Inside Abridge: The AI Listening to 100 Million Doctor Visits — Abridge's Janie Lee & Chai Asawa

The first and most important thing is context is everything, as Chai alluded to. And I also think about how do we go from being reactive alerting to really proactive intelligence at the point at which it matters most. One thing we like to say is we want our product to feel like air conditioning.

It should be in the background just making things better. And maybe if and if there is something that has great clinical risk and we're acutely aware that intervening now and not later is incredibly important, we should decide to act.

Before we get into today's episode, I just have a small message for listeners. Thank you. We would not be able to bring you the AI engineering, science, and entertainment content that you so clearly want if you didn't choose to also click in and tune into our content.

We've been approached by sponsors on an almost daily basis, but fortunately, enough of you actually subscribe to us to keep all this sustainable without ads, and we want to keep it that way. But I just have one favor to ask all of you.

The single most powerful, completely free thing you can do is to click that subscribe button. It's the only thing I'll ever ask of you, and it means absolutely everything to me and my team that work so hard to bring The In-Space to you each and every week.

If you do it, I promise you we'll never stop working to make this show even better. Now let's get into it.

Okay. This is a special crossover Late in Space, Unsupervised Learning pod.

Very, very excited to do this. Like, you know, once a year at this point we get together and-

Once a year

... and this is a fun occasion to get to do it on.

Uh, we-- I really wanted to talk to Abridge, uh, but I felt very underqualified because it's-- healthcare is not something we cover very intensely, and it just so happens that Redpoint's our big, big investors and supporters of, uh, Abridge, so-

Anytime you wanna have a portfolio company-

Yeah.

... on, on your podcast, please, by all means.

So we'll introduce our guests. Um, Chai and Janie, welcome to the pod.

Thanks for having us.

Thank you.

We're excited to be here.

Uh-

Thank you.

Yeah, so for, for listeners, uh, what do you guys do, just to situate you guys in a, in a company?

Uh, Abridge is a clinical intelligence layer for health systems. We really started with documentation and building for clinicians, and we think that, you know, as we think about reducing the burden that clinicians have, they're spending ten to twenty hours a week on documentation.

There's a massive doctor shortage in the country. We also think that conversations between patients and clinicians are probably the most important workflow in healthcare. It's obviously where care is given and received, but if you think about the twenty percent of our GDP that goes towards healthcare, almost everything is a derivative of that conversation, whether it's the claim, the payment, the actual diagnoses given, the treatment.

And we've started with a conversation to reduce the burden for doctors on documentation, but we're really excited about the path ahead as we become this broader clinical intelligence layer.

I'm Chai. I work on clinical decision support at Abridge.

Yes.

And so I think a- as Janie said, that we had this-- we're, we're uniquely situated where we started off with the clinical note. What I'm really excited about and where we're expanding towards is what are all the things you can do before the conversation, during the conversation, and after the conversation if you did have access to all the context about patients, payer guidelines, medical literature, and put that together and to serve, you know, what-- how healthcare could look fundamentally different.

Yeah. And that's, like, the context engine that you guys have?

Yes.

Is that what it's called? Yeah, okay.

Mm-hmm.

Uh, so histo- historically, as I understand it, company started in twenty eighteen. Uh, a lot of people would be familiar with, like, the AI voice notes form factor that p- that doctors would be like, "Well, do you consent to be- being recorded?"

It replaces handwriting and what have you. Uh, but I-- it, it sounds like more recently there's been a big transition in the company. Or again, just tell me about, like, the, the broader transition.

Yeah. So from a transition perspective, we really think about our journey as how do we... You know, first chapter was-- first act was how do we help save time, and that's where a lot of that original product was.

Yeah. Which like, b- by the way, one of those interesting stats-

Yeah

... on your landing page was, like, people spend-- doctors spend, like, time after hours-

They call it pajama time.

Okay. Why is that pajama time?

Uh, doctors after work in their pajamas-

In their pajamas. Oh

... at home are just writing and catching up on their notes every day.

Yeah.

And, you know, we think some of our favorite customer love stories, we have a Slack channel called Love Stories. We have clinicians telling us, "Abridge has helped us, you know, from retiring early," or, "We're now finally able to- " ...

go home and eat dinner with our kids for the first time." And I think-

Save the marriage in some cases.

Yeah.

Yeah. One of the quotes was like-

Yeah

... "We're not divorcing anymore."

Yeah.

And I'm like, "Why?" Like, 'cause you're working too much, I guess. Yeah.

But, um, in terms of where we're going and where we're expanding, we really think about our second and third acts around how do we help health systems save and make more money. Health systems are operating with, you know, record low operating margins.

It's getting harder and harder to serve patients, and they have regulatory some tailwinds, but also a lot of headwinds coming their way, and we think AI is ripe for helping on the saving and make more money piece. And then ultimately, how do we help save lives?

The fact that our software and our product is open millions of times a week before, during, and after a patient walks in the room, um, gives us massive opportunity with products like clinical decision support, which Chai is building, but so many others to actually improve patient outcomes, and probably one of the most important workflows and, and problems to be going after right now.

I mean, I think one thing that's, that's interesting, Chai, is obviously you came over to Abridge from Glean. When I think about clinical decision support, uh, which is, you know, for our listeners, is basically, you know, in the context of a visit, helping a doctor figure out the right type of care.

It's really a search problem in many ways, right? Of, of going through lots of different data sources. Very analogous to your previous role as, as, as one of the, uh, earliest engineers over at Glean. Um, I'm sure a lot of our listeners are curious what's, uh, similar about the problems that you're going after now and what feels different, uh, now that you're, you're in healthcare.

Yeah. Um, very similar. And I, I think taking a step back, I think with every wave, there's a lot of, like, very similar patterns that happen across different products. A lot of social networking products with the same, a lot of, like, credit-based products with the same.

And I think we're seeing that very similar in the agent era with many companies, of course, in Redpoint's portfolio and so forth. Um, and the key insight between both companies is that, like You have amazing models, but like context is king.

Context is what actually puts them to work. Um, so I see it in a lot of ways, a lot of similarities in like this is a healthcare-coded version of Glean, but I think the differences are really interesting. A couple things that come to mind.

First and foremost, uh, like the rigor at which-- in which in, in the setting we are in. Um, the downside risk is extremely high here in healthcare. It can actually be fatal in some cases. You pr-uh, prescribe something that the patient is allergic to, for example.

Whereas at Glean, it's like, "Oh, you got the question wrong." It, it wasn't the end of the world in most, uh, most cases. And so what does that mean? That shapes, uh, our evaluation strategy, both offline evaluation, progressive rollout, and there's a lot more we could kind of go into there.

Second thing that comes to mind is like vertical versus horizontal. Um, in both cases, there's, there's a large variance. But when Glean is, I mean, it's a much more horizontal company, there's a v-var-variance of personas, companies that you're working with.

Um, we also have a variance of, uh, personas, different types of specialties, different hospital systems. But the variance is a little n-more narrow. So from a product perspective, you're able to focus far more, especially when you have a maturing technology and you're building new products that never existed before.

It lets you go specific-- uh, go after them much more easily. And especially in healthcare where so many problems have-- were solved with labor and process, that's actually extremely ripe for AI to keep helping augment and enable. Um, and then the final thing that I think that's really interesting, Abridge specifically compared to many other companies in the AI area, is the modality we started with.

We're, we're ambient, and we're always listening in the background. And I think many more AI products will go that way, but it's actually how we, uh, started. And, and I think that's actually the, the, the greatest form of AI we can create, AI that's actually seamless.

