Nobody Wanted Their Booking App. So They Built an AI Front Desk | with Thomas Wojtowicz, BOAC #76

0:00If people can't get through to you because it's the evening, it's completely reasonable that you're not answering the phone. But they don't care. They're going to go somewhere else to someone that does. The day of voicemails is gone. That's archaic technology. But out of hours, when otherwise it would simply ring and no one would pick it up, it's a no-brainer. So that's the other thing to consider, is being able to capture stuff which previously you couldn't capture at all. We're effectively an AI front desk, a receptionist for private healthcare. We built the products, we went to market and realized it was probably a bit rubbish, and then went back to the drawing board, listened to customers, do what they wanted.

0:30We've come across customers that said, oh, we're going to build this ourselves, and that fall, and I gave, we'll speak to you in six months. It's difficult, but you don't see practices going out and building their own PMSs, do you? It doesn't happen, but I think because of AI, people think that it's probably an easier thing to build. You give the world's fastest car to someone that can't drive, what are you going to do with it? Probably crash it. Their admin staff are just stretched. They simply don't have enough hours in the day to do everything they need to do, and just taking a slightly more automated task off of their hands just allows them to basically do their existing job better and more meaningfully.

1:03I very much believe that the AI or human thing is the wrong question. It's AI and human. The AI or the tech may have gotten it a long way there, but the human at the end made a big difference. It's never going to replace people completely.

1:23Hi, everybody. This is Jared, co-founder at Coherent Healthcare. Welcome back to another episode of The Business of a Clinic. Today's an interesting one because normally we invite healthcare operators, providers, sometimes investors into the space. We try and stay away from people who sell into the space because we don't want this to be a marketing channel, but it's also part of the conversation and it's an interesting part of the conversation given a lot of the things that we talk about on the show. And so today we've very excitedly asked Thomas, who's co-founded Book Solid, to join us. And maybe you can give a two-minute version of what is Book Solid so that we can bring people up to speed and that we can get going.

2:01Yeah, thanks for having me on, firstly. We're effectively an AI front desk receptionist for private healthcare. So automating phone calls, SMS messaging, website chat with the intention of automating those. kind of administration front desk tasks for private healthcare trainings. Got it. And where, in terms of the start, and it's been running for a couple years now, the start to the now, what was that journey like getting going? It was an interesting one, to be fair. So my co-founder, Oliver, his dad runs a gym, and basically they'd got some, to go way back, they'd got some developers to basically build a kind of a custom booking software for their gym.

2:38and about a week to go before launch they basically pulled out and all of us had said Oliver you're up you've got to you've got to build me something yeah so there he was he stepped in and in a week he built it and he thought okay let me see if I can sell this elsewhere basically so that's where the the product began and we thought okay let's basically build the kind of appointment software right we realized we were at least a decade too late for that but we know we learned quite a lot and within that kind of booking software there was this feature to automate conversations over WhatsApp and the scheduling process over WhatsApp.

3:11And that was the bit that everyone was interested in. We couldn't sell the rest of it at all because, unsurprisingly, people had it and it was better and we had nothing to offer above that. But the automated WhatsApp thing was really interesting and people loved it. So we thought, okay, let's just scrap the rest of the product and let's just double down on this. So that was the focus. And we thought, can we build an entire product around just the AI side of things? And this was probably two and a half years ago before maybe it was such an obvious choice now looking back it's like obviously that's what we would have done but back then it was we could we've seen how quickly ai has improved so it wasn't really that obvious initially so yeah so that's what we did we built out other channels so other messaging channels sms and stuff like that and we yeah we went to market and got some feedback and it was very much a classic sort of mvp startup kind of thing we we built the products We went to market and realized it was probably a bit rubbish and then went back to the drawing board, listened to customers, see what they wanted.

4:11And initially we launched through a listing on Tenacode. So they've got quite an open sort of marketplace listing where you can put whatever kind of product you want. And that was where we got our initial customers. So a bit of feedback from there. And that allowed us to focus on the healthcare. We didn't really know that healthcare was going to be necessarily where we were going to focus because they don't come from healthcare ourselves. But yeah, we got the good feedback there and we realized So it was a good opportunity to integrate with different practice management softwares and stuff like that.

