Abridge is one of the most important proof points in clinical AI — Shiv's obsession with accuracy over speed is exactly right. The compliance layer underneath these ambient AI tools is what makes or breaks enterprise deals though. The "is this HIPAA-compliant?" question comes up before any procurement conversation can even start. Really enjoyed this interview — the nuance on ambient AI vs. documentation AI is important.
Sergei. Incredible piece. If you are truly building in AI, there is sooo much info here. An important note that I learned from your piece is that Shiv Rao is a physician. As you have taught us, we need to be skeptics… but, the fact that Rao is a doc is reassuring to me. He is human and not perfect- for sure, but he is one of us. This space needs physician founders badly. Like really really badly. So I am officially an Abridge and Dr Rao supporter… until you give me reason not to be 😂. Hold them accountable Sergei- I know you will! Great job on this piece!
This is a guy with a big vision for a technology that is increasingly studied and being found to have minimal time savings (one study found 42 seconds saved per encounter) and multiple errors (omission, commission, emphasis). They can also increase healthcare cost.
That these technologies have been rolled out so quickly with poor (poor!!!) validation into clinic (including in my own system) and so little feedback from patients is disappointing. Everyone wants to reduce administrative bloat and restore better communication to the bedside. Not sure these are the best way to do it.
Abridge specifically is an AI scribe that was listening in a lawsuit against the Sharp medical system when a patient was not consented about its presence and only found out on reading their note. And they did not (could not?) delete the data off their servers for 30 days despite him asking.
The healthcare arena is one of the toughest areas for an entrepreneur to succeed. Because physicians aren't in control of what software they use. It's under the control of the hospital IT committee or the corporate office. You can spend time trying to build a product that will make physicians' lives better, but ultimately, they are not your customer. The hospital systems are. Or the companies that make their EHR software, like Epic or Oracle. Some hospitals are more relaxed about physician AI use, while others restrict it. Also, the intrusion of AI into the clinic is raising confidentiality concerns with patients, who feel less secure about how private their conversations will be (https://healthexec.com/topics/artificial-intelligence/patients-feeling-less-sure-about-medical-ai-these-days). Even some docs aren't sure that AI scribes are the solution (https://benngooch.substack.com/p/i-was-an-enthusiastic-early-adopter). There have been many EHR apps, but only 2 or 3 really matter today. I think the same will happen with scribe apps. And I fear that as these AI apps strive to take over more of the clinicians' "drudge" work, the cognitive off-loading will detract from the physician's mental investment into the patient care process. Like the doc who couldn't even reconstruct the clinic visit from his own AI-generated chart note.
I agree. We’re still trying to figure it out. But whether it’s Abridge or some other tool, AI (and whatever technology comes next), will inevitably become an integral part of the clinical workflow.
Abridge is one of the most important proof points in clinical AI — Shiv's obsession with accuracy over speed is exactly right. The compliance layer underneath these ambient AI tools is what makes or breaks enterprise deals though. The "is this HIPAA-compliant?" question comes up before any procurement conversation can even start. Really enjoyed this interview — the nuance on ambient AI vs. documentation AI is important.
Sergei. Incredible piece. If you are truly building in AI, there is sooo much info here. An important note that I learned from your piece is that Shiv Rao is a physician. As you have taught us, we need to be skeptics… but, the fact that Rao is a doc is reassuring to me. He is human and not perfect- for sure, but he is one of us. This space needs physician founders badly. Like really really badly. So I am officially an Abridge and Dr Rao supporter… until you give me reason not to be 😂. Hold them accountable Sergei- I know you will! Great job on this piece!
This is a guy with a big vision for a technology that is increasingly studied and being found to have minimal time savings (one study found 42 seconds saved per encounter) and multiple errors (omission, commission, emphasis). They can also increase healthcare cost.
That these technologies have been rolled out so quickly with poor (poor!!!) validation into clinic (including in my own system) and so little feedback from patients is disappointing. Everyone wants to reduce administrative bloat and restore better communication to the bedside. Not sure these are the best way to do it.
Abridge specifically is an AI scribe that was listening in a lawsuit against the Sharp medical system when a patient was not consented about its presence and only found out on reading their note. And they did not (could not?) delete the data off their servers for 30 days despite him asking.
The healthcare arena is one of the toughest areas for an entrepreneur to succeed. Because physicians aren't in control of what software they use. It's under the control of the hospital IT committee or the corporate office. You can spend time trying to build a product that will make physicians' lives better, but ultimately, they are not your customer. The hospital systems are. Or the companies that make their EHR software, like Epic or Oracle. Some hospitals are more relaxed about physician AI use, while others restrict it. Also, the intrusion of AI into the clinic is raising confidentiality concerns with patients, who feel less secure about how private their conversations will be (https://healthexec.com/topics/artificial-intelligence/patients-feeling-less-sure-about-medical-ai-these-days). Even some docs aren't sure that AI scribes are the solution (https://benngooch.substack.com/p/i-was-an-enthusiastic-early-adopter). There have been many EHR apps, but only 2 or 3 really matter today. I think the same will happen with scribe apps. And I fear that as these AI apps strive to take over more of the clinicians' "drudge" work, the cognitive off-loading will detract from the physician's mental investment into the patient care process. Like the doc who couldn't even reconstruct the clinic visit from his own AI-generated chart note.
I agree. We’re still trying to figure it out. But whether it’s Abridge or some other tool, AI (and whatever technology comes next), will inevitably become an integral part of the clinical workflow.
hmm