You're not actually looking at your screen. It's al-always there. It's always helping you out and being proactive. You know, the Jarvis vision that like every hackathon I went to over the past decade, there was always a Jarvis competitor.

But I actually think Abridge very much started from the opportunity and continues to go that way.

Yeah. And one thing I think is super interesting then from a product perspective is you have this always on, seamless in the background, and then you have to decide, like, when do you kind of break the wall almost and, like, say, "Hey, you know, uh, clinician, like you might not have thought about X," or, or whatever it is that you wanna do.

Um, and obviously, I think in, in healthcare traditionally, there's been this idea of alert fatigue and just like a million pop-ups, and then a-- and doctor just ignores all of them. I-it's probably a pattern that a lot of builders are thinking through now.

How do you think about, like, the right way to, uh, to intervene or to pop up in, in a doctor visit?

Yeah, it's such a good question. I think alerts are notorious in healthcare specifically. I think over ninety percent of alerts are ignored. I think the first and most important thing is context is everything, as Chai alluded to, and I also think about how do we go from being reactive alerting to really proactive intelligence at the point at which it matters most.

One thing we like to say is we want our product to feel like air conditioning. It should be in the background just making things better. And maybe if and if there is something that has great clinical risk, and we're acutely aware that intervening now and not later is incredibly important, we should decide to act.

But I think if you think about proactive versus reactive, instead of alerting a clinician during a visit when they're with their patient having a pretty serious and sensitive conversation, how do we actually prep a clinician before they walk into the room with that patient?

And so historically, clinicians might have to manually go through charts with a patient that they've had over the course of months or years, and they'll try to suss out what are the things they should be doing. You can imagine a world with Abridge.

We'll summarize all of the most recent context for you, tell you based on the reason for a visit the patient is coming in for, the types of things you should be discussing. And so you're actually going into that conversation prepped rather than walking in cold to that patient visit and then having this product interrupt you five or ten times throughout the, the visit.

And there might actually be times where it's really important to interrupt. We have a product called, uh, prior authorization, and so this is when you may go into a doctor's office with knee pain. They'll prescribe you an MRI, and I think so many of us have had this experience before, where in four weeks you'll get a call saying, "Hey, Sean, that MRI that you were prescribed wasn't approved.

And why don't you come back in? We'll figure it out." In a world with Abridge, we might choose to actually quietly but, um, still alert a doctor in that visit. And a-alert is probably not even the word we would wanna use.

Before a patient leaves, we would wanna tell the doctor, "Hey, doctor, before Sean leaves, you should ask him, has he had physical therapy, and has his pain lasted for more than six weeks?" Because the- ... Aetna plan that he's on in California requires six things.

We've already confirmed four of them have been met 'cause we have all the context, but these two last criteria, if you can address with Sean before he leaves the room, we could actually guarantee that your MRI is approved before you leave.

And so when you think about clinical usefulness, um, impact to the patient, I think there are instances in which if we can catch a doctor while the patient is still in the room, um, you know, as we think about save time, save money, save lives, we kind of get to check all of those boxes.

But, um, when, you know, doctors have 15 minutes between visits, we have to be really, really thoughtful about when it actually matters.

I think there's this interesting product opportunity that AI can have is reduce latency in the world. Um, for example, prior authorization is an example of where care gets delayed, and so great AI can reduce that. And I think the problem with alerts before partially is a technical problem.

Like it's a-- the quality of your alerts really matters. They're gonna get ignored if you get alerts that, similarly in engineering, where they're noisy alerts that you can't act on. But if you can make really high quality alerts with, with both the context, as Janie said, and really high quality models, then I think you can create a whole 'nother game.

Yeah. And I really like that experience because I think it kind of starts to tease apart like what makes this so hard and unique. I think like, one, to make that prior authorization example possible, think about all the data that you need to have.

You need to integrate within th- with electronic health record to know all of the patient context. Do we have access to your previous labs, previous imaging? And then to actually match you and to know that you're on Aetna, we have to collect all of the different-

Interesting

... payer policies, and they vary by state. Some of these payer policies live on websites. Some of them live in unstructured 50-page PDF files.

I thought this episode was to make-

Yeah

... sure we didn't scare people from healthcare, but, uh

But it's-- I think when you think about the things that make it hard, it also gives you the moat.

Totally.

And then I think the second is the, uh, AI and the model quality we need to h- be able to hang our hat on. And so the bar, I think similarly, when I worked at Opendoor, I worked on pricing models.

Like every outlier wiped out the margins of 30. And so similarly here in healthcare, the, the bar for accuracy is so high. And then I'd say the last is workflow is everything. You know, if insurance companies deploy AI, it typically happens too late, and this is when you have the notorious kind of like comical examples of AI just fighting each other when it's too late.

But if we can pull forward the use of both the AI, but also the ability to solve problems when the patient's in the room, you can start to collapse what typically takes weeks or months after your visit, um, ideally down to, to minutes or real time.

And I think, um, it's, it's where healthcare is both very difficult but also extremely rewarding if you can crack it.

Just to get some baseline on the form factors, because I, I've seen some videos on, on your website and stuff. Uh, wh- you guys talk a lot about ambient, um, AI. Uh, is it primarily on the phone? Is there any other form factor that people get Abridge in?

Like, is there like Abridge room setup where it's just like always on, like, I don't know.

Abridge podcast studio.

Yeah.

Right.

Primary form factor is mobile and desktop. Usually-

Yeah

... clinicians are walking in and out of rooms with mobile, but at the end of the day, when they're closing out their notes or wanting to prep for the day ahead, they might use desktop. We have been having a lot of, um, really interesting partnership conversations with a lot of these in-room device companies as you think about what is, you know, the power of multimodality and even more data, as you think about all of the what is today not captured context.

Um, really, really fascinating to think about, especially even as we go into, uh, building and scaling our nursing product. It's one where nurses constantly, you know, as they're walking in to check in on a patient for two minutes or maybe even 30 seconds, starting an Abridge, uh, experience is probably gonna take longer than the visit.

And so what can we do with in-room devices that are always on-

Right

... starts to beg really, really interesting and, and fun product questions.

I was thinking like the way in, in tech companies we have all these Google Meet-

Yep

... and other things, we might as well set up entire rooms with just Abridge tech.

Very much, and, and I also think similarly about like actually also AR glasses and so forth, I think is also quite relevant, where part of it is how do we bring it in a way without like a screen, but like also bring the information to the clinician in real time, but also let them focus on the patient.

Do you think they want that? I'm, I'm just like very s- I, I'm tend to be skeptical AR, but you know, I'm curious what you've tried.

You know, admittedly, it's not a near-term product roadmap-

Yeah, yeah, yeah

... by any means, and I'm here being far-fetched.

There's some, there's some sick AR stuff for surgeries, actually.

Uh-

Really?

When people are trying to visualize like the, you know, you're about to make an incision, but you wanna see like what the cut might look or what the body might look like inside, and they can basically layer in imaging.

Uh...

Yeah, that's cool.

Yeah.

Yeah.

Some point in the future.

But there are a lot of, uh, our largest customers and, a- and at the largest health systems integrating already. And so even as we think about building into it, I think unlocks a lot of product capabilities.

