VIDEO FEED
===

[00:00:00] Since I have started working with ai, I feel like we have come a long way. One of the funny quotes from a provider, I will tell you, when we launched this, she said, oh, there's no way I'm gonna use AI in my practice. I will never do that. And we encouraged her to try with us a little. She did. And now if you go doctor, she's like, you can pride from my cold at hands.

Welcome to Launch Pod, the show from Log Rocket, where we sit down with top product and digital leaders. Today we're talking with Dr. Deep, deep Tani, VP of Clinical and AI Solutions at NextGen Healthcare. In this episode we'll talk about how Dr. Tani's team proved that AI can be trusted with sensitive medical data.

So why are you still avoiding it in your product? The key lessons behind NextGen's AI that turns Dr. Patient conversations into medical grade accurate structured data and how Deep DE and her team won over skeptical doctors and turned them into AI champions all while saving them hours per day. So here's our conversation with Dr.

Deep, deep Ani.

Jeff: We got a fun one today, Dr. Deep, deep Ani. [00:01:00] How are you? Welcome to the show.

Deepti: I am good. Thank you for having me here.

Jeff: I am really excited. So we always kind of joke that there's always, everyone has their own path in a product and it's always a little bit different, but you are definitely that I know of at least the first person who started as a medical doctor. You're trained as a doctor and then you ended up in product management in tech.

That's a wilder ride than I've heard from most people. So maybe you wanna just give us a little quick background on like, how does one go from getting your, you know, MD in India to running product for a large tech company here, here in the us?

Deepti: Jeff, that's a very good question, right outta med school. Um. When I was looking for some jobs, it was like, or I was trying to decide what I want to do. I was getting a lot into OB GYN practices and I was like, that doesn't really catch my interest.

It's not what I want to do forever. Then I started working with this company, which was a little way ahead of its time at the time, and starting a novel concept of telehealth in India, servicing the, [00:02:00] population that works in the industrial area, in in manufacturing units because they can't take time away from their work schedules for like fever, cough, cold headache, et cetera.

Right? So they wanted to be able to service that people and that sounded very cool to me. So I joined them and we started with a team. I had a 14 team of 14 doc doctors at the time. And when we were launching the service, even during the concept phase, we were discussing. Let's say, how are we gonna identify the patients coming in and how are we gonna treat?

So when we were putting our entire concept in place, it was like, how are we going to get to the history of the patient when somebody calls? We need to know who's calling, what are they calling for, and what was their family or previous history and realize we need something on computers. So. My founder, director at the time said, you know what?

Let me introduce you to some engineers and dev people and maybe you can tell them what you want and they can give you something on the computers. That's how it all started. I had no idea I was building a lean EHR at the time.

Jeff: Yeah. You, you, you didn't even know you were building a [00:03:00] tech company accidentally. Right. But I mean, it shows that, at the heart of it, product is, is solving people's problems. It's not shipping features. It's not can you get the, the call line set up?

It was like, how do you make it easy for these people who need medical care and don't have much time? How do you make it easy to happen and keep the company going and keep the people going and all of that? So, you know, fast forward to now and you find yourself at NextGen. And I would say, you know, at some level of problems, kind of, kind of similar in, in some ways really just wildly different.

Because now you're sitting at the center of a ton of efforts around AI and how do we help, you know, doctors just do exactly that, deliver care better. So we're gonna skip like a whole bunch of years here and just jump right in, maybe to next gen. And you wanna just give a little background like what the company is and what, what's going on right now and kinda where we're at.

Deepti: Yeah, so NextGen is one of the providers into the Atory care market. We service different specialties for all the clinics and practices that they have. [00:04:00] We have enterprise clients and what we try to do and the lens that I've, we were just talking about, we focus on what's the user trying to do, what are the problems they have and what we are trying to service them for.

So, as I said, we are not inpatient. Everything in ambulatory serving, multiple different specialties, behavioral health musculoskeletal, ophthalmology, and federally qualified health centers, community health centers, and so on and so forth. So we have a wide variety of clients that is using our solutions.

