By early 2024, an estimated 70% of U.S. healthcare organizations were exploring or using generative AI in both clinical and administrative areas, according to a report from McKinsey.
As AI tools increasingly take on administrative and clinical tasks healthcare organizations must rethink how they purchase and implement technology.
During a recent presentation hosted by Becker’s Healthcare, Thomas Kelly, MD, co-founder and CEO of Heidi Health outlined why conventional IT procurement models fail to deliver results with AI and what health systems must do differently to succeed.
Here are four key takeaways:
1. What matters most
Dr. Kelly emphasized that healthcare has long been structured around a limited supply of clinicians. Because physicians undergo extensive training and are relatively few in number, access to their knowledge is tightly constrained.
This results in layered care journeys, where patients go through multiple providers before reaching a specialist. The scarcity also shapes operational realities like long emergency department wait times and delays in primary care access. This structural limitation not only affects patient experience but also increases risk during gaps in care, when critical issues may go unaddressed.
“When you are not seeing a doctor or you’re waiting for investigations to come back, that’s when bad things happen,” Dr. Kelly said. “I’ve seen many patients in emergency who had a terrible complication because they were treated too late.”
By handling documentation, coding, patient follow-up, and care coordination tasks, AI can free clinicians to work at the top of their licenses, or even beyond, if paired with decision-support capabilities.
2. Beyond ambient listening
While many organizations focus narrowly on using AI scribes to generate notes and improve coding, Dr. Kelly urged leaders to take a broader view.
Capturing “clinical memory”— the full spectrum of interactions between patients and clinicians, including conversations, procedures, and care coordination moments, creates a foundational dataset that AI systems will rely on to drive future functions. Achieving this requires consistently transcribing every clinical encounter across all roles, not just for billing or note generation, but to build a comprehensive and actionable data asset.
“It’s not about just focusing on your primary care docs and coding your visits,” Dr. Kelly said “It’s actually how you capture that memory and data in all of the interactions that happen in your organization.”
3. The key to adoption
Traditional health IT prioritizes integration with the electronic health record. But with AI, Dr. Kelly argued, user experience is far more important.
Heidi Health emphasizes customization for each specialty, building hundreds of note templates tailored to specific workflows. That attention to user needs has paid off: Heidi reports 60% to 80% adoption rates, compared with 20% to 40% for more rigidly integrated competitors.
“In this version of the world, AI only works if it’s replacing the work I would do. And that’s a really personal thing. User centricity is the main issue,” Dr. Kelly said. “Make sure that you can get good integration, but don’t treat it as something that is a chicken versus egg integration. You need to make sure they can do it.”
4. A clinician-centered partnership
As AI becomes more capable, the tools that will drive the greatest value are those built around clinician behavior and able to evolve. Dr. Kelly emphasized that AI is truly becoming a partner that learns from a clinician’s workflows, takes actions and connects with patients, just like a human would.
Choosing AI tools that capture data broadly, delight users, and align with a long-term vision for workforce transformation is critical.
“I went to medical school because I had this vision of being this pillar in the community,” Dr. Kelly said. “Reducing the amount of things that take up this scarce time is critical. AI today at the point of care can really influence downstream wherever it’s implemented.”
Watch the full session from Becker’s Digital & AI Virtual event.
At the Becker's 11th Annual IT + Revenue Cycle Conference: The Future of AI & Digital Health, taking place September 14–17 in Chicago, healthcare executives and digital leaders from across the country will come together to explore how AI, interoperability, cybersecurity, and revenue cycle innovation are transforming care delivery, strengthening financial performance, and driving the next era of digital health. Apply for complimentary registration now.