Yeah. A- and just to establish the terminology, uh, sorry, and, and I know I'm, like I'm asking basic questions somewhat for myself, but also for the audience who might be-

Yeah

... less integrated. H- when you say health systems, it's like the Johns Hopkins, the Kaiser Permanentes-

Mayos, the Kaisers of the world

... these are your customers, right? And the outcome that you de-deliver for them is happier doctors, uh, reduce like whatever cost, like I guess of like, of processing, reduce mistakes. I think it's, it's, it's weird in a sense that- I feel like there's also, like, a secondary customer, like the customer of the customer, and I don't know if you-- Do you think about it that way?

Yeah. We have-- I think the other interesting and complex part of, uh, building product is we have our buyers, who are the chief medical information officers-

Mm

... the chief financial officers, the CIOs of these large health systems. Our users today are clinicians, but if you think about who downstream is impacted, it's patients. And so as we build, um, with every product in mind, we think about who are we building for, who's the secondary user, and what does that mean either in terms of experience, security and compliance, uh, ROI that we have to make tangible.

And, and so like you said, like, time savings is one of them. But for CFOs, they care a lot more than just time savings. We have to show for every dollar you put into Abridge, because you have more compliant documentation or because you have fewer queries coming from your billing team, um, we actually save or add real dollars, um, to your bottom line or top line, I think are things that we're constantly thinking about because of the, the dynamic across all three sets of users.

And I think there's a whole another axis too with the, with the payers and pharma-

Yeah

... as well, and, like, connecting all these three big stakeholders in healthcare is-

Yep.

Do the payers ev-ever see your data? Sorry, the payers meaning the insurers, right?

Mm-hmm. Yes.

They, they also see Abridge data?

Uh, n- uh, no-

I think direct to the patient, yes

... um, they, they wouldn't see the, the raw Abridge data, but where when you're working together on something like prior authorization, whatever information they need, we'd w- or we'd communicate to them.

Oh, right. Yeah, cool.

Yeah.

That's cool. I guess, you know, would love to dig into just obviously you have, uh, you still have to solve a lot of problems on the AI side. And so maybe to start at the highest level, what's one of the hardest problems you have to solve in AI at Abridge today?

To make things simple, let, let's take, like, building off the prior auth example. So one thing Janie talked about is, like, okay, this data is all over the place, and there's this combinatorial explosion of, like, uh, procedures, uh, payer policies, and even sometimes different health systems.

There can be some cross-product of all of these different considerations you have to take into account. But what's really, really, really hard about this problem is actually doing it real-time in the conversation. So, you know, in any AI product, usually the three KPIs you care about are quality, uh, latency, and cost.

Now, what we're saying is we want you to do this real-time in the conversation, guiding the clinician. How do we do it in a way that does not break the bank? Um, but we're using-- But we also need very intelligent models because you're working with this cross-product of data and this, like, all this context layer as well.

So you need high intelligence and high quality because you don't want the alert fatigue, but you also need to be fast and cost-effective. And so that's where I think a lot of clever engineering goes, actually. It's like, okay, without getting into all the details here, can you model these policies in, uh, some intermediate representation or other things that you can do that can actually make this problem tractable?

Um, and of course, the parameter frontier is always changing, but we're also trying to do this now.

Yeah. What implications has that, has that had for what you take off the shelf, um, and say, "You know what? We don't need to be world-class at X. We'll just take this from the model providers or from some infrastructure player," and what you're like, "No, this is where we spend most of our time focused on"?

Yeah. Um, th-this is, uh, the fun challenge in AI, right?

It changes every three months, right? So

Of course, with the, with the, with the shifting landscape. We try to be extremely thoughtful on predicting the trends of where third-party models are going and where we can uniquely go. Um, and you know, sometimes I feel like when we-- when you talk about AI models, we're like, "The models are just gonna get infinitely better."

But I don't think-- Maybe in the grandness of time you could say that, but actually, within every month, every quarter, there's specific ways they're getting better. You know, they're training on a lot more, uh, coding data to be better coding agents, for example.

Um, and so we have to think about where are the things that won't-- unique data that we're uniquely training on, um-- Or actually, to step back a little, like, where is a proprietary model bringing advantage to us is if it can give higher quality or lower cost and latency for similar quality, very similar to many other companies.

Um, and when we can do that is when we have proprietary data. So for example, we have on the order of eighty million or hundreds of millions, actually, now getting close to, of medical conversations.

That's insane.

This is a unique data set, and this data set, it's, it's very interesting because this data set is effectively a large part of the trace between the patient and the provider. That's where the quote-unquote debugging happens in healthcare.

We actually have these traces at scale, as in, like, as, uh, our CEOs even called it, an exhaust that comes out of our product. And so when you have these traces, that's how you can actually train better agents on certain use cases, um, whether it's your transcription diarization use cases or, or so on, um, or, like, note generation models, and we can do that much cheaper and faster.

But we're always also working with these third-party model providers. You know, we, we closely collaborate with them, and that's how we kind of predict where the trends are go. The thing that I think about a lot is that, um, I know that the model providers are gonna train much more on, like, agentic workflows and so forth.

So that's great, so that you have a better agentic harness. But the other thing that's interesting is you know that the model providers, because a large class of the consumer model providers is healthcare queries, you know that they actually might, uh, optimize to train a lot of healthcare d-data to actually encode the knowledge in its weights.

And I think this is just a great thing for us as well, uh, where the off-the-shelf models can keep betting-- getting better at he- general healthcare information such that what our strategy is, we have a constellation of models.

We can use something for this, that, and, like, we only care about, at the end of the day, the best product experience.

Yeah. And obviously you have, like, overall capabilities improving. I'm curious, like, as, as these models get better, is there something you look at and you're like, "You know, three months ago, we really couldn't do that, but God, like, the, you know, the latest, the latest models really allow us to do it"?

So here's something interesting that I've, uh, k-kind of been toying with. Um, so all models are Yeah, this wasn't super, super obvious a year ago, but now it's become clear and clear that almost every agent is a coding agent underneath, uh, underneath the hood, right?

So you, you give it whatever file system, it can write its own code, and so forth. So when you think about within, within healthcare and the use case that we have, you can think of the EHR effectively like a file system.

Um, it's just-- it's a storage of all this information. It's actually a lot of information there. It cannot fit into the context window, at least of today's models, and you want to use that context effectively for all these product use cases we're talking about.

And so if you have better agents that can actually, like, manipulate data, read that data, treat it as a file system, as we see they're going, and we know model companies are investing this way, um, then that actually very directly benefits us.

Yeah. Yeah. Okay, cool. Again, just establishing basic things, but we're going back to the model stuff. Um, I, I'm really interested in double-clicking more on, like, the real-time, uh, element, which is pretty important for, for both of you.

Is it-- is real-time basically just batches of, like, every one minute, every five minutes? Is that how we actually do it? Or is there some more native, like, genuinely real-time in the sense that OpenAI has a real-time API or Gemini has a real-time API?

Yeah. Yeah. Yeah. So today it is more on the, uh, on the batch basis, but there's interesting prototypes that we have that we're, we're still not fully, uh, full-time, you know, voice in, text out or v- uh, um, in, in that sense.

But actually, can you, can you trigger mo- your models, your agents or agentic workflows depending on actually the right times in the conversation? Um-

Yeah

... and so you can imagine, you know, different techniques to bring this latency down and, like, you know, you want to bring the feedback loop down as much as you can. Um, uh, and so a lot of clever engineering there without fully-- Maybe one day we'll do full voice in and text out, train a model to do something like that.

Do people don't want voice in, voice out?