And how I started my story is I put myself in the shoes of the user because I was the user at the time and I was like, oh, I need to solve this problem. And I still do the same. I train my team to do the same as well. Don't think you're building a feature or a solution. I. Just think if you had to use it, what would you do?

What would you want experience and go through that exercise and then build that feature for the user. So what we have learned is over the years, clinicians really do not wanna use EHR systems. [00:05:00] They see it as a barrier, then as an enabler, just because nobody had been using the lens as much.

Jeff: It is funny because every. So many different industries have exactly this problem, right? If you work in like a B2B Tech company, you have a problem where your sales teams don't wanna use your CRM and fill in all that data. If you work you know, in a customer support, you know, organization, typically people don't want put in as much data as you, you know, central might want them to.

This is kinda a problem across the board of there is a use for all the data that's coming through. There is a human layer, you know, a human interaction layer, which is really important, right? If I were to go to a doctor and just talk to a screen, I'd be. I'd be kind of, I'd be kind of unhappy, I'm gonna be honest.

But at the same time, I, I want the doctor paying attention to me. I want the customer service person paying attention to me or the salesperson paying attention to me, not trying to take notes. So yeah, I mean, you, you guys are tackling a very, I know it's medical focused for, for NextGen and you have some specific problems coming off of that, but this is, it's a pretty macro problem you're tackling.

Deepti: We are just to tell [00:06:00] you, our philosophy is not to put the tech in between the provider and the patient. We want to take that away and our philosophy is the new UI is no ui. It is actually focused towards trying to free the providers from having to use the system when they're actually with the patient in the room, in the exam room, so that they can focus.

There is a. Higher level of engagement between the provider and the patient, so they are able to provide better care and which is what they want to do. , , the clinician's main job is to actually provide patient care, not fill bunch of screens and documentation on the EHR. That's to support them.

So we should be enabling and not actually creating more barriers. And that's where we need to focus on how can we create those efficiency And next year we actually have solutions towards creating provider efficiency. And my focus, being a clinician is a lot more on that We. We have solutions started from intake, like throughout the journey from patient intake.

, towards the end of where the claims and billing and all that happens, and the AR comes into the picture. But what I [00:07:00] focus as my role with the vice president of clinical solutions, is towards the clinical efficiency and productivity and how can make the system more intuitive and user friendly for them is where I focus most on.

Jeff: You talked about kind of putting yourself in their shoes and thinking about it from their point of view. And we actually had a guest on a couple, couple months back now also named Deep De I just realized. But she's she's a VP of product over at company called Hunger Rush, and they.

Do you know, software for kind of franchise restaurants and, and things like that. But she talks about her and her team actually going into one of the restaurants and literally working a double and making pizza and working the drive through and working all the stations and understanding it. And that helped them make a better solution.

And in this case, you know, you were a doctor, you lived a lot of these problems and you can really empathize with, you know, like you said you have to read a, just a huge amount of case notes before you in and see a, a patient or you're always saying they're writing, writing, writing or taking notes.

You know, typing while someone's trying to interact with you [00:08:00] and talk to you and have, you know, a human interaction at a time when they feel vulnerable, they're, they're at a doctor. So let's get into it. Maybe 'cause it does seem like this world of living the problem can really, really help. How did you kind of take I guess, the high level idea of this is the thing we're going to do and what it needs to do, and kind of start to really get into the more nitty gritty of here's how we're gonna deliver it, here's what it's gonna look like, here are the constraints.

And especially AI brings all sorts of privacy, accuracy, all sorts of stuff.

Deepti: So how I approach this is I'm in that shoes. So I feel like what are the most common things that I do on a regular basis? What are those repetitive tasks? How much time does it take, for example.

Before I go into the exam room and just mind you, I'm not practicing anymore, but I used to for a.

So before I get into an exam room, I would be looking at knowing the patient first. What's the context? What's the history? Where, who, what am I talking to them about? What was their problems? What was the plan that I had given to them?

