Show Me the Nurse

Every company in healthcare technology has become an AI company.

Well, every press release has.

The products are taking a little longer.

I have now read hundreds of announcements about AI transforming the patient experience, revolutionizing clinical workflows, and reimagining the future of care delivery.

The slide decks are magnificent.

There are neural networks. There are glowing purple icons. There is always a brain made out of circuit boards, hovering over a hospital for some reason.

Then you ask a simple question.

Can I see it running in a hospital?

And suddenly everyone needs to check with the product team.

The demo of the demo

Here is a pattern I keep seeing.

A company announces an AI platform. The announcement wins an award. The award appears in the next announcement.

The product is in “limited early access.”

The waitlist is open.

The general release is coming in 2027.

The company has raised another round of funding based on the waitlist for the product that does not exist yet.

I want to be clear that I am not against ambition. I am against announcing the destination and skipping the trip.

Building real things is slow and unglamorous. We just spent four days on an Epic integration we thought would take one, because Epic is very specific about what gets written into a patient’s chart. And it should be.

That is what building actually looks like.

It does not photograph well.

There is no hovering brain.

The innovation lab has excellent coffee

Then there is the innovation economy that has grown up around all of this.

Health systems are spending millions of dollars on innovation labs. I have visited a few. They are beautiful. Glass walls. Writable surfaces. Excellent coffee. Furniture that costs more than a nurse makes in a year.

Here is what many of them produce:

A vision document.

A pilot of a prototype.

A steering committee for the pilot.

A consultant’s report recommending phase two.

And phase two, conveniently, requires more consultants.

That is the part nobody says out loud. The consultant class has discovered that innovation pays best when it never finishes. A deployed product can be measured. A roadmap cannot. So the roadmap gets extended, the lab gets renewed, and everyone’s interests are served except the hospital’s.

Meanwhile, three floors up, a nurse is documenting the same conversation twice because nobody connected two systems that have both existed since 2008.

That connection is not innovative enough for the lab.

It would merely help.

We have even lost to our own ideas

Here is my favorite version of this phenomenon.

I cannot tell you how many times we have lost a deal to an EHR product that looks remarkably like something we built ten years ago.

The digital whiteboard. The patient room integration. The rounding workflow.

We built it, deployed it, and spent a decade refining it inside real hospitals. Then the EHR vendor releases its version, and the hospital chooses it because it is already part of the platform.

I understand the logic. Nobody ever gets fired for buying more of the system they already own.

Sometimes they explain it differently. They tell me the EMR version is going to cost them less money.

I always ask the same question.

“Has your bill ever gone down?”

Nobody has ever said yes.

There is usually a pause, and the pause says everything.

Sometimes they go one step further and tell me the software is actually free.

Free is an interesting word.

The free software just happens to run on a $1,000 tablet PC mounted in every room.

The tablet lasts about six months in a hospital.

Multiply that by a few hundred rooms, and the free software has one of the largest hardware bills in the building. It just shows up on a different budget line, where apparently nobody checks.

The module is free the way a puppy is free.

But think about what actually happened in these deals.

Ten years of refinement lost to twelve months of bundling.

The product that shaped the category lost to the category’s newest arrival, because the newest arrival came with familiar login credentials.

I have decided to treat this as market validation.

Very expensive market validation.

It is also exactly why the new HCI exists. Being first is not enough. Being better is not enough. Being deployed is not enough. If we cannot tell the story as well as we build the product, someone bigger will eventually tell it for us.

Not this time.

Our standard

At HCI, we have exactly one test for all of this, including our own work.

Show me the nurse.

Not the deck. Not the lab. Not the committee. Not the announcement.

Show me one nurse who finished a shift with less unnecessary work because of what you built.

If you can do that, you are innovating. If you cannot, you are marketing.

And look, we have AI too. This blog has almost certainly been within ten feet of AI. We use it every day, and it is genuinely making our small company move faster than I ever imagined.

We have also made our own mistakes. We have moved too fast, promised too much, and built a backlog that occasionally resembled an idea buffet. I am not throwing stones from a glass innovation lab.

The difference is where the work ends up.

Ours ends up in patient rooms. Real hospitals, real nurses, real televisions with real pencils broken off in the earphone jacks.

That is the whole strategy.

Build things that work. Install them where people are. Fix what breaks. Repeat.

Nobody is going to give that an innovation award.

We will be upstairs anyway, plugging the television back in.

See HCI ONE in a Real Hospital