Right now we aren't creating experiences that are, like, kind of during the conversation, kind of inter... It's almost like-

Might be too disruptive

... too, too disruptive until, like, who knows, maybe eventually you could have full voice agents once we-- the quality and we improve the comfort of the technology. Um, but right now I think gra-- that change is much more gradual, and it's more text focus, uh, text out.

And I think so much of currently what our product is trying to do is allow a clinician to focus on their patient, and I think maybe at some point, but I think right now patients, clinicians don't want a third voice, at least in a literal voice in that room.

And so how do we be there with all the contacts and information ready at hand when there's the right moment?

Yeah.

Janie, one thing I'm curious about is how you think about, like, personalization of the product. I imagine, uh, every doctor is a special snowflake in their own way, has their own way they like to do things. Um, you know, uh, there's probably a bunch of different approaches you could take to doing that, um, you know, both within the model layer itself, but then also just with, like, clever prompting or engineering.

Like, how do you actually-

Yeah

... deliver on that?

Yeah, it's such a good question. Personalization is massive for us. We think about personalization at three levels. The first is at the individual, the second is at the specialty level, and then the third is at the health system or the organization level.

To your point, there are a lot of individual preferences. Um, when a note is produced, it almost is a reflection that is so deeply personal of a doctor's work and how they give care. And so do they have preferences on things like style?

They might want bullets versus paragraphs, really concise versus comprehensive. Um, they also might have phrases that they really like to use or the templates that they want every note to be structured. And I think, you know, we see it in our feedback all the time.

"We want two spaces in between sentences, or I refuse to use this tool." And so that's something that we've had to build in. And I think the tricky part is: how do you make sure that stylistic preferences don't actually interrupt accuracy and quality?

And that's something that we've really had to refine and hone over time. Second is at the specialty level. A cardiologist note or workflow is gonna look very, very different from a dermatologist workflow.

I assume cardiology notes are the highest stakes for you guys, given your CEO is a cardiologist. It's like, "Oh, man, we gotta make sure we get this one right."

Yeah. Shiv, our CEO, is still a practicing cardiologist. He rounds once a month. And so, uh, first call when we want just quick and easy user feedback too. But, uh, specialties require a lot of personalization, both in terms of what does the product actually look like, and so we make sure that as new users onboard, we catch that, and the product kind of proportionally reflects that.

But also on the back end, like evals at the specialty level, they are hard-earned to actually calibrate and get right. What does a really great dermatology note look like? What actually makes it complete? What makes it compliant and billable is very different than a primary care doctor.

And so it's not just about what does the product experience look like, but on the back end, tuning and really, really deepening our understanding for the specialists. What does great output look like? And that's, um, obviously a problem that we need to calibrate internally, externally, online, offline, but, um, takes lots of cycles, but is necessary in a high-stakes environment.

And then at the health system level, um, for products like clinical decision support, you have health systems who've spent years or decades refining their best practices, and they wanna know, "Hey, we love your clinical decision support product, but how do we embed our own hospital guidelines into them to actually inform clinicians before, during, or after a visit what brest-- best practices should look like?"

And I think as you think about, like, deepening moats as well, when health systems, uh, trust us with that data, allow us to kind of productize it and directly into the clinical workflow, uh, I think makes us a really, really great partner to health systems who wanna build something that truly meets their needs, their practicing guidelines.

And I, and I want to add onto that, uh, the-- for the clinical documentation problem, it's, it's very similar to, like, AI writing, AI writing that doesn't feel like your own, and then we call that slop. But the way I describe one framing of slop is like AI without context.

But we have all that context, you know, and both the clinicians, uh, can have it and can guide it. And so part of the other interesting exhaust for us is, like, memory is like actually one of these new systems records-

Yeah

... almost.

Yeah.

And we also have all the edits people make on our product, and when you think about a data flywheel and how we get better over time becomes really, really powerful as a mechanism to just going deeper in personalization.

Yeah.

Evals.

It's interesting. I love this idea of, like, working with systems on, you know, the guidelines they built up over a long time. I feel like so many of the best AI app companies today are, you know, the question is how do you take, you know, the expertise that a law firm or, or a bank has built up over many years, and then add that as context and also a special sauce over, like, a, a, an AI tool.

And so seems like y'all are really doing that very effectively.

Yeah. We're now starting to have our customers ask, like, "What are other customers doing?"

Yeah.

"And how are they doing it?"

Ah.

And I think as we think about having visibility across such a large set of care being delivered right now, a really interesting place we could also partner.

I'm just kind of curious. I-- this, this may be a nothing question, but, uh, how different are health system guidelines from each other? Like, don't they all converge to the same thing? And if not, where, where do they differ?

I think at a really high level, they're, they're gonna talk about very similar things, but the, the difference is probably in some more of the details. Like, oh, you should refer to specialists only when XYZ conditions are met or so forth, and maybe different organizations have different practices and guidelines around that.

But high level, talking about similar things, but the details are actually what, of course, that shapes the context and the decisions you make.

Yeah. And this all goes into the context engine, and it might affect the notes, but maybe not.

The-- for these local pathways, we're definitely thinking about it a little more for our clinical decision support product.

Yeah, yeah.

So yeah.

Which is your stuff, yeah.

Yeah, yeah.

Um, okay. And then the, the memory which you, you raised, let's just tell us more about that. What, what have you tried in memory? What's the structure of the memory? Um, what works? What doesn't work?

Yeah, there's like, of course, many different ways you could do memory, where it's like, okay, can you actually bake it into the model weights, or can you do it in some external s- uh, store? Um, I think for us, what's interesting is, of course, when you think the models are rapidly changing, whether it's in-house or third party, baking into the model weights, sometimes you worry that it could be a little throwaway.

And so, like, how do you... You need to find a way that you decompose the problem, the preferences from the underlying models and so forth. The thing we're right now most ex- uh, both that's easiest to start with and we're excited about is having, like, a separate store for memory, where you actually have, for example, like, a memory sub-agent that's, like, working in the background, figuring out what are the important parts of the clinician's actions that we want to remember for the long term.

And then you can also imagine, uh, other things where in the-- you have background jobs that are running that are collating these, like, memories similar to, like, Sleep, of course, uh, and what other pa-pattern, uh, patterns products do as well.

Learning over all these action, all the action data we have, again, like note edits, the conversations they did, and the actual transcripts.

What about evals? How in the world do you-- like, this is such a complex product surface area. Um, we would love to hear you riff on that. And also, how has that evolved? Like, I'm sure you've gotten better at it, so any, any kind of learnings along the way?

From an evals perspective, uh, we-- I think from day one when we build any new product or feature, we think about, like, what does good look like. And there are table stakes things like clinical safety, but then you start to get deeper into what does good quality look like.

And when you go into something like our core product, there's stuff like style and completeness, and there's things like does this note actually become something that can be billable, which is obviously very, very high stakes for a health system.

We have a number of ways in which we get confidence for this. We have, uh, internal in-house clinicians who do what we call an LFD process to give us our very first pass at is this or isn't this a good enough output, uh, look at the effing data.

LFD?

Yeah.

Great.

That's why I was smiling.

Yeah.

I was like, is Janie gonna mention what it stands for?

Oh, no, it's not... There's like a million acronyms-

Yeah

... how am I supposed to know that I don't. So like, oh yeah, of course, like an LFD.

Yeah.

I've never heard of LFDs.

Yeah.

It's Abridge specific.

I, I think I got through three days, and then I had to ask someone. I thought it was just me that didn't know, but-

Yeah, yeah

... it's our, our internal process.

But look at the data is a meme in ML, 'cause you, you tend to not look at it. You just kind of look at number go up.

