Not another AI article…just a quick realization check

Our entire industry of Healthcare IT is racing to sprinkle “AI” on everything in sight. Much like hard to fill consulting staffing needs, this is the tech world’s version of finding a unicorn. But in reality, a lot of what’s being marketed as artificial intelligence, in the various EHR systems today, might just be closer to automation vs sentient software.

Epic has embraced the AI label across its platform, from clinician assistants to patient-facing tools. At their 2025 Users Group Meeting, the company highlighted hundreds of new AI features, and at HIMSS demos, some 60+ new products, including Emmie, ART, and Penny, that were all aimed at helping clinicians with charting, documentation, and administrative tasks.

The reality? Most of these capabilities resemble what many refer to context aware automation, including:

• drafting prior authorization responses based on structured data
• suggesting codes after a visit
• summarizing chart sections for easier workflows.

These are truly just workflow automation, not the much hyped AI future of technology releases that we read about every day. These are all still ultimately guided by human decisions and rely on existing clinical processes, not independent clinical judgment.

Check out this great Epic article on CNBC.

Since Oracle acquired Cerner, the messaging has leaned heavily into new AI driven capabilities. The new Oracle Health EHR platform is described as “AI-driven intelligence” built into workflows, able to surface insights, handle voice-activated commands, and trim documentation burden.

This fall the press releases and news articles were endless, including this fun read Oracle Ushers in New Era of AI-Driven Electronic Health Records

But a closer look shows most of this is still just automation, including:

• converting speech to text and structuring notes,
• pulling up relevant chart data on keyword requests,
• auto-suggesting orders or navigation clicks.

These are valuable productivity enhancements, which everyone can benefit from, but they don’t think or diagnose on their own. They automate repetitive tasks that clinicians used to do manually.

There are genuinely some AI-centric tools that I’ve seen, including systems that analyze medical images, flag deterioration risk patterns, or predict clinical outcomes based on massive datasets. Check out Nuance/Microsoft, which partners with both Epic and Oracle. But let’s all be clear and recognize that most EHR-embedded features today are task automation with AI branding that tout age old promises of reducing clicks (ever hear that one before?), creates draft texts (been around for years), and directly improves workflows.

Thoughts on how “AI” is driving your optimization project support efforts? Share your comments here.