child has a temperature and a rash.”

The AI system instantly triages this, not with tick-boxes, but with real clinical reasoning:

• “Chest tightness + exertional symptoms = urgent same-day review”

• “Fit note extension = administrative task, no GP needed”

• “Fever + rash = ask follow-up red-flag questions”

And then it acts:

✔ Routes to the right service

• Same-day GP

• Nurse practitioner

• Pharmacy consultation

• Physio

• Mental health team

• Admin team

• Community service

✔ Books appointments automatically

No phone queue.No refresh-and-repeat website behaviour.No first come, first served.Just clinical need determining access.

✔ Provides self-care advice where appropriate

Fully personalised, with safety-netting.

✔ Produces a clinical summary for the GP

Including symptoms, timeline, red flags, and assumed urgency.

✔ Alerts the practice team if capacity is exceeded

AI can predict demand, highlight pressure days, and redistribute slots dynamically.

This is how you eliminate the 8am scramble, not by stretching receptionists thinner, but by intelligently managing the flow of demand.

AI won’t replace GPs - it will replace the broken model around them

Critics often jump to the conclusion that AI equals job losses or diminished care.That is not the reality.

GPs deal with:

• multi-morbidity

• safeguarding concerns

• differential diagnosis

• clinical risk

• uncertainty

• nuance

• relationship-based care

AI isn’t replacing that. AI is replacing the system pressures that prevent clinicians from doing more of that.

AI can:

• filter out low-complexity queries

• automate admin

• prioritise based on risk

• reduce DNAs

• produce clinical documentation

• manage long-term condition recalls

• handle prescription requests

• communicate follow-up plans

It means GPs spend more time being clinicians and less time being receptionists, administrators, and traffic controllers.

This Is not a dream scenario, it’s already happening in parts of the NHS

Some practices and PCNs are already piloting AI-led triage tools:

• Accurx AI

• PATCHS AI

• AskFirst

• EMIS-enhanced AI triage

• Babylon’s early prototypes (with all their flaws, they still proved the concept)

But adoption is patchy. Culture is resistant. And many systems are held back by legacy infrastructure. The technology is already there. The barrier is implementation.

From a failed bot… to a future without queues

My experience today was frustrating but revealing. A bad chatbot couldn’t help with my power outage. A human stepped in, understood the context, and solved the problem.

But here’s the key insight: An advanced AI system could have solved it just as well as the human, possibly better.

And if AI can handle a dynamic, multi-variable problem like diagnosing a local power fault, then it can certainly handle the daily workflow of general practice.

The 8am scramble is not inevitable. It’s not a law of nature, it’s a symptom of an outdated system waiting to be redesigned.

AI gives us the opportunity to build something better:

• 24/7 access

• intelligent triage

• fairer allocation of appointments

• reduced stress for staff

• safer decision-making

• more time for complex care

• fewer complaints

• a smoother patient experience

For the first time in decades, technology isn’t the bottleneck, it’s the unlock.

Final Thoughts

If today proved anything, it’s that bad bots frustrate people, but good AI can transform entire systems.

The NHS doesn’t need more phone lines, longer queues, or more paperwork. It needs intelligent digital infrastructure that can absorb demand, triage efficiently, and ensure patients get the right care at the right time. AI won’t replace the human touch, it will free humans to deliver it where it matters most.

And if it can do that, then the 8am scramble really could become a thing of the past.

About the author

Michael O’Connor is a partner at Grey Sergeant, specialising in PR, communications, and engagement across the healthcare and non-profit sectors. Through his consultancy Grey Sergeant, he helps healthcare organisations define their brand, strengthen their reputation, and communicate with clarity. For more information, contact michael.oconnor@greysergeant.com