4:35So we doubled down. And that was end of 2024. And that's when we brought the voice channel in. And we onboarded our first customer, our first paying customer, end of 2024. And that's where today's version of the product began. And then from there, we just doubled down. We kept investing into it, building new features, listening to what people wanted, fixing things that didn't work. But yeah, that was really the starting journey. Got it. I know when we first met you before this conversation, you said that a huge part of the product maturing has actually been the underlying models improving and your housing for those capabilities being very healthcare accessible, sort of fit for purpose, doing that right sizing.

5:18I am curious. I think a lot of people often think about, and in healthcare, they often say, oh, the models are getting better. The models are getting better. But I don't think people actually understand what that means because they're not necessarily building with it hands on. And I'd be curious to understand from your perspective, what has that evolution looked like? What have you seen change from out of the box performance capability from 2024 to 2026, which still needs stuff around it a la Booked Solid to make it work. Yeah. But the engine itself has gotten better without, I guess, without any of your investment per se, with sort of other investment.

5:54And how has that played out? Surprisingly, less of that has been the actual underlying LLMs. They haven't changed. They are changing quickly. But most of the development has been us working out how to actually use those LLMs and make them good at being an agent. And there's a huge amount that goes into that isn't just an LLM. That's a good question. Plug it into ChatGPT or Claude and it just works. There's so much infrastructure on top of that to make sure that conversationally, especially in phone calls, like verbally, you've got things like latency to consider you've got so many considerations so making sure each of those models is used as well as it can be that's probably where most of the improvement in the core tech has been seen over the last like year as opposed to just new models coming out basically that is obviously helpful but often new models come out and they're still not quite necessarily ready because they're new models basically and sometimes it takes different iterations of those models to become good.

6:47So yeah, most of that has been us working out how to use the models and turn them into an agent really well, basically. Got it. It's interesting because I think every now and then we encounter a healthcare provider in the work that we do where they will say, yeah, we do that ourselves. We build it ourselves. What, the AI agent? Whatever the thing, insert blank here. And I think our view is actually quite similar to what you just said, which is the underlying LLM might take you 80 % of the way there, but that last 20 % is very hard and needs a lot of very technical depth, constraints, understanding, so on and so forth.

7:25So I very much agree. And I think for us, one of the things that we always ask in response to, oh yeah, we just did it ourselves. What did you do exactly? Tell us exactly what you did. And if the answer is, oh no, we just did a chat GPT. That's usually the answer. One of them's a bicycle, one of them's a four by four. They both wheels, they're not the same thing. So I think that makes a lot of sense. So, okay, so you've been on this journey now where the models have gotten better-ish, but really the journey for you on a product level has been how do you control, constrain, focus, direct, and actually anyone building an AI in healthcare more generally is really thinking about that.

8:03What's that last 20 %? How do we get this thing that can do so many things to do this one thing or two things really well? yeah i think there is definitely this it goes back to the fact that it's such new tech but there is definitely this whole thing where people think they're just building with a is just easy because it just doesn't always like the amount of people that i speak to and they're like oh you can just you don't need to be a developer anymore you can just and it's a tool fundamentally it's a tool some people will be better at using the tool than other people right if you give use the bicycle analogy you give the world's fastest car to someone who can't drive what are you going to do and I'll probably crash it, you know what I mean?

8:36So, yeah, there's a huge amount more that goes into it. And, yeah, like you said, we've come across customers that said, oh, we're going to build this ourselves, and I thought, okay, we'll speak to you in six months. It's difficult. It's really difficult. And it takes a lot of, yeah, a lot of learning. But also, you can build a lot of things with AI now. It would probably be easier to build something that didn't have AI in it, right? But you don't see practices going out and building their own PMSs, do you? It doesn't happen. But I think because it's AI, people think that it's probably an easier thing to build.

9:07But it's really difficult to get it actually good. You could make it clunky enough that it could probably do things. But making it really good is, that's where everything lies, right? So yeah, it's challenging, but that's the benefit of building in-house. Obviously, you get a lot of agent builder tools out there, like sort of wrappers, which you can use to fast track that process. And a lot of companies out there probably do use those. We've always stayed away from that because we felt that we can build a better product. in-house, not just that we can build a better product, but with all the right resources, the best product is always going to be the in-house one versus the kind of outsourced one.

