Healthcare’s Agent Bytecode: X12 as the Harness for AI Agents — Vasant Kearney, Onlay

>> Hey everybody. >> Hey everybody. How's everyone doing today? How's everyone doing today? How's everyone doing today? Good. So, this is a bit about my background, So, this is a bit about my background, but I think it's always really good to but I think it's always really good to but I think it's always really good to learn what the audience background is if learn what the audience background is if learn what the audience background is if it's more on the technical side, which I it's more on the technical side, which I it's more on the technical side, which I know this conference is, healthcare know this conference is, healthcare know this conference is, healthcare side. Let's get a quick show hands to side. Let's get a quick show hands to side. Let's get a quick show hands to see who Who here is on the healthcare see who Who here is on the healthcare see who Who here is on the healthcare side? side? side? Ooh, wow. That's a lot of you. There's Ooh, wow. That's a lot of you. There's Ooh, wow. That's a lot of you. There's more than I expected. Wow. Okay. Who more than I expected. Wow. Okay. Who more than I expected. Wow. Okay. Who here is on the technical does the here is on the technical does the here is on the technical does the genetic workflows? genetic workflows? genetic workflows? Wow, okay. Overlap. All right. As should Wow, okay. Overlap. All right. As should Wow, okay. Overlap. All right. As should be expected at this conference. Who here be expected at this conference. Who here be expected at this conference. Who here has models running right now somewhere has models running right now somewhere has models running right now somewhere doing some work? doing some work? doing some work? Wow. Wow. Wow. It's like the whole audience. Okay. All It's like the whole audience. Okay. All It's like the whole audience. Okay. All right. So, I know who I'm talking to. right. So, I know who I'm talking to. right. So, I know who I'm talking to. Wonderful. Wonderful. Wonderful. This is the right crowd. This is the right crowd. This is the right crowd. So, what is the goal of this? Like if we're what is the goal of this? Like if we're working in healthcare and we're doing working in healthcare and we're doing working in healthcare and we're doing some genetic workflows, we have to keep some genetic workflows, we have to keep some genetic workflows, we have to keep in mind the goal.

in mind the goal. in mind the goal. And that goal, at least from my And that goal, at least from my And that goal, at least from my perspective, is to drive the overall perspective, is to drive the overall perspective, is to drive the overall cost down. So, cost down. So, cost down. So, in this talk will be about insurance in this talk will be about insurance in this talk will be about insurance cost specifically, the cost of of cost specifically, the cost of of cost specifically, the cost of of interacting with insurance. interacting with insurance. interacting with insurance. Um, but also to improve the patient Um, but also to improve the patient Um, but also to improve the patient experience because insurance relates experience because insurance relates experience because insurance relates back to the patient and how they back to the patient and how they back to the patient and how they experience the whole process. So, let's experience the whole process. So, let's experience the whole process. So, let's keep that in mind when we're solving keep that in mind when we're solving keep that in mind when we're solving problems. I know that if we're on the problems. I know that if we're on the problems. I know that if we're on the technical side, we like to get really technical side, we like to get really technical side, we like to get really experimental with things and get excited experimental with things and get excited experimental with things and get excited about just this the technology itself, about just this the technology itself, about just this the technology itself, but has to be grounded in one of these but has to be grounded in one of these but has to be grounded in one of these two concepts. So, a little trip down memory lane just So, a little trip down memory lane just starting from this long journey of AI starting from this long journey of AI starting from this long journey of AI machine learning and where we are today. machine learning and where we are today. machine learning and where we are today. And we're obviously going to talk about And we're obviously going to talk about And we're obviously going to talk about the last point, this the last point, this the last point, this agentic execution layer. But, we can see agentic execution layer. But, we can see agentic execution layer. But, we can see this evolution from the neuron this evolution from the neuron this evolution from the neuron convolutional neural networks convolutional neural networks convolutional neural networks large-scale unsupervised learning back large-scale unsupervised learning back large-scale unsupervised learning back in 2011 2012.

in 2011 2012. in 2011 2012. Then, the introduction of attention is Then, the introduction of attention is Then, the introduction of attention is all you need, one of my favorite titles all you need, one of my favorite titles all you need, one of my favorite titles for a paper, the introduction of the for a paper, the introduction of the for a paper, the introduction of the the transformer. the transformer. the transformer. Then, we go into this modern chat Then, we go into this modern chat Then, we go into this modern chat interface with these large language interface with these large language interface with these large language models. And then, finally with the models. And then, finally with the models. And then, finally with the Claude codes, the Codexes, Claude codes, the Codexes, Claude codes, the Codexes, and our system internally, and a lot of and our system internally, and a lot of and our system internally, and a lot of systems you have open claw, Hermes systems you have open claw, Hermes systems you have open claw, Hermes agent, all all that kind of stuff. agent, all all that kind of stuff. agent, all all that kind of stuff. Which really brings into the into the Which really brings into the into the Which really brings into the into the picture this execution layer. So, we're picture this execution layer. So, we're picture this execution layer. So, we're going to be talking about how to do this going to be talking about how to do this going to be talking about how to do this execution layer safely and reliably in execution layer safely and reliably in execution layer safely and reliably in health care. So, let's think back to some of the So, let's think back to some of the earlier examples of like getting really earlier examples of like getting really earlier examples of like getting really excited about some AI technology and excited about some AI technology and excited about some AI technology and then realizing it has all these little then realizing it has all these little then realizing it has all these little bits and pieces which make it a lot more bits and pieces which make it a lot more bits and pieces which make it a lot more trickier than maybe it is is obvious at trickier than maybe it is is obvious at trickier than maybe it is is obvious at first. So, like you have a check and you first. So, like you have a check and you first. So, like you have a check and you want to cash it. You want to deposit it want to cash it. You want to deposit it want to cash it. You want to deposit it into your bank account. Um you might into your bank account. Um you might into your bank account. Um you might say, "Oh, we have solved the handwritten say, "Oh, we have solved the handwritten say, "Oh, we have solved the handwritten digit problem. We can recognize digits digit problem. We can recognize digits digit problem. We can recognize digits from zero to nine." Right? Wow, oh oh, from zero to nine." Right? Wow, oh oh, from zero to nine." Right? Wow, oh oh, so now we're ready to so now we're ready to so now we're ready to um deposit this check into this person's um deposit this check into this person's um deposit this check into this person's account and transfer money. Well, not account and transfer money. Well, not account and transfer money. Well, not quite because as you dig in a little bit quite because as you dig in a little bit quite because as you dig in a little bit deeper you see that you have to identify deeper you see that you have to identify deeper you see that you have to identify all sorts of characters in the check.

all sorts of characters in the check. all sorts of characters in the check. You have to make sure it matches up with You have to make sure it matches up with You have to make sure it matches up with all these other pieces of the all these other pieces of the all these other pieces of the infrastructure. You have to make sure infrastructure. You have to make sure infrastructure. You have to make sure that it is um that it is um going to the that it is um that it is um going to the that it is um that it is um going to the target account that you're interested target account that you're interested target account that you're interested in. in. in. So, parts of this can be thought of as So, parts of this can be thought of as So, parts of this can be thought of as as the harness. as the harness. as the harness. Um in claims, we have a similar Um in claims, we have a similar Um in claims, we have a similar challenge that there's a lot of these challenge that there's a lot of these challenge that there's a lot of these little AI steps involved in fulfilling little AI steps involved in fulfilling little AI steps involved in fulfilling that whole patient journey of that whole patient journey of that whole patient journey of eligibility to getting the insurance eligibility to getting the insurance eligibility to getting the insurance company to deposit money in the company to deposit money in the company to deposit money in the provider's bank account. A whole bunch provider's bank account. A whole bunch provider's bank account. A whole bunch of little steps. And we have to make of little steps. And we have to make of little steps. And we have to make sure that we're safely doing this, that sure that we're safely doing this, that sure that we're safely doing this, that we're operating like in these tight con we're operating like in these tight con we're operating like in these tight con these strict confinements. So, another thing that comes up, just So, another thing that comes up, just sort of setting the stage, is that um sort of setting the stage, is that um sort of setting the stage, is that um multimodal multimodal multimodal context in context in context in comes up very frequently with claims.

comes up very frequently with claims. comes up very frequently with claims. So, you might have an image and it might So, you might have an image and it might So, you might have an image and it might seem like at at first for cost reasons seem like at at first for cost reasons seem like at at first for cost reasons or something else that you'd want to or something else that you'd want to or something else that you'd want to take that image and reduce it down to take that image and reduce it down to take that image and reduce it down to the findings, like here's the anatomy in the findings, like here's the anatomy in the findings, like here's the anatomy in the image, and maybe even extract some the image, and maybe even extract some the image, and maybe even extract some geometries from that anatomy. Here's geometries from that anatomy. Here's geometries from that anatomy. Here's pathologies. And then you would take pathologies. And then you would take pathologies. And then you would take that and then combine it with some other that and then combine it with some other that and then combine it with some other machine learning with some other data in machine learning with some other data in machine learning with some other data in a different downstream machine learning a different downstream machine learning a different downstream machine learning model. model. model. Like Like Like um EHR. um EHR. um EHR. And that might make sense from a cost And that might make sense from a cost And that might make sense from a cost perspective and also just like model perspective and also just like model perspective and also just like model capabilities. capabilities. capabilities. Um but Um but Um but in a lot of situations, in a lot of situations, in a lot of situations, it you lose context. So, it might be it you lose context. So, it might be it you lose context. So, it might be that you're extracting all this that you're extracting all this that you're extracting all this information and missing something that information and missing something that information and missing something that relates to some downstream procedure relates to some downstream procedure relates to some downstream procedure that you didn't that wasn't the upstream that you didn't that wasn't the upstream that you didn't that wasn't the upstream model wasn't aware of it. model wasn't aware of it. model wasn't aware of it. So, that introduces this concept of just So, that introduces this concept of just So, that introduces this concept of just multimodal processing. So, uh uh another multimodal processing. So, uh uh another multimodal processing. So, uh uh another place this comes up in healthcare, but place this comes up in healthcare, but place this comes up in healthcare, but not related to anatomy or anything like not related to anatomy or anything like not related to anatomy or anything like that, is desktop use.

that, is desktop use. that, is desktop use. You can see that sometimes, you know, You can see that sometimes, you know, You can see that sometimes, you know, things are buttons or things are buttons or things are buttons or or or, you know, it might have some or or, you know, it might have some or or, you know, it might have some shapes that are only obvious when you do shapes that are only obvious when you do shapes that are only obvious when you do this multimodal. All right. What is the agentic execution All right. What is the agentic execution layer? layer? layer? So, this can this can take on a lot of So, this can this can take on a lot of So, this can this can take on a lot of different forms. different forms. different forms. It is It is It is the ability of this model to take the ability of this model to take the ability of this model to take actions. So, it might be you're starting actions. So, it might be you're starting actions. So, it might be you're starting out with out with out with um um um database queries. And let's say it's database queries. And let's say it's database queries. And let's say it's just completely open. You're querying just completely open. You're querying just completely open. You're querying the database, you're finding your the database, you're finding your the database, you're finding your schema, schema, schema, you're figuring out what this what the you're figuring out what this what the you're figuring out what this what the data looks like, and then you might even data looks like, and then you might even data looks like, and then you might even have access to your code. So, then have access to your code. So, then have access to your code. So, then you're querying your code with the you're querying your code with the you're querying your code with the respect to your data. respect to your data. respect to your data. Uh and you might actually, in our system Uh and you might actually, in our system Uh and you might actually, in our system or other systems, you might be making or other systems, you might be making or other systems, you might be making insurance transactions. You might be insurance transactions. You might be insurance transactions. You might be making a phone call. You might be making a phone call. You might be making a phone call. You might be looking at a web portal. You might be looking at a web portal. You might be looking at a web portal. You might be interfacing with an EHR. These are all interfacing with an EHR. These are all interfacing with an EHR. These are all actions you can take.

actions you can take. actions you can take. And some of these actions have right And some of these actions have right And some of these actions have right implications. If you're interfacing with implications. If you're interfacing with implications. If you're interfacing with different PMSs, you're going to the different PMSs, you're going to the different PMSs, you're going to the desktop, you can have at least user logs desktop, you can have at least user logs desktop, you can have at least user logs at the minimum. at the minimum. at the minimum. Uh and then the next concept is memory. Uh and then the next concept is memory. Uh and then the next concept is memory. So, cloud code or codex, they use local So, cloud code or codex, they use local So, cloud code or codex, they use local memory, they write to your desktop. In memory, they write to your desktop. In memory, they write to your desktop. In enterprise healthcare, we can't really enterprise healthcare, we can't really enterprise healthcare, we can't really do this, so we do memory in a database, do this, so we do memory in a database, do this, so we do memory in a database, just so we have that logical separation. Uh important concept here is that when Uh important concept here is that when you're introducing new and improved you're introducing new and improved you're introducing new and improved better models more sophisticated more better models more sophisticated more better models more sophisticated more parameters parameters parameters you can't you can't just replace the you can't you can't just replace the you can't you can't just replace the model and assume it's going to be better model and assume it's going to be better model and assume it's going to be better it's different right it's a it's a on it's different right it's a it's a on it's different right it's a it's a on certain evals it's a better model as certain evals it's a better model as certain evals it's a better model as measured by these different metrics but measured by these different metrics but measured by these different metrics but it doesn't necessarily mean it's better it doesn't necessarily mean it's better it doesn't necessarily mean it's better right for all the situations that you right for all the situations that you right for all the situations that you want it to be better at because of the want it to be better at because of the want it to be better at because of the way you've designed your system so you way you've designed your system so you way you've designed your system so you really have to redo everything from really have to redo everything from really have to redo everything from scratch scratch scratch just make sure your evals your testing just make sure your evals your testing just make sure your evals your testing your validation is all set up so that your validation is all set up so that your validation is all set up so that you can introduce these new models and you can introduce these new models and you can introduce these new models and not break your system.

So So this concept of harness this concept of harness this concept of harness different groups have different different groups have different different groups have different definitions of this so I'm going to use definitions of this so I'm going to use definitions of this so I'm going to use a super broad definition here which is a super broad definition here which is a super broad definition here which is like all the different nuts and bolts like all the different nuts and bolts like all the different nuts and bolts that that surround this agentic that that surround this agentic that that surround this agentic reasoning and that is the reasoning and that is the reasoning and that is the concept of memory that we discussed the concept of memory that we discussed the concept of memory that we discussed the different tools the checks the different tools the checks the different tools the checks the permissions the handoffs the evals permissions the handoffs the evals permissions the handoffs the evals but also in the context of health care but also in the context of health care but also in the context of health care and claims it's x12. and claims it's x12. and claims it's x12. So just like we have these old school So just like we have these old school So just like we have these old school languages or formats like COBOL languages or formats like COBOL languages or formats like COBOL or other stricter maybe strict languages or other stricter maybe strict languages or other stricter maybe strict languages typescript typescript typescript llms really thrive they work well and llms really thrive they work well and llms really thrive they work well and when they're confined they have clear when they're confined they have clear when they're confined they have clear limited limited limited values that they can predict and x12 is values that they can predict and x12 is values that they can predict and x12 is exactly this so it provides this exactly this so it provides this exactly this so it provides this underlying structure this contract underlying structure this contract underlying structure this contract between what you're trying to between what you're trying to between what you're trying to communicate and the insurance company.

So, when you're reasoning in this in So, when you're reasoning in this in this healthcare, your your objective is this healthcare, your your objective is this healthcare, your your objective is to do something to do something to do something with handle a claim or research your EHR with handle a claim or research your EHR with handle a claim or research your EHR with respect to claims. It might be that with respect to claims. It might be that with respect to claims. It might be that you have like 50 steps or something like you have like 50 steps or something like you have like 50 steps or something like that. There's a lot of different steps. that. There's a lot of different steps. that. There's a lot of different steps. And so, you can And so, you can And so, you can um um um you at at each of those steps as you you at at each of those steps as you you at at each of those steps as you make mistakes, those mistakes can make mistakes, those mistakes can make mistakes, those mistakes can propagate down your system. Um and so, propagate down your system. Um and so, propagate down your system. Um and so, it's very good to have something it's very good to have something it's very good to have something grounded that can be rejected to. So, if grounded that can be rejected to. So, if grounded that can be rejected to. So, if you have a really strict got guardrails, you have a really strict got guardrails, you have a really strict got guardrails, you can reject something that happens you can reject something that happens you can reject something that happens that's incorrect. So, So, if you're reasoning over, let's say, the if you're reasoning over, let's say, the if you're reasoning over, let's say, the previous example, 50 steps, that and previous example, 50 steps, that and previous example, 50 steps, that and they're multimodal. You're considering they're multimodal. You're considering they're multimodal. You're considering and everything like that. That can get and everything like that. That can get and everything like that. That can get really expensive. It can also take a really expensive. It can also take a really expensive. It can also take a really long time. And folks might not really long time. And folks might not really long time. And folks might not want to wait. You know, it could be too want to wait. You know, it could be too want to wait. You know, it could be too expensive and people don't want to wait expensive and people don't want to wait expensive and people don't want to wait that long. And each time it each step is that long. And each time it each step is that long. And each time it each step is an opportunity to introduce an error and an opportunity to introduce an error and an opportunity to introduce an error and you can have problems.

you can have problems. you can have problems. Um but, if you hardcode your whole Um but, if you hardcode your whole Um but, if you hardcode your whole system, you say we're you're going to system, you say we're you're going to system, you say we're you're going to throw out this whole agentic process, throw out this whole agentic process, throw out this whole agentic process, you limit yourself or your code can you limit yourself or your code can you limit yourself or your code can explode to be just unmanageable. So, now explode to be just unmanageable. So, now explode to be just unmanageable. So, now all of a sudden, you just have this all of a sudden, you just have this all of a sudden, you just have this crazy bloat and you have to have this crazy bloat and you have to have this crazy bloat and you have to have this giant engineering team, which poses its giant engineering team, which poses its giant engineering team, which poses its own problems. own problems. own problems. Um Um Um so, what we want to do is strike this so, what we want to do is strike this so, what we want to do is strike this balance between what we should be balance between what we should be balance between what we should be completely free, completely free, completely free, like um like um like um with just pure agentic reasoning and with just pure agentic reasoning and with just pure agentic reasoning and execution, and what is hardcoded. execution, and what is hardcoded. execution, and what is hardcoded. So, we do that internally with So, we do that internally with So, we do that internally with introducing memory, just this uh partner introducing memory, just this uh partner introducing memory, just this uh partner level memory, level memory, level memory, organizational organized memory and user organizational organized memory and user organizational organized memory and user memory. So, we say if a user, we find memory. So, we say if a user, we find memory. So, we say if a user, we find people in in multi-site health people in in multi-site health people in in multi-site health organizations, they tend to do the same organizations, they tend to do the same organizations, they tend to do the same thing day after day. And it might be if thing day after day. And it might be if thing day after day. And it might be if they mention a few words, "Oh, they they mention a few words, "Oh, they they mention a few words, "Oh, they usually do eligibility and they usually usually do eligibility and they usually usually do eligibility and they usually do it within this context, they probably do it within this context, they probably do it within this context, they probably mean this."

mean this." mean this." Right? Where another user, they probably Right? Where another user, they probably Right? Where another user, they probably mean that. So, we want to be really mean that. So, we want to be really mean that. So, we want to be really careful here because as you introduce careful here because as you introduce careful here because as you introduce memory, you also persistent memory memory, you also persistent memory memory, you also persistent memory across chats, across days, across chats, across days, across chats, across days, you also introduce bias. So, maybe that you also introduce bias. So, maybe that you also introduce bias. So, maybe that person doesn't want to do the exact same person doesn't want to do the exact same person doesn't want to do the exact same thing that they did yesterday and now thing that they did yesterday and now thing that they did yesterday and now you steer them to do the exact same you steer them to do the exact same you steer them to do the exact same thing they did yesterday. That's a thing they did yesterday. That's a thing they did yesterday. That's a problem. So, you want to strike a problem. So, you want to strike a problem. So, you want to strike a balance somewhere in there and you want balance somewhere in there and you want balance somewhere in there and you want to make sure that the use any user can to make sure that the use any user can to make sure that the use any user can break out of this. So, for folks that are unfamiliar with So, for folks that are unfamiliar with the whole claim life cycle, it's many the whole claim life cycle, it's many the whole claim life cycle, it's many steps. So, each step steps. So, each step steps. So, each step does have an X12 correspondence with it. does have an X12 correspondence with it. does have an X12 correspondence with it. Starting with the schedule, when let's Starting with the schedule, when let's Starting with the schedule, when let's say you're showing up to the doctor's say you're showing up to the doctor's say you're showing up to the doctor's office before you even show up. That's office before you even show up. That's office before you even show up. That's insurance starts then. insurance starts then. insurance starts then. Um when you're getting treated, that Um when you're getting treated, that Um when you're getting treated, that also relates to insurance, what you you also relates to insurance, what you you also relates to insurance, what you you know, the different procedures that you know, the different procedures that you know, the different procedures that you are candidate for depending on your are candidate for depending on your are candidate for depending on your insurance.

insurance. insurance. Um your documents, sometimes the x-ray Um your documents, sometimes the x-ray Um your documents, sometimes the x-ray itself is the document and you would itself is the document and you would itself is the document and you would send proof of that in. send proof of that in. send proof of that in. Submitting the claim. Submitting the claim. Submitting the claim. And then finally getting the payment in And then finally getting the payment in And then finally getting the payment in the provider's bank. So, So, this I think this concept is a little this I think this concept is a little this I think this concept is a little bit bit bit I I I I found it to be obvious in retrospect I found it to be obvious in retrospect I found it to be obvious in retrospect but let me talk you through it. Maybe but let me talk you through it. Maybe but let me talk you through it. Maybe you find it's it's you find it's it's you find it's it's interesting or not but interesting or not but interesting or not but if you're calling an insurance company if you're calling an insurance company if you're calling an insurance company that it that boils down to a transaction that it that boils down to a transaction that it that boils down to a transaction an X12 transaction. You'd say hey this an X12 transaction. You'd say hey this an X12 transaction. You'd say hey this is the patient I'm talking about. Great. is the patient I'm talking about. Great. is the patient I'm talking about. Great. That's a like an eligibility request a That's a like an eligibility request a That's a like an eligibility request a 270. Oh I need to do you're requesting a 270. Oh I need to do you're requesting a 270. Oh I need to do you're requesting a claim status or whatever it is you're claim status or whatever it is you're claim status or whatever it is you're doing that has an X12 grounding. doing that has an X12 grounding. doing that has an X12 grounding. And that is the whole concept here this And that is the whole concept here this And that is the whole concept here this this X12 harness. So you call the this X12 harness. So you call the this X12 harness. So you call the insurance company you insurance company you insurance company you have an agent interact with a desktop have an agent interact with a desktop have an agent interact with a desktop you have an agent interact with the you have an agent interact with the you have an agent interact with the browser browser browser your imaging system that's a 275 your imaging system that's a 275 your imaging system that's a 275 and and and and and your insurance your bank your and and your insurance your bank your and and your insurance your bank your ACH so ACH so ACH so that's not X12 but it's still that that's not X12 but it's still that that's not X12 but it's still that structured structured structured transaction.

>> [snorts] >> [snorts] >> So this is just a reiteration of these >> So this is just a reiteration of these >> So this is just a reiteration of these different transactions. And the other beautiful thing about it And the other beautiful thing about it it's not you know it can you ask an it's not you know it can you ask an it's not you know it can you ask an agent to do something let's say you're agent to do something let's say you're agent to do something let's say you're genetically programming or let's say genetically programming or let's say genetically programming or let's say you're just you're you're programming you're just you're you're programming you're just you're you're programming how you know maybe half the companies I how you know maybe half the companies I how you know maybe half the companies I spoke with here spoke with here spoke with here still program today just everything by still program today just everything by still program today just everything by hand hand hand and they use these clock code or code X4 and they use these clock code or code X4 and they use these clock code or code X4 research. research. research. If you look up any of these transactions If you look up any of these transactions If you look up any of these transactions they're all public. Like this is not the they're all public. Like this is not the they're all public. Like this is not the beautiful thing about this is like this beautiful thing about this is like this beautiful thing about this is like this is not my schema. If you ask agents to is not my schema. If you ask agents to is not my schema. If you ask agents to make a schema for you you're going to make a schema for you you're going to make a schema for you you're going to get like get like get like all sorts of stuff. But now if we ground all sorts of stuff. But now if we ground all sorts of stuff. But now if we ground it in something standard you can look up it in something standard you can look up it in something standard you can look up all of these and you would know just all of these and you would know just all of these and you would know just right off the bat my schema. Let's say right off the bat my schema. Let's say right off the bat my schema. Let's say you're a new engineer coming in, like, you're a new engineer coming in, like, you're a new engineer coming in, like, you know. So, X12 is a So, X12 is a is is is a system of rules a system of rules a system of rules and and and it doesn't mean that when an insurance it doesn't mean that when an insurance it doesn't mean that when an insurance company gives you an X12, it's true.

company gives you an X12, it's true. company gives you an X12, it's true. So, So, So, that concept is that concept is that concept is when insurance company tells you when insurance company tells you when insurance company tells you something, it's coming from different something, it's coming from different something, it's coming from different teams potentially. They can have an teams potentially. They can have an teams potentially. They can have an engineering team that's It could be even engineering team that's It could be even engineering team that's It could be even a different company. a different company. a different company. A different company that the insurance A different company that the insurance A different company that the insurance company contracts out designed their web company contracts out designed their web company contracts out designed their web browser, their phone system, or their browser, their phone system, or their browser, their phone system, or their X12 layer, X12 layer, X12 layer, or their fire. or their fire. or their fire. And we have to understand that there's And we have to understand that there's And we have to understand that there's no ground truth. They also within all of no ground truth. They also within all of no ground truth. They also within all of these systems, they can they can all these systems, they can they can all these systems, they can they can all actually agree on the wrong information actually agree on the wrong information actually agree on the wrong information as well. Like, let's say they all say as well. Like, let's say they all say as well. Like, let's say they all say this patient is covered. All three You this patient is covered. All three You this patient is covered. All three You call them. You look in the browser and call them. You look in the browser and call them. You look in the browser and the X12 and they all say, "Yes, this the X12 and they all say, "Yes, this the X12 and they all say, "Yes, this patient is covered." And then you treat patient is covered." And then you treat patient is covered." And then you treat the patient, they say claim is denied the patient, they say claim is denied the patient, they say claim is denied due to due to due to the patient wasn't covered during that the patient wasn't covered during that the patient wasn't covered during that time. Um time. Um time. Um so, they can all disagree, but sometimes so, they can all disagree, but sometimes so, they can all disagree, but sometimes you'll learn some idiosyncrasies of you'll learn some idiosyncrasies of you'll learn some idiosyncrasies of these different payers that some of these different payers that some of these different payers that some of these systems are more reliable than these systems are more reliable than these systems are more reliable than others. others. others. But, regardless of if it originates as But, regardless of if it originates as But, regardless of if it originates as an X12 or not, you can boil all those an X12 or not, you can boil all those an X12 or not, you can boil all those transactions down to your own internal transactions down to your own internal transactions down to your own internal semi-correct X12. Correct until semi-correct X12. Correct until semi-correct X12. