According to Forbes, Dr. Michael Pfeffer has served as the Chief Information Digital Officer at Stanford Health Care and an Associate Dean at the Stanford University School of Medicine since 2021, all while maintaining his role as a practicing physician and clinical professor. He described Stanford’s network, which includes the main campus, Tri Valley Hospital, and hundreds of clinics, as “quite an amazing organization.” His core mission is closing the gap between clinical needs and tech systems, moving beyond the “digitized” era of electronic records into truly “digital” care processes. A key project enabling this is ChatEHR, an integrated large language model tool embedded directly in the electronic health record to help with tasks like admission summaries and discharge notes. He also highlighted Stanford’s Secure GPT environment for safe experimentation and the FIRM framework for ensuring AI models are fair, useful, and reliable.
The Human Translator In The Machine
Here’s the thing about most health tech leaders: they don’t see patients on Tuesday. Pfeffer does. And that’s not just a cute detail—it’s the entire thesis. His value, as he sees it, is being a translator between two worlds that have historically spoken different languages: clinical ops and IT. But he thinks that chasm is closing because tech is just everywhere now. It’s not special anymore; it’s the water we swim in.
So when he talks about “moving from digitized to digital,” he’s pointing out a failure we all felt. Putting paper charts on a screen was step one. A necessary, painful, expensive step. But it didn’t make anything better. It often just made the old, clunky processes faster. The digital step is the hard part: ripping up those bad processes and designing new ones where the tech actually reduces friction. The goal isn’t to boast about AI adoption rates. It’s “bringing people together around a shared goal of really improving the patient experience.” And, crucially, the clinician experience too. You can’t have one without the other.
ChatEHR And The Seconds That Matter
The proof is in the pressure points. Pfeffer talks about his own work as a hospitalist. Admission and discharge. That’s where the time gets sucked into a vortex of chart review and note-writing. With ChatEHR—which pulls real-time patient data into an LLM right inside the EHR—he says you can get a summary in seconds. Not minutes. Seconds.
That’s the shift. It’s not about a flashy chatbot. It’s about embedding the intelligence exactly where the clinician already is, so they don’t have to open another tab, log into another system, or copy-paste a single thing. The system learned this because they built Secure GPT first, a sandbox for staff to play in. And the signal was deafening: “Put it in the EHR, or don’t bother.” Seems obvious, right? But how many tech projects ignore that basic user truth?
Guardrails And The Un-Automatable
Of course, this is healthcare. You can’t just move fast and break things. “Healthcare is slow to change because it should be,” Pfeffer notes. You can’t really make mistakes. That’s why he points to the FIRM framework—Fair, Useful, Reliable Models—as a non-negotiable part of the lifecycle. The goal is to avoid “sophisticated” automation that just adds new risks.
But for all the tech, he draws a bright red line at the human connection. “I think the human-to-human connection is never going to be replaced by AI,” he says. That’s the whole point, in his view. If AI succeeds, it gives clinicians back the minutes and hours they need for empathy, for nuance, for dealing with the scary, messy reality of illness. Care isn’t data transfer. It’s a relationship. The tech should serve that, not the other way around.
Rethinking Bots Fighting Bots
Looking ahead, Pfeffer’s vision gets more interesting. He’s not excited about using AI to just turbocharge our broken systems. He challenges the dystopian idea of “bots fighting with bots” in prior authorization battles. Instead, why can’t evidence at the point of care make the right path obvious from the start? “We should be reducing the cost of health care with AI, not increasing it,” he argues. That’s a radical thought.
It will require redesigning processes, sure. But also policy change and real governance. Living in Silicon Valley, the hype is inescapable. “It’s the talk of the town.” But he stresses partnering with folks who see healthcare as more than a market. His final prediction? “If we’re practicing medicine the same way, five years from now, I’d be shocked.” The pace of AI-augmented discovery, he thinks, might outrun even operational change. The future isn’t just digital notes. It’s a fundamental rethink of what’s possible. You can follow the conversation with thinkers like Pfeffer on platforms like the Forum on World Class IT or via analysts like Peter High on X.