9:41Because the providers that build these outsourced kind of wrappers, they're so general for different industries and stuff that there's so much more that goes into actually making it. We can fix bugs incredibly quickly. We can ship new features incredibly quickly. We have things like context injection, which you can't get through all these kind of wrappers and stuff. So that's the benefit that we've had of building it in-house. and yeah but yeah it's a complex process when we get when we began it was easy next week we're live we were live but there's a lot of we had a bit of customer churn at the beginning because the core tech wasn't good enough so yeah we had to really work hard to to make the core tech beyond the the shiny features and stuff actually good basically at doing its job that's the most important thing makes sense so okay got it so i understand where we're on the journey we've gone from the beginnings.

10:29It wasn't really working. It got better. It started working. Now we have a customer base. We have some consistency in the product. We've crossed the chasm of maybe the thing works to now we have a bit more reliability, confidence in the thing delivering on the promise. And that's where we are now. So I think what will be very interesting and I think has come up in different episodes that we've done with different group operators, group leaders, who will say that one of the biggest challenges for them is figuring out how to make AI work in their organization. There's obviously a technical answer to that, but I think it would be more interesting to maybe focus on the non-technical.

11:08It would be very interesting to understand in your journey and from what you've seen, what are the things, if I'm a group COO, group CEO, I'm thinking about how do I bring augmentation to my organization? What are the things that, what are my top five things that I really need to pay attention to. Check these five things before you unplug the fire extinguisher. How do we know that an organization is positioned for success? What does it need to implement to a high standard? What have been your learnings around that sort of AI implementation story? It sounds like the product development side, but then it's got to take hold.

11:50It's got to actually be brought into the organization. People need to learn how to work around it, work with it. There's a whole learning curve around the implementation piece. What have you found that might be relevant for others looking to do AI implementation more generally in their groups? I think the most important thing is it really highlights any kind of, I don't want to say operational cracks, but any kind of inefficiencies elsewhere in the kind of tech stack. So the PMS obviously is the thing that we plug into and there, if there are right ways and wrong ways to set that up. And we've spoken about this before, but it's some people have an absolute spaghetti on their PMS and it tends to be the larger organizations that we work with that have much more efficient, simple workflows on there.

12:33The way their appointment types and their practitioners set up is really neat and tidy. And then you have the ones where it's like loads of different appointment types that haven't been like archived and, you know, they might have the practitioner name inside the appointment type instead of assigning them properly through the way it's meant to be done. so that is really important because that just gets reflected over basically so we try and the way our system works in terms of getting availabilities is we basically map what the PMS has so it basically is one less thing that the AI needs to remember and get right and it's such an important thing that given it's already mapped out on the PMS we'd rather just pull that over but then if they have a weird way of setting it up on there then obviously that reflects over as well so that's one of the big things and yeah like I said the bigger customers that we have they generally tend to be a lot more efficient there and you can see why they're as big as they are because it's so simple from a patient perspective to book in and to that whole kind of patient intake process is just so much smoother and then you have the ones where it's like the instructions and the prompts can get out of hand because this person only offers disappointment type unless it's this type of patient and it's just you know what I mean so that can make things more challenging So I'd say that's probably the main thing really is making sure that everything else is simplified.

13:49And that's not just to make life easier for the AI, but that's also a reflection of every other kind of friction point in the clinic, you know what I mean? So that's probably the main thing. So it sounds like there's like a hygiene factor of in order for this thing to work, the foundation needs to be clean tidy usable not hyper custom one-off everything i mean within reason our platform is very customizable and obviously we're not going to sit in and say every clinic is the same they're very different and we have to have functionality to support those differences but then there is also rights and wrongs for example where you know we were speaking to somebody once you had a different practice management software for every clinician there's a whole different account instead of just adding a practitioner and it's like that's that that's the extreme things but it does just show that and actually i remember they asked us how does this work with your other customers then and i was thinking not again not like that yeah so there is like you said a hygiene factor but obviously things are going to be different so you know we have the ability to add custom prompts to the back end of every account that's really handy for kind of catching those differences and any kind of behavioral things which customers need so that's really useful but yeah there is there are rights and wrongs of the PMS and it tends to highlight those very quickly and obviously that does affect how easily people engage with the AI basically.

15:17It's funny I'm thinking back a customer that we worked with it was a sort of small-scale private hospital and very nice organization but when we started working with them we came to discover they had probably 160 different service lines many of which I think had been written in once when the appointment was scheduled and never reused again. Yeah. And the practitioners weren't available here and they weren't available there and they were available on the third Friday of the month, but not the first Monday of the month. And this one with the lunch break needed to be four hours on this day and three hours of that.