Exactly.

Um, but yes.

Yeah. But so, uh, we make sure we look at the data, and then as we think about all of the components of good output, we, one, create LLM judges across all of these, and we make sure with annotated data and either internal or external evaluators, we feel like these judges are calibrated.

And then depending on the stakes, we also work with in-house and third-party evaluators across all of these before we ship any big change. And I think the goal is, in terms of evolution, how do you go from this process taking months down to weeks down to days?

Some of it is, like, a true science and ML problem. A lot of it's also just, like, hard operational work. Have you planned ahead in terms of what you need? Have you really optimized the capacity that you need across all of the different specialties you need?

Have you gotten a really good sense of, like, which third parties are great to work with for what use cases? I think this takes a lot of domain, like, expertise and, like, to be frank, lots of mistakes and errors in figuring that out.

And so I think as much of it is an ML problem, like, so much of it has also been operational gains that I think are hugely, hugely important, where domain-specific expertise is, is everything.

Yeah. But it's funny 'cause I feel like people talk about healthcare like it's one giant market, and the reality is-

Yeah

... it's like, you know, dozens and dozens of, of sub-markets. And so it feels like in your evals, you obviously have to, you know, uh, build that up across the board.

Totally.

And is specialization the, the primary mmm cardinality at, at... That's the word that comes to mind

Sometimes, depending on the product or the use case. And so if we're making a note improvement or feature for a particular specialty, definitely. But we have products that are for nurses. We have products that, um, are really, really aimed at making the document or the output a lot more billable.

And so we'll actually wanna work with coding teams and not necessary clinicians. And so-

Coding meaning healthcare coding-

Yes

... not-

Yes, yes, yes

... other kinds.

I see where you're...

Yeah, but is this output proportional to the work that was actually delivered? Is there sufficient documentation to justify the amount that a health system may end up charging? And so, um, specialty sometimes, but also domain very different across all of the different products that we're working for.

And building out that network is, uh, not easy, and I think is, is where a lot of our operational investments have gone into.

And I think, um, I, I view a lot of analogies to self-driving cars here, where, like, part of it is y- we actually really want progressive rollout o- of features to actually test in the real world, is this useful, is this gonna work?

You know, one, one big difference compar- compared to past lives is before I'd build a product, maybe I'd alpha it, and then I'd, like, GA it the next week 'cause I'm like, "Go, move fast, ship," and whatnot. But the mentality is, is like you...

I wanna make contact with the reality as quick as possible, but I want a progressive rollout because as much as I get as large of an offline eval set, I want the distribution of that to actually match real-life distribution.

And over time, by rolling out early, I actually think similar to Waymo has this tagline, the world's most experienced driver. I actually think another thing that can actually, like, uh, at least linearly increase for us is, like, both the size of our evaluation offline and online, um, that a- and it all feeds back.

I think something that's been earned over time, speaking of evolution, is just the trust we've gotten with customers. Um, historically, a lot of these health systems, when they bring on new vendors, their release cycles are quarters, sometimes twice a year.

We've gotten our customers onto monthly release cycles, which is pretty fast for health systems. But I think what is more exciting over the last, call it, few quarters, has been, um, a subset of our customers have said, "We actually wanna innovate with you.

We trust you." And we have a pretty, like, decent chunk of our customers who say, "We'll actually develop with you outside of these monthly release cycles. We have a higher tolerance. We know that the stakes are very high, but we want to be the first ones using these products, giving you feedback."

And so for, a, a pretty substantial set of our customers, we've been able to convince them to, to be able to ship, uh, you know, in this gradual way, way before GA. I think something we talk about a lot internally is, uh, trust is earned in drops, earned in buckets.

And so we still can't do what I used to do when I worked at Loom. We had 30 million users. I'd just be, you know, rolling out experiments left and right. Like, the bar is still quite high for iterative rollout, but I think, like, because of the, the trust we've earned, we're able to learn at pretty high volume very quickly.

Yeah. I mean, your scale is still pretty, pretty huge.

Yep.

One thing I wanna-- we were gonna go into scale, right, in a sec. Uh, one thing I wanted to follow, follow up on evals, which a- again, just coming from a generalist engineer point of view, just thinking through what would people be scared of in doing this, uh, the privacy and HIPAA elements of this, uh, I have zero experience in that.

What do you have to do? What is actually surprisingly not that bad?

So one thing that's really important here from a compliance perspective is very much that any of the data we use needs to be de-identified, any real-world data we use as a basis of, um, online eval sets we're learning from.

And so you have to... And there's, like, actually very clear, you know, government guidelines, like, what counts as PHI. And so we've actually even have built models that can take, for example, a clinical transcript and actually remove all the key PHI indicators, and so you have a scrub/de-identified version.

And then once you-- And so one thing that's important is first you've got to get confidence in that model in the first place, right, and prove that out because, you know, now you have, like, multiple probabilistic systems on top of each other.

But once you, once you have that, then you can actually train on it, use it for evaluation and so forth, provided one of the cool things also that you can do from the business side is the right data contracting as well with your partners.

Is the anonymization one way, like, once it's done, you cannot undo it? Or is there someone-

Yes

... who holds the master key that can... Yeah, okay. So it's, it's one way.

Yeah, it's one way.

It's one way.

Yeah.

Yeah. I, I guess-- I mean, I guess that's how it works. I just wanted to... Because, like, you know, there, there's a lot of this, like, learning from feedback and everything that, like, you would want to debug more, but you can't because you just physically don't allow yourself to.

Yeah.

Right?

Well, some of it's also written- ... in our customer contracts in terms of who can or can't access PHI data, how long do we retain it, um-

Very good

... before it gets de-identified. And so we have a pretty high bar for who can access that PHI data, um, just to make sure that we always respect our customer data and privacy. But that's something that we partner with our customers on, too, to make sure that as we want full, as close to precision as possible in, in that quality-

Cool

... we can still use it.

Yeah. But it'll be fascinating to see how that space evolves, right? Because you think about, uh, I, I used to work at a company that, like, did a lot of healthcare data in the cancer space, and if you ask, like, the average cancer patient, like, "Hey, do you want people-- do you want other patients to be able to learn-"

Take it.

Yeah. "... learn from your experience?"

Take it all.

Yeah.

They're like, "Please," like, "I'd love, like, nothing more than for other people to be able to learn from-

Yeah

... the, the experience that I had." And so obviously, in the past, it was a lot harder to do that kind of learning. But I think with this technology, uh, that actually might really be practical. And so it'll be fascinating to see how that, uh, how that continues to evolve.

Yeah. There's so much in our data set of 100 million conversations. You, you can imagine things like insights that you can give to the clinician. How could you, oh, how could you have reacted to this in coaching? Or insights around, like, uh, which treatments are effective or, like- Because you have this, again, this data source that was never captured before, but that's, like, where, like, intuition or experience is created from, going back to this idea that the conversation is the agent of choice.

Yeah. I guess, you know, back to the 100 million conversations, I mean, I feel like you have this insane scale that maybe only a few other AI app companies have and everyone else dreams of. So not, not everyone has had to confront this yet, but maybe just talk about some of the, like, challenges of, of operating at that scale and what, what do our listeners have to look forward to if they ever get to, uh, to this level of scale.

I think, uh, at large, in larger in scale, so of course there's a general, like, infrastructure reliability. Like when you... in any given startup, you're kind of building the plane while while, while it's flying. So, so there, there's some notion of that.