Was there any labs that were done, et cetera. So I had to go through the previous [00:09:00] encounters, bunch of documentation if they were seen somewhere else that I got their chart from another organization. And it would take me at least 10 to 15 minutes to go through their chart to prep up. Or I would have a nurse practitioner or physician assistant who would be actually calling.

Saying those things out loud to me before I walk into an exam room, depending on my schedule. Now, what we have done is with the help of our AI technologies and our models, we have launched patient summaries. It has brought the time to prep for the patient's visit before getting into the exam room from 10, 15 minutes to one to two minutes.

You could just click of a button, get the patient summary, get acquainted with what the problem could be, what's the reason for the visit, what you have seen them for, what was the medications they're on, what plan you had given previously. And then walk into the room knowing the patient, starting an informed conversation, knowing what they had been here for before, and then be able to follow up on the previous, uh. problems as well. So that really, really immensely helps them. So that was one of the examples of [00:10:00] what's the reparative most common task that you do? So we can at least remove that cognitive burden that they have going through two to three visits before looking at various documentation, clicking on bunch of screen links to get to that information, right?

So all that is removed from them so they can just. Get to the summary, understand the patient, walk into the room. And the second thing everybody does is because you're in an exam room, you are conversing with the patient and then making notes, right? That's the main part of the day. Reviewing patients charts and getting talk to the patients, understanding, giving them their plans, treatment, medications, et cetera.

And then walking out of the room and finishing the paperwork. So now you're doing paperwork at the time inside the room as well as afterwards. Which is the second frequent uh, workflow that they keep doing day in, day out. Right. So the second part is where we have our second solution, which is an ambient assist.

So when the providers are with the patients in the exam room, they can actually state the chart upfront, actually add to [00:11:00] using ambient to say, I'm gonna see this patient in the room. They are here for this reason. This is what we have discussed in the past. And then walk in, have the device running and not have to look at the technology at all.

They can just have a very organic conversation with the patient. Check in with them about their problem. Learn, discuss the previous plan, modify medications, plan treatment plans, whatever you have to do, and then walk out of the room. You can pause your ambient note taking at the time when you get out of the room.

You can actually add more if you wanted to, or then you can finalize the. Documentation right from there, review it, make sure that everything looks good, make quick edits, and then finalize, and it's seamless into the EHR. There is no copy pasting needed, et cetera. . So we took two big problems.

Solve them with AI and brought so much efficiency back into the day. For the providers, they sometimes save almost two hours plus during the entire day, and then they go back home five minutes after their last appointment rather than two [00:12:00] hours after their last appointment or taking their systems back with them, and then after hours, literally spending another hour or two to finish documentation at home.

Jeff: I don't know about people out in the, you know, listening, but on my end, you know, I, I had a habit of watching Grey's Anatomy and say, what you will about realism or not. I always thought it was funny that they showed, you know, all the doctors always in bed at night before they go into bed, just send their typing, typing, typing, putting in notes.

But I knew that was real because I also, my uncle was an ER surgeon. One of my best friends right now is an ER surgeon and several others are, are, you know, clinicians in other ways, and they all talk about documentation is just this silent killer of time. But you have to do it. But you're saving just a giant amount of time, which is great because, you know, either a doctors can get a little time back. B. You know, from the company's perspective, maybe they can schedule a couple more clients in there. Or but also just from a doctor patient interaction, that's so much nicer to have the person actually speak with you and be able to [00:13:00] talk and not have to like, go back and forth between the computer and spin around.

And then, you know, similarly on the other side, if their, you know, prep isn't 15, 20 minutes, they can actually come in, like you said, it reduces cognitive load. I would love to right now kind of dive into how some of this works. Let's talk about, I guess the, the summary piece first. We've all, you know, sat there and played with chat GPT and, 

hallucination is a real problem. Here, right, you might prescribe someone the wrong thing or you might. You know, go in and think the patient is afflicted with something different. What did that process look like to really dial that in somewhere where you need really high fidelity? 