15:46It was impossible. It wasn't human readable, let alone machine readable. That's the thing, right? That's what it is really. It's the same really for humans. And the whole, it should be set up to make the patient intake process as smooth as possible, whether it's through AI or not. The same fundamental applied, really. But I think when AI is handling it, it's just people probably tend to assume that everything's the AI's fault as opposed to maybe looking at things and going, okay, is there a way we can make this better? You know what I mean? So it's interesting that the idea of fault is a very hot topic in general right now in the world.

16:19In life. In life. In life, in the world. I think for many reasons, and I think this idea of fault or a nicer way of saying that, accountability,

16:30when you use old non-AI software, use software in general, if something goes wrong, it's a bug. And you report the bug and the software vendor fixes the bug, hopefully. And then they apologize, sorry, we had a bug and this affected your account in this way. And now it's resolved. And assuming that they're telling the truth and that they've resolved it correctly, it's now resolved. And the likelihood of the possibility that bug shows up again is effectively zero unless someone goes in and does something very stupid, which can also happen. So I guess it's not zero. So that is also possible. What we have found in our work, and I'm curious how it shows up in your work, and likewise for clinics that are implementing AI either in front office or in back office, really I think anywhere that's patient facing, which is I think a little bit of a different implementation than doing it in a part of the clinic where the extent of the impact is the wrong medical supplies are purchased or the wrong account is reconciled, which are big problems, but it doesn't show up in the patient experience.

17:29And how do you think about this idea of accountability for a system that you can try and direct and point and hope behaves and probably does behave as expected, but may on occasion not? How should a health care provider think about buying something or using something that can't quite give the same 100 % certainty? maybe can give 99 % certainty, 95 % certainty, but there is the possibility of unexpected behavior with a technology provider. What is the sort of tension there? How do you think about that? Do you articulate that to customers? Is it just something that they need to expect going in? How have you load balanced the 9.5 out of 10 times is it's going to hit the thing on the head, but 0.5 times it might be slightly off piece or might go slightly this way, slightly that way?

18:23Or is that not something that providers need to really worry about because that's where the human comes in? What is the how do you think about faults and accountability as it relates to AI product? Good question. I think the assumption is often that humans are always right, especially when comparing with AI. And so AI is much more heavily scrutinized, very scrutinized. So people will watch it with a really close eye and the tiniest of things they will raise, which I guess is good and bad. It allows us to keep an ear out and get good feedback, but it also makes our job kind of more challenging in terms of making it as good of a perfect experience as possible.

19:03But yeah, I think that there are mistakes and there are mistakes, right? So the mistakes that it will make are the mistakes that a human can make. Tiny little conversational errors do happen. It's just the nature of AI. And everyone's been using any kind of AI, LLM or whatever, and you would experience small little things. and you'll correct it and stuff like that. And it'll go, oh yeah, you're right. That's just part of life, quite frankly. We do try and obviously improve that process as much as we can. But I'd also be lying if I said that has never happened before. So there is that. But then there's also the big mistakes that people worry about.

19:38These are the ones that people are like, should I implement it? Especially with all the kind of every week in the news article, one of the anthropics, new models, and everyone's... Jail breaks it and goes, who knows what it is. Yeah, that's it. I'm not using it ever again and stuff. and obviously we have systems in place to ensure that those kind of big mistakes that happen for example people are often concerned about the ai having access to things like clinical notes and just exposing them to coolers and from a tech perspective in theory if configured in a certain way that could potentially be a risk we've gone by the route of let's not allow the ai to access anything like that in the first place so that even if it makes the mistake of thinking i'm going to exposed this for some goodness knows what reason, it doesn't have access to it anyway.

20:24So we have very strict kind of security controls in terms of when you're conversing with AI, basically getting information about that patient record, which in theory should always be the person calling or parent guardian of the person calling. But we have to assume that also it could not be, right? So we make sure that we don't allow it to have any access to any information that we wouldn't want to effectively share. So that prevents risk of any kind of serious stuff being exposed, basically, or serious mistakes happening by the AI. But small kind of conversational things, saying it's Thursday the 9th and it's actually Wednesday the 9th kind of thing, that can happen and it will correct itself and it will say, sorry, I got that wrong.