Um, but I think what gets interesting on the AI and ML side for sure is this, as you get in more and more scale, so one, you, you have the data to first and foremost do this, but, like, you actually start thinking about costs or infrastructure in a whole different way at scale versus, like, a prototype.

Yeah.

You can use the most expensive model, you can burn as many tokens as you want, but when you're doing 100 million conversations-

Yeah, token max and leaderboards are less exciting, uh, in that context.

Right. When you're doing that, and so that comes for we have the data and we also have the team that's able to actually, like, post-train based on this, and you can actually optimize for efficiency, especially in areas where you believe that maybe a lot of the quality headroom is less so, um, and you don't expect the other off-the-shelf models to go that way, such that you wanna do, you know, efficiency maximization, uh, in terms of, uh, compute and tokens.

Yeah. I mean, I feel like you guys live in, in the future in some way where most, most use cases today are really just in use case discovery mode, where it's like, "God, I really hope I can find something that can get to scale," and so you're always gonna use the most powerful model and then the, the few things that do get to this level of scale, you start to do those kind of optimizations.

Yeah. It's a natural trajectory where it's like zero to one, we're not talking about any of these optimizations. But when, when maybe we're in the one to 100 or so forth, then we're in optimization mode, and, like, what works out really well is you've got all this data from zero to one that lets you do this.

Yeah.

So.

That's fascinating. I mean, I feel like there's, there's, you know, one thing that's so interesting about the Abridge footprint is, like, you're in the doctor-patient visit in real time. There's probably, I always like to say, there's like probably 50 years worth of product you could build on top of that.

What g- gets each of you, like, I don't know, what are you most excited about building, you know, either in the short term or medium term or even, you know, long down the line?

I think something that I get really excited about is that the same conversation can serve so many stakeholders. Uh, if you think about the conversation, a doctor needs to know what is the documentation, how do I make sure that this fully represent the care I gave?

A patient needs to know, like, "What the heck just happened? This was really overwhelming. What are my next steps?" A payer needs to know was this the proper and appropriate care given. Um, a pharma company might wanna know why isn't this drug being properly used, or is there actually a good candidate for this clinical trial that I'm about to run?

And I think where I get excited is that our product and our platform and our infrastructure can be the same product across all of those things and start to what's today, like, separate, very expensive, complex systems that serve each one of these stakeholders in very different ways, start to kind of collapse all of that into a singular platform that enables not just more efficiency across the board, but also better outcomes for, for everyone.

And I think, you know, I think all of us experience healthcare in probably very painful ways, and I think knowing that there is a world in which we can simplify a lot is, is really exciting to me, and it all starts with the conversation.

You know, it's interesting. I think of it very similar to going back to the KPIs that any AI product cares about. How do you increase quality of care? How do you reduce latency to care? And how do you, uh, reduce costs, which is a huge, uh, in healthcare-

I think they call it triple aim in healthcare.

Um, but, but, but, but very similar to, to building AI products, and the thing that really excites me is when we talk about that latency piece, we talked about one example earlier of, like, prior authorization, can you reduce the latency to care?

But you can imagine so much more, like, oh, as soon as the lab value gets updated, do you have, like, a background agent that, like, kicks off and uses all the context to be like, "Oh, hey, actually the patient should do this next," for example, and f- uh, flagging that to the clinician who's always in the loop, but reducing that latency, um, to care.

And then you can imagine this is much further down the road, but it's like even connecting that to the direct patient and the consumer. And so how can you, how can you build, build a bridge to all of these things?

Very cool. I think, like, the connections piece is, is just an ever-growing thing. Um, and one, one of the key partners is the EHR, and I, I wonder what that relationship is like. Like, um, will, will they, will they, like, you know, look at this as, like, something that is valuable enough that they wanna own someday?

Yeah. I think our partnerships with the EHR is, like, we know that we have to be extremely close partners with all the EHRs who we partner with. Um, being able to not only pull and push all of the data into the right places is, like, not only table stakes, if we can't do that- Health systems don't wanna use us.

I think the second and the reality of today is clinicians spend a lot of their days in the EHR. So much of what allowed us to win in the largest health systems was pretty direct and, uh, very, very close partnerships with some of the largest electronic health records that allowed us to pull and push data with, you know, APIs that weren't ready out of the box.

And clinicians wanna save clicks. Anytime we introduce a new product that, you know, adds two clicks for them in their day, they're like, "We're not going to use it."

Yeah.

Um, you know, they have 15-minute back-to-back appointments with their patients. They're spending, you know, hours during pajama time doing documentation. Like, every second and every minute counts. And so we really think about being deeply integrated into the EHR as also table stakes to actually getting real usage and adoption.

And I think anything that we build or introduce, we really talk about earn the right internally a lot, which is we have to provide so much value or save so much time that people will use us. Um, but I think those are the two things that are, are, are close to us, is we know that the product won't be used unless it is deeply interoperable.

And, and strategically, to your, to, to your point, it's like, what does EHR wanna own vers- uh, versus us? You know, EHRs are really focused on the clinical workflows and so forth, but some of the things that we're talking about here, um, at least traditionally are outside of the domain, where it's like, oh, connecting payers and providers together with like, uh, provider policies or the clinical trial matching, as Janie brought up.

And so these are like enti- entirely... We position ourselves as building this entirely new intelligence, clinical intelligence layer across, again, providers for pharma and, um, payers.

Right.

And so that's a, it's a whole different ballgame that we try to play-

Yeah

... in combination with them.

And it's like a different layer of scope.

Yeah.

I'm curious, you obviously, uh, are, are both relatively newcomers to healthcare. Um, people have these like, you know, uh, there's lots of futuristic healthcare AI takes of, "Oh, everything will look different." You know, now that you've been in healthcare for a bit, you obviously live at the edge of AI, what have you, like, changed your mind on around this, you know, as you think about what healthcare looks like in 10, 20 years?

I mean, any, any updates to your mental model from, from the time actually being close to the problems?

One thing that I was hesitant about before, and actually it's a common thing when I'm trying to recruit engineers that people ask me around, is definitely like, oh, healthcare heavily regulated space. And, and it is, rightfully so. You wanna keep, uh, the patients at the end of the day safe.

Um, but one of the interesting things that, uh, I actually think is a y- that surprised me how much it is coming to the company is actually there's a lot of really favorable regulatory tailwinds as well. Where you think about, like, government actually really wants interoperability between all these systems that we talked about, and so agents can access this information.

Um, the government actually just in January, the FDA released updated guidance on clinical decision support, what I work on, in such a way that they used to have guidance from like 2022 that required you to have like mention all these options and do all these other things, but actually it's a very forward and forward-looking way.

And so I think for me, what's been really cool to work on is I think this, there's this very special moment both in AI in general, we all know that, but there's a special moment also regulatory in healthcare as well.

I think one thing I would call out is I think for the very reasons things are higher stakes or, you know, potentially considered more difficult in healthcare, I think it's where some of the hardest AI problems will get solved first, just because the bar is so high.

I think when I first joined, I was like, oh, this is where we'll be on the tail end of where, like, all of the AI innovation will actually be able to be applied. But when you think about, like, zero error evals or multi-step workflows that have really, really low tolerance, I actually think a lot of the innovation will, will happen here, just because we have to or else we can't ship.

Yeah. 'Cause like in other domains, you'd much rather to solve the 80%-is-good-enough problems-

Yeah, 80/20 doesn't work here

... versus, yeah. Yeah.