Deepti: So, we monitor our models on a consistent continuer basis. We train the models with de-identified data, right? So it's a continuous training and learning loop. And also we put the provider in the middle of it. So we learn from providers behaviors, what they are changing when they're changing, and we seek continuous feedback.

There are a few measures that we have put in one, as I was mentioning, we continuously monitor and retrain. We look for hallucinations. [00:14:00] We, we, we actually feed the data back into the system. As I said, we train on the identified data, so nobody's. Being able to identify all these recordings that are ambient transcriptions that are happening are deleted after the, after the fact, and other thing that we are doing is we seek feedback from the providers.

There is a feedback loop, so let's say we generated a summary provider looks at the summary right there at the bottom. If they can actually say thumbs up, thumbs down, and provide some textual feedback to say, in this summary, this is what I feel could have been better. Right? We take that and then we train our model on that feedback.

So we are constantly monitoring that. Second thing we always say is, this is informational. So there are disclaimers put in the providers. Actually, if something looks off, do a double check, make sure that look at the document, and then they can give us the feedback. And on top of it, what we also do is we make sure that, let, let's say we were talking about ambient soap note, right?

We do nudges on the codes and meds, labs, orders [00:15:00] and diagnosis, et cetera. we show them what's suggested based on the interaction that happened in the exam room. We do not finalize until the provider signs off selects on those codes.

So, so there's and balances built in and all of that is very secure.

Um, continuously monitor and trained as well.

Jeff: Right. So where it comes to, you're kind of taking this unstructured data of the conversation or the kind of, well, I guess, the, the, the pre-care data is probably structured. You probably have fields that you're pulling from for that, so that's probably a little bit easier to ensure quality on, but you're taking like an unstructured conversation.

And I guess, do you have, you know. Is it one agent? Is it several agents got a kind of parsing to fill in these structured data fields that you're looking for, or how does that work there?

Deepti: So, it's a combination. We have several different agents. For summary, we have different one for amis, we have different one. We have also some smart suggestions on top of it as well. So what happens is there are templates. [00:16:00] So when the unstructured conversation is happening, there are queues, there could be macros, there could be some additional functionalities and features, enabler capabilities that can be used on top of ambient assist as well.

But at the base of it is there is a template which the providers have personalized, and once the conversation happens based on the cues, how the providers talk and how those cues happen, all the data goes into those specific structured fields as well. So they don't have to come and copy paste it is based on their preference and it goes into the soap note as the desired to go, which they can review and then sign off on the notes as well.

So there's a little bit of setup involved just to make sure that. This is the provider's preference because , every user has a preference as well, and we wanna make sure that we are, again, making the technology enabler and not a barrier for them

Jeff: Yeah, you don't wanna just make it, okay, now you get to take 15, 20 minutes rather than 15, 20 minutes writing down the summary. You get to take 15, 20 minutes reviewing the summary. That doesn't help anyone, but if you can, you know, a couple minutes reviewing the summary is probably well spent though.

Deepti: Right or copy pasting it to [00:17:00] the right sections, which is what we do for them, not, not make them do it.

Jeff: exactly. No, I mean, I, I think, I love the approach because it's one I've seen across a lot of different application types is. I mean, it's even how we operate here, which is, it's really interesting to see like the intersection of these kind of. Some technologies that are horizontal, like AI and the platform there, but then things that are, you know, very vertical in your case.

EHR software and on our end, a completely different thing, you know, monitoring applications and, and digital experience. But we kind of take a very similar approach, we're not prescribing medicine but we record the session. Our AI agents watch it, and we have a couple agents that look for different things.

Once it has it, and it kinda has the summary and has the, the actual kind of data points around the session itself, it starts to have a conversation with itself. It asks itself like, oh, what would be an issue? Where did the user have friction? Was there friction here? If yes. What would I call that? You know, and it starts to [00:18:00] fill out these very structured fields and areas.

So product managers can get that data and see it very easily in a very, you know, structured calculable way as they would expect to. But it takes the kinda raw, you know. In our case session data in your case, you know, conversation data and extracts that and, and puts it in that place.

But it's always interesting to me to see the kind of like similar approach across such varied different use cases, because at the end we are kind of just trying to distill certain data and certain structured data understandings from, you know, very nebulous data. What, what were some like big problems you ran into maybe early on?

Going through this, like any kind of funny issues you found where you people kinda realize that's, that's really not what we want. And and how'd you kinda look into fixing those?

Deepti: I mean, I would tell you my story as well. When I started thinking of using ai, I was like, oh no, I don't trust it. Right? I.

Jeff: Oh yeah. I mean, I'm not no different than anybody else, right? , and especially being a physician, right? The first thing that you [00:19:00] wanna look at is I can't prescribe anything, which is not right.

Deepti: It's about safety, right? So patient safety is paramount, and if you don't see the right information, you might end up making wrong decisions for patient care. So the biggest barrier that we see is building trust and making sure that we showcase with results. To our providers, to our users. What we are putting out there in the market is tested.

It's accurate. We have measured it. We are keeping an eye. We are always learning from what is happening. And then we put, as I was saying earlier, the decision making with the providers rather than making the decision for them. We are suggesting we are helping, but we are not making the decision prescribing medications for them.

So, so that was one of the biggest things. But , since I have started working with AI for the last so many years and trying to put those models out in the market, I feel like we have come a long way and the [00:20:00] adoption has gone. Far more from 30% or 33% earlier to more than 60% now, and going towards 80% in the next year or so, and providers opening up to actually adopt this technology.

One of the funny quotes from a provider I would tell you is when we launched this, she said, oh, there's no way I'm gonna use AI in my practice. I will never do that. And we encourage her to try with us a little. She did, and now if you go talk to her, she's like, you can pry it from my cold dead hands.

Jeff: Well that's, that's the proof point of something, you know, really viable is, is it's easy to turn people who are on the bleeding edge of all the tech stuff and say, Hey, try the new thing we've built. It's another thing to convert a skeptic and get them loving it. \ , back to the parallels, it makes sense where before you are prescribing anything though, like the AI is not prescribing, it's saying we think this is what you said.

I. Is that right? And that's, you know, kind of, I feel like across [00:21:00] the board in a lot of places, you know, that's still that last outcome piece is you always have to kind of have the human check it and, and say yes.

You know, make it so, which, which is probably a really good spot to be in. I guess looking at kind of the, how you guys are using it next gen, another issue that must have come up is the data privacy piece because you know. For a lot of companies, they can just kind of send the data to OpenAI or Gemini and have a process there.

There is a exponentially higher burden on privacy for, for companies like NextGen than anywhere else. How did you solve that problem? Are you locally hosting and running, you know, one of their models or what did that look like to solve the, you know, huge privacy concern there?

Deepti: So we have security teams, we have our chief security officers team here. We store everything in the cloud. All our environments are HIPAA compliant, and we are hosted in AWS. We make sure that we are not storing any data that is [00:22:00] not meant to be stored. Our legal teams constantly give us feedback and input on our practices.

We run all our processes by our legal and security teams to make sure that we are compliant we are adopting the right, standards and implementing the processes around security and compliance. Right? We are responsible ai based practice. So what we do is we make sure that if we are training our models, all the data is de-identified.

If we have transcripts, we are deleting the transcripts as soon as the written transcript is done. So all the best practices that we have. We need to use, we are complying to all those standards as well. So for us our teams are backing, constantly monitoring what we do, and we meet periodically, at least once a quarter to make sure that we are in alignment with all the practices for responsible AI implementation.

Jeff: , it makes sense and I wanted to bring that up because one, you know, part of what I've been able to do, thanks to, you know, the show is [00:23:00] travel around the country and, and we've had, you know, kind of several networking dinners with product leaders across, you know, all over the country.

This has come up kind of again and again is, oh, my company says we can't use AI because of privacy concerns or because we'd be exposing too much data to, you know, the, the AI provider. And you know, the kind of point I wanted to bring forth is, you know, NextGen is working with. Patient data, right?

This is, this is not even aggregated healthcare data. This is, you know, individual patient data. It might be a higher hurdle to, to do it safely and it should be, right? This is people's healthcare data, but it's doable. You can do it responsibly and you know if it's gonna bring the value, like in this case, right, you're saving.

Taking what doctor pre-visit prep from 15, 20 minutes down to one or two. You're saving them hours on note taking, you know, during and after the visits. This is a huge gain for doctors and yes, the companies too, but you know, patients as well. It's worth putting the work in [00:24:00] to understand that, but you know, it's worth kind of not just taking on face value.

Oh, we can't do it 'cause of privacy. There are ways to do it. You guys. It sounds like adhere to really strict protocols. Be safe. Think about all the data and how do you keep it incredibly, incredibly safe? But there are ways to do it. Like NextGen has figured it out and that's, it's great to see kind of pushing through what many companies would maybe say, ah, it's too risky.

We're not going to, but finding ways to do it responsibly too. If Deep and the team at NextGen can, can figure out how to process healthcare data this way. You know, your B2B SaaS is probably okay.

Deepti: I, I would tell you this is one of the questions and topics that we get raised anytime we are in a conversation in a room with A-C-T-O-C-I-O-C-M-I-O of a company, all of these questions are definitely coming up. So when we are talking to somebody about these AI solutions that we have, and we always make sure that we touch these points upfront.

So that they don't have to ask us. We are already in giving them the information.

Jeff: Yeah, I think that's a big part [00:25:00] of it too. Is, is, but that's a big part of a lot of things is can you deliver, you know, you build alt trust if you can. If you can deliver, you know, the answer that someone wants before they give the question, right? Earlier you answered the question before I asked it.

But also, but in like privacy data. You know, if you can practice, Hey, this is our plan, this is how we're doing it, this is how we've thought it through. You're gonna alleviate a lot of, you know, your security people's concerns. If you can show that. We did it, we really thought it through. Similarly, you know, on our end the goal has been, you know, how do we take data from not just sessions, but from across your product stack, synthesize it so you know, you're not having to dig through analytics data or session replay data to see how yesterday's I.

Release went we'll just deliver you the update of, you know, here's, we saw you did a release. Here are the notes that you put into linear or Jira. So we know this is the goal, here's how it's trending against that, and that kind of stuff. But really what AI allows you to do is bring this data forward proactively and kind of, you know, look a little magic, bring it forward before someone asks.

And, and, you know. You can kinda look [00:26:00] at an asked question as a, you know, an area to improve, to get ahead of it. I do have one Nick question before we move on from kind of talking about the summarizer specifically. Is it like an actual physical device? Like how does this work? Because I, I ran into a guy at a trade show a few months back now, and he had this like little. Box on his, for lack of a better word, a little like box on his shirt. I could tell it was a mic, but you know, he started to talk and then it turned out it was like this tool to record

trade show interactions. So you didn't have to take notes afterwards, which sounds similar, but like it, what does the, what does the device look like and how do patients react to that?

Deepti: Oh, so for us it could be as simple as provider's cell phone. 

You can literally initiate an ambient listing session on your mobile app on NextGen mobile app. As you walk in, you look at the patient's schedule and on your mobile app, and. You go in with the phone or iPad or whichever device you're using, you click on your ambient listening button and nobody has to do anything.

And you can put [00:27:00] the phone on the side, of course, you're gonna ask for consent from the patient. You could always disclose saying, Hey, I'm gonna be using a listening ambient listing. It's gonna take our notes, it's gonna record our conversation, and it'll help us facilitate this visit better.

So they start with that, and at the end of it, they can walk out with their device and. That's done. So no other extra device to be taken, nothing to be carried into the rooms other than the device that you're using for patients. 

