The AI You Design This Year Will Define How Patients Experience Care

The AI you design this year will shape how patients experience care for the next decade. That is a large claim for a technology most teams are still piloting, and I believe it anyway, because the choices being made right now, about what to build first and whose problem to solve, tend to harden into how the system works for years afterward.

Most AI in healthcare starts with the system rather than the person, and that is where it goes wrong. The first question in the room is usually some version of what the technology can do. A better first question is how this will feel to a patient in the moment they actually need it. Those are not the same question, and the order matters far more than it looks.

Most Patients Meet Us on a Bad Day

People do not arrive when things are going well. They arrive when something is already wrong, often frightened, often in a hurry. In that moment the accuracy of a model is not what they carry home. What they remember is whether the care felt human, or whether the technology in the room made them feel like a case number. Build for the system first and you optimize the part the patient never sees, while the part they do feel goes unattended.

The Promise Is Not the Mission Statement

Almost every health system already says it puts patients first. The words are on the wall and in the strategy deck, and the belief behind them is usually real. The gap is not belief. It is operations. A commitment to center the patient does not decide which problem an AI project takes on first, who sits in the design review, or what the dashboard ends up measuring. Those choices are where the mission becomes real or stays a slogan, and they are most often made by people looking at the system rather than at the person who will live with the result.

From Infrastructure to Impact

I did not start here. I spent the early part of my career in data warehousing, system integration, and analytics architecture, building the pipes that move insight across complicated environments. That work matters, and I still believe in doing it well. What changed for me was watching elegant architecture deliver almost nothing when it did not reduce friction for the people using it. A clean pipeline that makes a clinician click three more times has not helped anyone. Over time my attention moved from how the system performs to how the work feels for the person inside it, because that is where the value either shows up or quietly disappears.

Design for the Operator, Not Just the User

The most useful idea I have come across on this came from a caregiver I spoke with on The Signal Room. She made a distinction that has stayed with me. Families do not simply use the healthcare system. They operate it. A parent coordinating care for a child with a complex condition is running that process every day, across specialists who each work in their own lane, and she can point to exactly where it breaks, because she is the one absorbing the break.

She gave an example that sounds small and is not. She asked for a transfer of care, and the system read it as a request for a second opinion, two phrases that route to very different places. The wrong reading put her on a slower path and cost weeks. The fix, she learned only later, was a single direct call between two physicians. Nothing in the process was built to tell her that, because the process was designed around specialties and handoffs rather than around the person trying to move through it.

That is the part worth sitting with. If we build AI on top of a system shaped that way, we do not repair the blind spot. We automate it and scale it. The same gap that cost one family three weeks becomes a default, applied faster and to more people, and much harder to see once it is buried in software.

What People-Centered AI Asks Of Us

None of this is an argument against the technology. It is an argument about where to point it. The AI worth building gives time back to clinicians and patients instead of adding another screen to check. It helps the people in the room listen better rather than process faster. It treats the reduction of friction as the goal and not a happy side effect.

Getting there is less a technical problem than a design one, and design starts with who is in the room. The people who live where care happens, the caregivers, the coordinators, the front desk staff, the clinician working at two in the morning, hold knowledge a design team cannot get any other way. They know which step is redundant and which alert everyone has already learned to ignore. Bring them in at the start and the tool reflects the real work. Leave them out and you get something that demonstrates well and frustrates everyone who later has to use it.

What we measure has to change too. If the only thing we count is throughput, we will build tools that move people faster through a process that was already failing them. The harder questions are whether the burden of delivering care actually went down, and whether a patient left the encounter feeling seen. Those are not soft concerns. They are the difference between a tool people trust and one they quietly work around.

Why This Is Worth Getting Right

This is not about loving technology or fearing it. It is about what is at stake when we get it wrong, and what becomes possible when we get it right. Technology will not replace the human experience of care. It can make that experience worse, or it can make it better, and the design choices being made this year are what tip it one way or the other.

The systems we build now will decide how people meet healthcare on the days they are most vulnerable. That is not only a technical decision. It is an ethical one, and it is worth treating like one.

Christopher Hutchins Founder and CEO, Hutchins Data Strategy Consultants