And tr- tr- building off that, I think traditionally, I think there was a bit of stigma that, oh, healthcare companies are not that interesting from a technical perspective, or I've seen that or faced that myself. But these are actually really, really hard and fun problems from a pure technical perspective beyond just the impact.

Like, how do you bring the latency of this thing down, um, and make it really high quality?

Yeah.

How do you bring the latency of things down?

Yeah, yeah, yeah. Um, so okay, l- let's answer the latency question. Um, and, and maybe ho- hopefully not too redundant with some of the things I've said earlier, but some part of it is w- with any latency, you have to like what is actually, what is really your bottleneck.

In a lot of workflows, it's sometimes it's the model itself, and so that's where our, like our data flywheel, our post-training team and so forth come in so that can you make the models far more efficient. So that's one aspect of latency.

But there's whole 'nother aspects of latency where it's like, okay, on top of that, if you use a constellation of different models, can you use, can you first use like a, so like kind of thinking fast and slow.

Can you use a cheap, fast model that triages and hands it off to a larger model that where you get more intelligence and so forth. And so all these clever tricks to, to make it work.

Yeah.

Um, and by the way, we are totally-- we also realize that the product frontier is changing, and so these tricks will, were, may not get us to where we wanna be in five years, but th- w- we need to if we wanna build a useful product right now.

Should we go to the quick fire or you wanna ask more about Abridge? We can stuff everything that's non-Abridge into the, uh, into the quick fire, but-

I don't mind. I, I mean, I, I was just, I think, I feel like Janie was on the topic of more long tail stuff, which is-

Yeah

... not the 80/20 thing, and that really matters, and I'll... You know, if you have any tips or cool stories or just general approaches that have worked for you that's interesting to dig into.

Yeah. I mean, I think one of them is even just how we staff our teams looks different than a traditional software engineering team, I'd say.

Let's go.

Uh, we have a bunch of folks with different roles who are clinicians And so we have this role called the clinician scientist, and I think I heard one of our leaders refer to them as mutants recently. But they are people who've had clinical backgrounds, so MDs typically, who are also deeply technical, somewhere, um, on the spectrum of, like, basically a full stack engineer all the way to, like, extremely, extremely scrappy prompter.

But having each of these people embedded within our teams instantly raises the bar for everything that we build because not only are they determining, like, is this product clinically useful, but they're deeply, deeply embedded in our whole evals process.

And so when we talk about LFDs, when we talk about what is our actual evaluation criteria, like, you don't want Chai or me creating what those are because we don't have clinical background. But I think that is probably unique to Abridge but has been game-changing.

And when you think about where the puck is going, you have people build with clinical backgrounds who are technical, and where AI tools are going, they just become more and more, uh, critical and, like, the killers of the team.

And so I think that's one. And then I think the second is just the scale at which we do evals. To catch that long tail up front before ev- anything ever gets into production is something that we've pretty much, like, really, really started to fine-tune, both from a scale, but when do we know we need to get several hundred versus several thousand offline responses, what helps us make that quick decision and make this less of an art and as much of a science as possible.

But I think that's also been something we've had to tune over time.

And you have partners who basically opted in to give you those evals?

Yeah. So we, um, work either internally or with third party for offline evals, and then we have he- uh, customers who also agree to give us, whether it's, like, thumbs up, thumbs down to, like, choose this or that, um, a lot of data to get us to what is as close to, to fully confident as possible.

Yeah. The, the term that comes to mind is-

Yeah

... kind of like active learning on things where you know you're weak. I feel like it's, it's kind of a lost art, but-

Yeah

... uh, is a lot of the polish that comes into doing something like this.

Really.

Yeah.

Hundred percent.

I mean, maybe, uh, on a, on a totally unrelated note, like, Chai, I think you obviously had a very, uh, uh, storied run at Glean before heading over to, to, to Abridge. And so, you know, I'm curious, like, that's...

it's obviously was one of the early AI app success stories. I guess reflecting back on that experience, like, what do you think Glean got most right, maybe most wrong? Um, yeah, curious for your reflections.

I think the... I attribute Glean's success really to, uh, very, very strong technical foundations, um, that have really stood the test of time. Um, and so it started with, with... it started with a known problem and, like, finding information where work is hard.

The best technology at the time was to build really, really high-quality search. I think a lot of times enterprise search startups failed because the quality wasn't great enough. But the learning that people took away from that is, "Oh, enterprise search is not good enough."

And so, like, quality, I think, really changes the game of, like, if something can be useful or not. It's like similarly, like, people may have taken it that way, like, "Oh, Alexa voice assistants are not that useful," but when you have quality, things can change the game.

And so I think Glean's early foundations, by bringing people who had built search at Google, the best place to have ever built search, um, and being really creative and having a very concrete problem to solve, but with the right technical backgrounds, laid the foundation for all of its success for the many, many, many years to come.

And I think what's interesting is always figuring out, like, "Hey, how does a company adapt in this," as we all know and we've talked many times, in this changing landscape. And so for Glean, like how do you put this context layer to the use, um, has been the thing that we, we've really, l- last few years, has been the fun from the challenge that we're like...

You could say, uh, like, that's been the opportunity for the company as well as the challenge as well.

Yeah, definitely a, a competitive market. It feels like one at, at the epicenter of the foundation models and, you know, uh, the hyperscalers, so it'll be interesting to see how it all, all plays out.

When you think about can you build something that helps everyone at knowledge work as well is, is, is a, is a massive opportunity.

Yeah. Always my mental model is, like, there's a few m- markets that are, like, the foundation model companies have to win or are, like, big enough to, to go after, and it's probably, like, consumer code and that.

Yeah.

Um, and so it would definitely be, uh, be interesting to see how it plays out. I, I guess one thing we often think about on the investing side is, you know, the pace of, of progress in models changes so fast, and so the building patterns adjust so fast.

And it's always hard to figure out, like, what pieces of the way people are building today, the infrastructure tools they use, are gonna prove persistent versus, okay, six months later we're doing something completely different because-

Mm-hmm

... you know, models have, have improved. I'm curious, of the stuff you use today, like, how, how do you think about the pieces of AI infrastructure software that feel a little bit more persistent?

Okay. So generally, if you take the thesis that the models are gonna be more and more agentic, I think before we had to build a lot of scaffolding around that. Like, in, like, in previous case, like I've-- we've effectively, we made our own DSL effectively, and you can view the like, um, uh, because the models were not capable enough, so you needed to simplify things.

Um, and you can view it similar to, like, other agent frameworks. But over time, if the models become more and more agentic and can use the similar tools that we already have, where it's like computer use, writing code itself in sandbox, I think much more around, I think far more about, like, what are the right context layers and the tools to give agents.

And then I think the other things that I think about are actually how do you really build truly event-driven real-time systems, and especially at Abridge, again, where you're doing something real time in the conversation. And so I think there's a lot of event-driven technology.

And by the way, stuff that we've always used in the past, whether it's Kafka, Temporal, Sockets, and so forth, how do you bring that together is, I think, actually also durable. Um, or thinking about patterns in which humans collaborated with each other on Google Docs.

How do you think about like CRDT and so forth when you have conflicts, when you have multi-agent systems? So all these things that we've built for... I actually think the things we've built for humans are actually the things that are con- gonna be continue to be durable.

Right. Just with like a thousand times more the scale of agents running at them instead.

Yeah.

They're, they're gonna really work.

So make sure that they scale, of course, and fast and whatnot. Without a doubt, yes.

Does Abridge become more agentic over time? Then, you know, what, what does the, the next more agentic version of that look like?

Um-

'Cause you're already pretty proactive. It's, uh, with, like, the notifications, I guess.

