Faster Isn't Enough. Why Ambient AI Needs More Than Speed

Imagine walking into a library or a video store.

Instead of sections for mysteries, science fiction, biographies, or documentaries, everything is organized by format: books over here, DVDs over there, VHS tapes in the back. Within each, it’s chronological by date. If you want a mystery novel or a book on small business? Good luck — unless you already know the exact title and author, you’re out of luck.

That’s what our electronic health records feel like today.

The EHR as a Disorganized Library

EHRs sort clinical data into tabs:

• Notes in one place

• Labs in another

• Imaging reports in another

• A medication list, often alphabetical

• A problem list, often unordered

There’s no unifying schema that connects information by clinical problem. The result is scatter and duplication, forcing clinicians to spend hours sifting through piles of data just to reconstruct the story of a patient.

Every chart is a library — but one organized by media type, not by topic. And the bigger the library gets, the harder it is to use.

The Printing Press Problem

Most ambient scribes today are building a faster printing press. They make it quicker to produce notes. That feels good in the moment: less typing, easier documentation, fewer clicks.

But here’s the catch: faster note-writing doesn’t fix the underlying chaos. In fact, it may make it worse. Faster note-writing is also faster long-note-writing. Without structural reform, scribes just accelerate the flood of disorganized content.

The Hidden Driver of Burnout

Most studies on ambient scribes look at short-term effects: a clinician feels less burnt out today because their notes are done faster. But what happens six months or two years in, when the library has swelled with even more copy-forwarded, context-free data?

The real burden in clinical care isn’t just writing. It’s reviewing — trying to extract meaning from notes, labs, and reports scattered across tabs. Burnout comes from being unable to find context, not just from typing too much.

Do We Really Need a God-Like AI Librarian?

Some propose that we can skip fixing the library and just ask an AI to search it for us. In theory, this works. In practice, it’s fragile and wasteful.

• It’s overkill: asking an AI to reconstruct a coherent story each time, rather than fixing the structure once.

• It risks missing information as volume grows.

• It can hallucinate or omit details.

The simpler, safer solution is to organize the chart properly at the start.

A Better Future: Patient Libraries That Evolve

Each patient’s chart is a library. And libraries need to evolve. The organizational schema should reflect the patient’s actual problems, shifting as their health changes. A 45-year-old with asthma and hypertension has a different chart structure than a 70-year-old with cancer and heart failure.

With modern tools, notes created by scribes or large language models can be organized at the point of creation into problem-based “volumes” that grow with the patient. That’s the kind of documentation clinicians need: not just faster, but smarter — structured in a way that makes context available instantly.

Closing

Without fixing the library, the faster printing press will just bury us. To reduce burnout, improve safety, and help clinicians make informed decisions, we need documentation that is problem-oriented, context-rich, and evolving.

The future of ambient AI isn’t speed. It’s structure.

Next in the series: “If EHRs Were Libraries, You’d Walk Out Frustrated.”