# From 10 minutes to 2: how AdventHealth uses ChatGPT for Healthcare to give clinicians time back!

Hey everyone, it's Shii-chan! Today I found a heartwarming story about how AI changed the way a hospital works, and I really want to share it with you!

## What was announced?

OpenAI's News published a story about how AdventHealth, a big hospital system, rolled out ChatGPT for Healthcare across its whole organization. AdventHealth runs hospitals in 9 US states and cares for millions of patients every year. The goal is to cut administrative burden and smooth out clinical workflows.

## The story so far

Big hospital systems face tight margins, rising demand, and growing administrative complexity. For example, physician advisors reviewing cases for utilization management used to spend about 10 minutes per case, reading charts, pulling out the relevant details, checking criteria, and drafting structured rationales. Multiply that by hundreds or thousands of cases and the time adds up fast.

Teams in finance, HR, and IT were stuck in what leaders call "constant operations mode," buried in drafting and summarizing. Inside the company more and more people wanted to try AI, but policies held them back, or they simply didn't know how to use it well.

## What changes

The thing AdventHealth cared about most was treating adoption itself as the product. They decided early that a few small pilots wouldn't move the needle; the real goal was getting a large workforce to use AI safely and consistently.

And the keyword is "time back." They framed AI not as automation, but as a way to return time to clinical staff.

> We don't talk about AI as automation. We talk about time back.

If you can shorten a 10-minute review while keeping quality, that time becomes capacity for care.

## Dive Deep

The most measurable use case was utilization management. With ChatGPT for Healthcare, physician advisors can generate structured summaries of charts, surface relevant clinical details, and draft initial rationales. The clinician still makes the final call, so AI just handles the prep work.

What I love is how they measure impact: not self-reported estimates, but real data baked into the workflow, like electronic health record (EHR) timestamps. They can see exactly how many minutes improved and whether the change is statistically significant.

Their rollout was clever too. Instead of big centralized training, they used domain-based peer groups. Finance worked with finance, HR with HR, sharing prompts, workflows, and best practices. They even track usage as "messages per user per business day" and manage it like a KPI.

Beyond utilization management, teams now start from a first draft instead of a blank page, turn policies and communications into usable formats, and summarize notes into action steps.

> The hardest part of AI in healthcare is getting humans to use it safely, consistently, and at scale.

There's even a story about a physician who used to do paperwork at home in the evenings, called "pajama time," and can now finish during regular hours and be present with family.

## Wrap-up

- AdventHealth expanded from ChatGPT Enterprise to ChatGPT for Healthcare (source: OpenAI's News)
- The keys: "treat adoption as the product" and "AI is time back, not automation"
- Impact is measured with real data like EHR timestamps; usage is managed as a KPI via messages per user
- In utilization management, 10-minute reviews got shorter while quality held, and the pattern spread to finance, HR, and IT
- Next up: patient access, clinical decision support, and new care delivery models

This story is less about the tech and more about how you make it stick. I think it really lands for leaders trying to scale AI across a big org, and for anyone working on digital transformation in healthcare!