21:03Just as a human, it happened to me the other day. I was picking an appointment and somebody responded and they said, it was November the 19th and they said, oh, Thursday the 16th. And I said, it was actually the 19th. It happens and those kind of things will happen. They will happen less and less, probably to the point where it's going to happen less one day than it would with a human. But yeah, in terms of those big errors, then we've got the kind of guardrails in place to prevent those from happening. Now a quick break to tell you a little bit about Coherent Healthcare, the company behind the business of a clinic.

21:29Coherent helps fast-growing healthcare providers engage, support, retain and nurture their patients from intake to discharge. As you get bigger, your patient body grows enormously, and it becomes much more than just another hire, just another front office team member. It's no longer a people problem. You need technology that can scale with you. Coherent helps you do that by keeping every patient on track. In short, we're the agentic CRM and patient support platform for outpatient hands-on health care. To learn more, visit us at coherenthq.com. Now back to the episode. So it sounds like what you're saying is when one thinks about putting AI unsupervised in front of a patient, the thing you need to ask yourself as a health care provider is what are the red lines?

22:18In so much as if I'm a health care provider thinking about buying in certain technology or certain provider services, I want to know what decisions were made about ring fencing protecting certain kinds of information. as an example, that gives me the confidence that even if it totally wigs out, it only does so in a how bad can it get? The answer is not that bad. And I think that's, for me at least, it's an interesting question because I think a lot of providers right now are being assaulted by new services, new solutions, new things. And very often the demo is great and the problems come six months later.

22:55But I wonder the extent to which questions like that need to be asked up front by providers to qualify or disqualify people. But it's hard to know, right? As a healthcare provider, you're not necessarily, it's like going to a supermarket and reading a nutritional label. You don't really know what you're reading. You don't really know to ask the question, what is poly, no, whatever, triple, dehydrated. You want to ask the question, but you don't know if you're asking the right question. Exactly. So it sounds like you're saying one of the things that you should be thinking about as a healthcare provider when you're implementing solutions and when you're designing the infrastructure and security measures more generally is what information should be inaccessible to prevent the risk, much like you would for an employee.

23:34You don't want to give your employee that you just hired your bank account details. You want to give them the view-only permission. I'm not going to give you to them and say, please don't open this. I'd rather just not give it to you. Interesting. So it almost sounds like, and we would talk about this a lot at Coherent, the idea that when you're buying in certain AI-native products or AI-native services today, you should think about the fact that you're buying something. you're almost buying is it there are some elements that are a bit like hiring someone as opposed to buying something because you want to think about how do you give the person context how do you help them get up to speed how do you make sure that they do the right thing how do you monitor their work in the first instance make sure it's correct that whole iteration cycle with the product that you're buying is not it's a new behavior i think for buyers at least from what we see in the journey that you've been on and the clinics that you now support, can you, without opening the house and the garage and the front door keys and everything that goes with it, can you tell us the sort of two ends of the spectrum, the horror story you've had, wow, we totally messed that up, maybe early in the journey, it went really badly, and here's what we learned.

24:41And then similarly, the magical moment, wow, this provider without names, it unlocked massive growth capacity, like two ends of the pictures. I think oftentimes we talk about AI, everyone knows. So like, it's all upside, it's all good, but actually getting to the good takes time, takes mistakes. So if you go back to the 2044 days, what were the sort of, ooh, we really stepped in it and now we have to make some changes and now what is the, where is the sort of great success story for a clinic that's coming on now or came on most recently and sees that headroom, that uplift, that growth that they are ultimately buying?

25:20Yeah. We never really had any kind of big, sort of nothing big that went wrong as such. But in terms of kind of things that needed an improvement, yeah, as I was mentioning earlier, it was just clunkier at the beginning. There were more bugs. It wasn't as conversational. And I think also it's important to note that people are experiencing AI in their everyday lives so much more now. So they're getting used to the concept of it. They're getting used to the capabilities of it. And they're more confident with the idea of speaking to it because they're probably more aware of what it can do right whereas two years ago I think it was different it was still there and it was probably a more exciting thing on the stage right because it was so new but a lot of people would still especially older patients would still probably shy away from it and prefer not to talk with it and that was one of the challenges we faced because we could see how well it was working in some instances and then we could see just some patients just wouldn't want to talk to it but we've seen that massively improve over the last couple of years just the confidence that people have in ai in general is improving so much because people are using it people know how powerful it is when they use it in everyday lives so that was one of the early stumbling blocks in terms of scenarios it's gone really well it worked really well at scale our product is we've designed it to be really good at scale we've got some enterprise customers that have got kind of 20 plus sites and it's a massive part of their patient intake now it handles a huge number of calls things like reminders as well the whole kind of process and just works really smoothly.