Jeff: Which I wager is nice for patients where, you know, some people are already a little nervous by clinicians not having another. Thing to, to focus on, but you know, the phone is, feels very normal, so that's cool. Okay. I do want to move on because you know, product, there's more to product than AI is despite what I think LinkedIn conversations and even some of the conversations we have at, at dinner would lead us to believe.

But back to, you know, solving. User problems and, and really just bringing a better experience to users, which lies at the heart of what we're all trying to do [00:28:00] here. At, at a previous role, you know, we talked about earlier. Another thing that can really, really bottleneck is, is clinicians are one thing, but also as a patient, I have run into a situations so many times before where I go, I try and book time and justice kinda lag time between contact and when I can actually get in.

Is just so big that sometimes you know what, these glasses are four years old. I think at this point, because every time I go to schedule eye appointment, it's so far out, I'm like, you know what? Nevermind, I'll do it later. And so I have glasses that kinda gimme a sleigh headache and, and I, you know, they're not perfect.

They work fine, but, you know, but my contacts is prescription are terrible. But you, you have experience kind of solving that problem in a previous role. And I'd love to chat about that 'cause it's one that's very near and dear to my heart.

Deepti: Access to patient care, especially here in this world. It feels like at this time, at this junction where technology has made such advancements and we still cannot access care, what we need is healthcare and we cannot access that. So we identified that Problem was like if [00:29:00] even if I wanted to go to a doctor, I couldn't go to a doctor because I couldn't get an appointment up to three months in.

And so we started to solve this problem. We were like, what can we do here? How can we make it less? And by launching certain streamlined applications and features like centralized scheduling, where the scheduler had eyes on everybody's schedule, every location, all the timings that they were working their off, their, non-working times, their vacations, et cetera, they could just find based on the patient preferences. So having a combination of a portal where the patients could say, Hey, this is what I would prefer, having a centralized scheduling application where they could see what the patient's preferences were and actually be able to offer them times based on their preferences.

And then the patient can pick and choose to what they wanna do. And in addition to that, reminding them, because there is a lot of times patients, I literally specifically forget what, when was their appointment, when. When are they supposed to go in? So reminding them a week ahead or two days ahead and being able to send them just a text to say, Hey, you have this appointment, and be able to send them [00:30:00] directions to the office where the appointment is.

Here is the location, this is how you can get there. Really helps in bringing the patients in faster as well. So at once we roll this out in in combination, some certain features, combination of patient portal, centralized scheduling, appointment reminders, and being able to fill the forms online. All of that streamlined applications at , one of my previous roles where the clients actually saw huge benefits in bringing in the patients.

And then the time from intake, which is the first contact for the patient with the practice reduced to 49 to 50% went from 11 days to five to six days. , 

Jeff: Looking at that kind of as an experience, like how, how do you identify the areas that are fixable versus not? I guess is would probably be the first thing I would focus on. 'cause some things just. Doctor capacity is, is kind of a fixed thing. 

But how do you start to analyze this problem to see like, where can we actually find levers of gain and, and how do we solve this in a real quick [00:31:00] manner that goes beyond just, oh, you need to hire more doctors.

Deepti: . So first thing we have to look at is what's the workflow look like? What happens in a bringing the patient in, they do the first contact point. They might be calling you for an appointment. If they called you for the appointment, the clinician schedules very quickly and understand when is the first available appointment three months down the line, if that's the answer, you are giving to everybody all the time.

Maybe you have a staffing problem, you have no slots, or are you looking at, we had slots, but I couldn't get to the slot because I had to go by provider. By provider. I only could look at one month at a time. I didn't have a search capability where I could give a combination of, here are the location, here are the providers, here are the reason for visits, give me the first available slot, so gimme 20 available slots, right?

So those kind of things. Do we have the right tools to enable the users to have an informed conversation with the patient to bring them in? If it's like I have to go schedule by schedule, provider by provider, forget it, will not be able to find an appointment. If I had a search capability where I could use a bunch of parameters to really [00:32:00] search the first 20 available slots, I could choose by providers, or I could just say, any provider, just gimme the first available slot so we have to walk through the problem and the systems first to say step by step what we can do.

Jeff: It is so funny because again, like not to always make it about, you know, the , horizontal ness of, of some of these solutions, but if you look at it, it's kinda the same thing as, you know, on the B2B SaaS side, we're always, you know, historically a problem was how do you. Route, you know, maybe someone who's interested to a rep. And there were a lot of solutions that, you know, used to, the way they would operate is they'd go, okay, come in, you know, pick a rep who they should go to, and then look at their calendar. The problem is, if that rep was on vacation the week the person was targeting or just was busy for a week, you suddenly are looking at just no time on this person's calendar.

Meanwhile, you might have four other salespeople who are completely free.

Um. You know, just reversing that work order of, okay, first of all, let's show you all the open calendar time of the people who you could meet with. And then you can pick what works and then we'll [00:33:00] assign you basically, if three people have that time available, then we'll assign you to one of those three.

But just reversing some of these workflow decisions actually has gigantic efficiency gains that I think go unnoticed sometimes.

Deepti: Aren't we all surprised to learn that similar problems across industries and we all try to solve them in different ways? based on just our conversation that we were having previously on how AI is used in what we are just doing podcast production, right.

Versus how AI is used in healthcare practices. All we are trying to do is solve the problem of making these applications user friendly and making the journey more efficient for whoever that user end, end user might be.

Jeff: It is really interesting to see a lot of, especially in the the world of ai, a lot of people are dealing with the same. Kind of problem. We can look to, you know, probably how, you know, you guys at NextGen looked at taking something like a conversation and distilling out , certain salient I pertinent points and probably goes the other way too.

You guys could look at how we've kind [00:34:00] of found ways to find what are the important things, what are the things you need to highlight, then how do you describe it and how do you make sure that's accurate and you're not delivering false positives or false, you know, kind of hallucinations there. Not to go all meta here, but this is kinda the basis of why we started this, this show here is because we found that so many questions that were coming up as we talked to product people do sit horizontally like that.

If you take the half a second to kind of like back yourself up and look at context you know, what you're doing in an e-commerce company. You can learn from a healthcare company or what you're doing on a B2B SaaS company. You know, we could definitely learn, like I said, from a healthcare company and vice versa.

So it's always cool to kind of have these. Conversations and be able to talk through, not just, oh, you know, if someone's looking for someone's in healthcare and trying to see who they should talk to you'd be a great resource to kind of network with. But that might even go for a ton of other industries where someone's trying to do this kind of distilling of unstructured data in ai.

And, and that would be a great resource even though you guys work in completely different industries. , I'd love to keep you and talk to you a lot more about some of this 'cause it's really, really [00:35:00] interesting.

I think we could learn from each other even more, but I don't wanna keep you all day. You have. Much more important things. Do you have like actual medical you know, assisting medical clinicians. But thank you so much for coming.

It was fantastic to have you on. Deep, deep. And you know, for anyone looking to ask questions what's the best place to reach out? Is it LinkedIn? Is there another place that works?

Deepti: They can reach out to me on LinkedIn. They can always reach out to our marketing team. We can talk more with through them as well. And I would say, I mean, I go to these conferences, seminars. I sometimes talk at some of the other organizations, so wherever you find me, do not hesitate. Okay.

Jeff: Sounds good. We'll have to, you know, hopefully we can hit your hit your area and get you to one of our dinners too, so we can meet, we can actually meet in person, so. Awesome. Well, it's great having you on. Thank you so much for coming on. We'll have to stay in touch and you know, I'd really love to see kinda how, how this AI stuff evolves on your end and what other areas you end up being able to, you know, help clinicians and patients in even further.

So thank you so much for coming on. It's a good blast.

Deepti: Thank you, Jeff, for [00:36:00] having me on the show. And I'm just gonna say, we don't all reinvent the wheels again and again, so we should just get inspiration from what others are doing and apply that in our fields.

Jeff: Love it.