Don’t Hide the Human Door
Plus: The NHS wants AI to route patients faster. Fair enough. But faster access still needs a person the patient can reach when the machine sends them sideways.
Your daily signal on AI and CX — minus the hype.
DCX Stat of the day: A Sussex general practitioner (GP) trial tied to the National Health Service (NHS) app rollout saw a 29% fall in patients queueing by phone for appointments. The Guardian
In this issue:
→ AI starts routing patients
→ Consent becomes a service moment
→ Scribes test patient trust
→ Heidi shows the workflow layer
→ Travel agents get transaction power
🔍 DEEP DIVE
When the App Becomes the Front Desk
The NHS plans to use AI inside its app to route patients toward the right service: a GP appointment, pharmacy, A&E, or another care path. First-year reach is expected to be about 200,000 patients, with availability for all users planned by April 2028.
On paper, that’s exactly the kind of thing healthcare needs. Fewer phone queues. Less appointment roulette. Better use of clinical capacity. Nobody’s sitting around wishing the 8 a.m. call scramble had more hold music.
Here’s the rub. The first decision in a care journey is often the scariest one. The patient isn’t asking for a slick routing layer. They’re asking, “Am I safe? Where do I go? What if this is wrong?”
So the test isn’t whether AI can move people through the front door faster. It’s whether the patient can understand the recommendation, challenge it, reach a human, and avoid getting penalized for being old, anxious, confused, offline, or just not fluent in app logic.
Bottom Line: AI can make healthcare access better, but only if the recovery path is as visible as the recommendation.
📬 Copy-Paste Take
Before putting AI at the front of care, define the patient escape hatch. Who can override the route? How does the patient know why they were sent there? What happens when the model’s wrong? Speed without a visible human door doesn’t remove the queue. It hides it.
🧭 OPERATOR PLAYBOOK
Write the Wrong-Door Plan First
If AI’s going to direct customers, patients, travelers, shoppers, or account holders, don’t start with the happy path. That’s the demo. Start with the misroute. That’s the business.
Audit every AI-assisted access flow for four things:
What the customer’s told when AI recommends a path.
Who owns the override when the path feels wrong.
What happens to people who can’t or won’t use the digital channel.
How quickly the business corrects a bad recommendation.
Then test whether the customer can recover without starting over.
Ask your team: Where can AI send someone to the wrong place, and who owns the apology plus the fix?
Signal: The front door’s no longer a page, phone number, or intake form. It’s a decision someone has to own.
📊 MARKET REALITY CHECK
The Note Taker Is Part of the Visit
The Guardian Australia reports that use of AI scribes by Australian GPs nearly doubled from 22% in August 2024 to 40% in November 2025, according to an RACGP online poll. That’s not a lab experiment. That’s the third presence in the room.
And look, the pressure is real. Clinicians need time back. Notes need to get finished. Health systems are drowning in admin. If AI can give the clinician more eye contact and less keyboard time, patients may actually feel the benefit.
But consent isn’t fine print here. It’s part of the care experience. Patients need to know when a visit’s being recorded, where the data goes, whether saying no changes access, and who checks the note before it becomes part of their record.
Why it matters: Healthcare AI won’t be judged only by productivity. It’ll be judged by whether patients feel respected when the system listens, records, routes, and stores their most sensitive moments.
Less typing + unclear consent = a trust bill someone else pays.
🧰 TOOL WORTH KNOWING
Heidi Health
What it does: Heidi Health is an AI clinical documentation platform that turns consultations into notes, codes, follow-up tasks, patient communications, and other care workflow outputs.
CX use case: The useful promise is simple: less face-in-screen time, faster follow-up, fewer dropped tasks, and cleaner documentation for the next person in the care chain.
Worth watching because: Tools like this sit right on the consent line. They can make the visit feel more human. They can also make the patient wonder who’s listening, what gets saved, and whether the clinician actually checked the output.
Bottom line: AI documentation becomes a customer experience tool when it gives time back to the visit without making the patient feel managed by a silent recorder.
The DCX AI Today - AI Tool Directory - If you lead a CX team and want a curated shortlist of tools worth evaluating, this is your starting point.
📡 90-SECOND CX RADAR
AI agents can now move from travel search to booking
eDreams ODIGEO says it’s working with Visa to support AI-initiated transactions across its travel brands, using Visa’s Trusted Agent Protocol, Agentic Directory, and Payment Passkeys. That moves agentic travel from “help me compare options” to “book this within my rules.”
Why it matters: Once AI can transact, CX needs agent identity, permission limits, payment recovery, and a clean way to unwind the wrong booking.
✅ YOUR MOVE
AI’s starting to sit at the first decision point in high-trust journeys.
That can make access faster. It can also make the wrong turn harder to challenge.
This week, pick one AI-assisted flow and write the recovery map before you write the success metric.
Check:
What decision can AI make or influence?
What does the customer see when that decision happens?
Where can a human override it?
What happens when the customer says no?
Who owns the fix when the path is wrong?
If AI becomes the front door, don’t make the human door a scavenger hunt.
Until tomorrow,
👥 Share This Issue
Think of one person who’s wrestling with AI in CX right now
and forward this to them.
I’m obsessed with Wispr Flow Pro! Get a Free Month on me.
If someone forwarded this to you, they thought you needed to see it before your next AI planning meeting. Get your own copy.