26:49And yeah, that's probably the highlight I would say is when we see an action in a really large scale, it really highlights how powerful it is. It's slightly harder to see if it's like one man band gets a few calls a week. You don't quite notice how much it cleans up the operational side of things. But when it's a massive clinic, it's getting like huge number of calls every day across multiple sites. You can really see how much it offloads from the admin teams basically. So that's probably my favorite in terms of seeing where it really shines. Interesting. You mentioned admin teams, and that is maybe the natural segue.

27:27So I'm also going to take off my hat for a minute, put a different hat. So I'm going to express my own view very strongly, which is that I very much believe that the AI or human thing is the wrong question. It's AI and human. and we have seen in the work that we do many instances where because the interaction was human at the last mile rather than the first 99 % but at the end when it mattered when it needed to show up the right way for the patient the human was an essential part of that completion of the work if you will the AI or the tech may have gotten it a long way there, but the human at the end made a big difference.

28:10And I think we've seen that not only in the patient interface, but also in some of these feedback loops or some of these hyper customizations. The clinic wants it to be booked in pink or booked 15 minutes after this or some weird rule that doesn't show up, but everyone knows that it doesn't say anything. We have found that last, going that last 10 % to deliver exactly as the clinic would has been very powerful. Granted, it brings its own set of issues. But you use the word admin. You talk about admin teams, front office teams. One of the things I know is true, both in the work that we do and I imagine in some cases the work that you do, is that people in some ways feel threatened by the arrival of new things that do some of the jobs they were doing previously.

28:58I'm curious in the work that you do and the customers that you support and the conversations you're a part of, where does the human and or AI thing show up? Do customers lay off lots of people or do they repoint those people elsewhere? Actually, are people doing the same job, but they're doing more valuable work because now some of the lower value? Where and how does that show up for clinics on a day-to-day basis? Yeah, it's a good point. I'm with you, by the way, on the whole AI and people thing, right? It's never going to replace people completely. The whole point, really our kind of mission ethos has been the the human resource is the most important resource right and it's about being able to use that as wisely and as meaningfully as you can whether it's delivering care or whether it's doing something that ai cannot yet do over time we will see an increase in the proportion of what ai can do versus human but it will never eliminate it completely obviously there are certain things that i can handle and when it gets to a certain point there are obviously things that we've built into the platform to make sure that those are automatically bombed off to humans but also there is the ability for any patient to simply request to speak to a member of the team they can be either transferred or however the escalation is set up within that particular customer's account so that's something that obviously we support it's not like when you call your local call center or something and you try and get through and it's like you're on the phone for five hours we don't believe that are people have people been afraid of AI in general is new tech and laying off the whole staff thing.

30:28Yeah, every time new technology comes out, some people are always afraid. You know what I mean? That is just, that's part of the process of bringing out new technology. And just because people are afraid doesn't mean it's the wrong thing to do. Just as it means that if it doesn't work as well initially, it also doesn't mean it's the wrong thing to do. The point is we all see division of it long term and we're all working to try and make that the best it can possibly be. So I think it really depends customer to customer. Some customers, like I said, are the 41-man bands, and they might be doing all the admin themselves.

30:59They might not have an admin team at all. And for them, it just allows them to buy a little bit of their time back, which is super valuable for them. And then we have, on the much larger scale, the kind of enterprise national clinics. Their admin staff are just stretched. They simply don't have enough hours in the day to do everything they need to do. And just taking a slightly more automated task off of their hands just allows them to basically do their existing job better and more meaningfully. and it probably does allow the clinic to grow without having to hire as many admin staff per kind of patient supportive yeah exactly which is also a good thing for them it really depends on the customer type but in all kind of senses we find that it does improve the experience for the clinic whether they're handling all the admin the clinic owner or whether it's an admin team either way allows their resources to be distributed in a better way basically it's interesting i think for us and often we speak with clinic groups, larger scale operators, I think you're probably right on the smaller side.