Yeah. And s- so I view that as, like, a piece of being agentic, but I also view it as maybe some of the things we mentioned before, like, oh, reacting to labs or, like, you know, m- doing work in the background, or doing-

Yeah

... even more capabilities on behalf of the clinician, who we believe has a super important role to play as, in terms of, um, patient connection and so forth.

Yeah.

I'm curious for both of you, like, what's one thing you've changed your mind on in AI in the past year?

I think the one I flip-flopped on, and this is much more product specific, is, uh, probably the hotter take is that prototypes are the end-all be-all and that PRDs are dead. I think we've tried switching and kind of...

We continue to evolve the way product is developed, and, uh, the products that we're building are extremely complicated and nuanced, and it is very, very difficult for a prototype to capture the full complexity of what can we or can't we do with this data.

Um, what and who... Is this the actual right problem to be solving for in a world where software has become so cheap? Like, yes, this is a cool-looking prototype, but should we be spending any of our precious hours here?

If so, why? And, like, how does this deepen our moat in a world of decreasing moats? Um, does this require custom implementation from our customer to actually use? None of that gets captured in a prototype, and so I think, uh, we've- we're continuously evolving the way that we develop product here.

But even if not written in the same traditional ways as it was two years ago, I think, as a team, we've gotten pretty, I think, high conviction that in a world of so much noise, like, crisp written clarity is more important than ever.

It might now live in a markdown file that more teams and systems can use as context, but that's probably one that is much more-

So you-

... function specific to me.

I love that though.

But, um-

You're disagreeing with the consensus.

That PRDs are dead-

That's good, yeah

... and that-

So you, you are, like-

... prototypes are the thing.

Like, yeah.

We should partner with AI to create great documentation, but I think first, probably most important, is strategically answering, like, why is this problem the one our company and our product should solve? What happens if the next 20 competitors build this?

Like, why, what is our right to win, and does this help us differentiate in any way? Or are we just adding noise? I think it's important-

That's a high bar. I don't know if I could answer that-

Yeah

... because a lot of the times the answer is, is let's do it first.

Yeah.

And I think when the cost of doing it first is so expensive, we just talked through the, the process of getting something out to customers. I think you need to have a higher bar for, like, as a business, should we invest here?

And I think i- as all of our roles evolve, like one of product or, like, all of our jobs become like, should we do this thing? And I think that's something that is worth the time spending up front on.

And then, um, as you think about prototypes, it's still really valuable to quickly show, "Here are the 20 ways we could do it. Clinician, like, I would love your feedback. Like, which one resonates more?" Or as you get into deeper fidelity, you can also make the prototypes deeper fidelity and, like, get it as close to production-ready as, as possible.

But, um, beyond that, to actually get it out to customers, there's a lot of implementation details, security compliance, edge cases, um, things that never get caught in a prototype that I think need to be written out somewhere. And so they look different, but I think still more important than, than ever.

Yeah, it's interesting. I imagine a lot of that also is, is, like, given the con- the context of the stage that Abridge is at.

Yeah.

And I feel like for so many early stage companies, it's just a desperate race to f- you throw, like, 30 things at the wall. You're like, "Please, something just, like, resonate with my end buyer." Um, and, you know, you find something and, and that's, you know, why the prototype first approach is so powerful.

But for you all, it's like anything you're gonna do is across 200 systems. There's, like, a whole, you know, implementation change management side of things, and you get a few big bullets to fire at, like, you know, uh, at, at what you want those systems to do.

And so I think being really thoughtful about that, uh, makes a ton of sense. And maybe the, uh, the, the prototype first takes will all grow into, to your-

Yeah

... view of the world when they're, uh, when they're a bit more scale.

Yeah. I think the weekend demo versus it works at the largest health systems is, like, a massive, massive gap. I don't think it means we can't go fast. Like, I think this is the fastest I've built in my career, um, right now, and I think the-

Compared to Loom?

Yeah. I think from a, like, ti- the complexity and the scale of the products we're trying to build and the problems we're trying to solve, I'd say, yes, maybe I, like, updated a flow or, like, shipped a new feature pretty quickly, but if you think about some of the products we're building, we're trying to collapse prior authorization, like things that used to take 45 days across maybe 20 different touchpoints into, to one.

Um, I think I'm building faster than I ever have, and so I think the thoughtfulness actually allows us just to go fast at the right things. Um, it sounds contradictory, but I think, uh-

No

... thought out front-

Go slow to go fast.

Yeah, exactly.

Yeah.

Yeah. I, you know, it's, it's interesting. I think in the, when a lot of things are changing and in the AI discourse, like, I think sometimes we lose sight of things that always sort of stood the test of ti- time, like judgment and clarity always matters.

Like, as an engineer, sometimes I don't want a prototype. Actually, I would like to see, like, I want the written, the clarity that comes from writing, and then we build that. And again, for some things, of course, where it's a small thing, like, yeah, just ship the prototype.

That's why, like, don't sweat the details. So I think the interesting thing, the nuance that gets lost sometimes in discussion is, like, sometimes we need to recalibrate our judgment for sure because the costs and gains have changed, but that doesn't mean we go all the way on one spectrum or the other.

Yeah.

Yeah. Outside of your specific tool, I always like to ask this question, any other AI tools that you guys are enjoying?

Claude Code. But, but, like, that, that feels like too, too basic of an answer

Is, is, is, is all of Abridge engineering very built on Claude Code?

Yes.

Wow.

Very much so.

Wow.

You know, I, I won't-

No Cursor.

Oh, we also have Cursor as well.

Cursor's good.

I, I mean, I mean-

I'm just checking the boxes here.

Yeah. Many, many, many of the tools available, but it's like you look at, uh, just earlier today, you, you see an engineer's screen. You see, like, six different, you know, Claudes running at it. Sometimes the same person, I've seen them on the sofa now with the remote control as well on the mobile.

But, like, uh, very much so, like, one of the interesting things for me is, like, as a r- relatively new person to companies, like, Claude Code actually helps me onboard much faster or any of these AI co- And, like, I feel like I learn so much-

I do love the memes of, like, you know, "Claude's gonna do this." So like, I'd like to see Claude, you know, uh- The, the venture equivalent is like, "I'd like to see Claude go do a company at a billion dollars pre-revenue."

Like...

Um, well we always like to leave the last word in these conversations to, to you both. And so, uh, any place you wanna point folks where they can go learn more about Abridge, the work you're doing, any of the research you guys have done, whatever.

The, the floor is yours.

A couple places. If you, uh, at- on our Abridge website, we have a lot of our whitepapers where we've done a lot of interesting work, such as, like, d- uh, reducing a hallucination rejection.

Very well-presented, by the way. I liked it. Yeah.

Th- thank you. Um, you know, our science team rigorously defined what actually, actually is a problem. And one of the interesting things, by the way, at Abridge, is we actually have multiple, uh, stats professors on staff as well.

Um, so in that, in that specific whitepaper, Michael Oberst, who's a professor at JHU. Um, and so we have multiple, a- and from that comes, like, very high rigor, and then also our taste for design comes from really good presentation.

But, but setting that aside, and we, we're gonna have many more technical topics there, please follow our Twitter account as well, um, AbridgeHQ. And then the other thing I, I'll plug a little is, um, we have a open house of d- diving deep into AI and healthcare coming up with Andreessen Horowitz.

Amazing. Uh, well thanks so much.

Yeah.

Thanks so much.

This was super fun.

All right. Thank you.