31:53There is always a competing agenda. One of them is I want growth in revenue. I don't want equal growth in cost. I want the lines to do that. And what we often say is the idea of growth absent cost means that your delivery model needs to fundamentally change. And obviously, right now at least, you can't change your practitioner delivery model because unless you are grossly misallocating practitioner resources, you presumably need to have a clinician or someone in the room with a patient at that point in time. So I think there's a limit to how creative one can be. And so then the question becomes, how do you genuinely augment or reduce the speed at which your admin cost climbs in relation to new patient serve?

32:40And I guess what you're saying is that a lot of the business case, why does someone work with you? The answer is I want to grow my patient engagement intake stuff. And I don't want my team to trail that linearly. I want there to be divergence in those two lines. Have I understood that correctly? Yes. The other thing as well to consider is that obviously it's 24-7. And we're now entering an age where kind of people's attention spans are incredibly short. It's amazing. People's attention spans are generally virtually non-existent nowadays. If people can't get through to you because it's the evening, it's completely reasonable that you're not answering the phone.

33:16But they don't care. They're going to go somewhere else to someone that does, right? And I think now that we're moving towards a place where the patient experience is going to be expected to be 24-7 at the bare minimum, you know what I mean? So not having something there, at least in the out-of-hours scenario, is probably a questionable choice. Because, like, why wouldn't you? However you choose to allocate that balance during the kind of office hours, that can depend clinic to clinic. We have some customers that maybe the phone initially goes to the admin staff and then if they can't answer it, it goes to the AI.

33:50Or it might go straight to the AI. There's kind of a bunch of different workflows there. But out of hours, when otherwise it would simply ring and no one would pick it up, it's a no-brainer, right? So that's the other thing to consider, is being able to capture stuff which previously you couldn't capture at all. It would be like a voicemail or something. The day of voicemails is gone. Right. That's arcade technology. Got it. So it makes sense. Constance of Time, and you've been very generous with yours, we're now at a place where you are seeing people talk about the healthcare front door, the front door of healthcare.

34:25You're at a place where you're starting to chip away at different parts of that door. And now you've got web chat intake, you've got people calling in, you've got different ways of supporting people who are trying to access, executing by patient access. What does, where does this go? What is the, you're in conversation with clinics all day. One of the benefits of having providers join the show is that you're talking to 300 clinics, whereas most of the clinic operators who come to the show or thinking about theirs and maybe two or three or four of their peers groups, they're not necessarily thinking about the problems that many providers are trying to solve.

35:06What is the problem conversation that you're a part of, that you keep hearing? Providers are trying to solve for X. They're trying to address Y in their organizations. How is that informing the work that you guys are now doing to support clinics? I think there's, going back to my point earlier, there's the whole kind of ratio of what it does compared to what humans do. and even within kind of the sort of the basic admin stuff there's more that it will hopefully be able to do in the future as the tech kind of progresses so that I would say it's been our main focus for a long time and will be we envision it being the the one place to manage the communications there's all sorts of other communications as well that you can run through like the outbound stuff reminders confirmations all that kind of stuff and then beyond that I would say from a kind of a long-term vision i would say that the triage is a big one that's one that's still handled by humans obviously and being able to automate that process i think would be huge because once again it's not that's still intake it's still pre in the kind of the room right so that's another thing that definitely could be automated in the future i don't think we're there yeah not us but generally but i think that will be super meaningful because that's definitely a key task that takes up a lot of time and being able to do that accurately well yeah just improve the whole experience for patients and for the admin staff that's a big one i'll tell makes sense tom thank you so much for joining folks this is again tom one of the co-founders at book solid helping you do better front office things a range of things more things to come to the future Feel free to reach out, I'm sure, if I've been interested.

36:47But I think also a huge part of the journey that Tom has been on with BookSolid, that we're certainly going on a coherent, is this idea of how do you bring this new stuff to the world in a way that doesn't create massive disruption, that adds clear value to the clinic. So if you're thinking about going on that journey and just want to do coffees or chats about how to avoid making some of the mistakes that we've made, maybe you guys have made, I'm sure Tom would be very happy to hear from you. Thank you so much for joining. Thank you very much. I'm John. Thank you so much for joining us on this episode of The Business of a Clinic.

37:17We hope you had a good time. We hope it was helpful. If you can think of anyone who should join the conversation, feel free to nominate them, send them our way. We'd love to invite them to participate. If you like what you saw, give us a like, share, comment. We hope we'll see you again next time. Thanks so much.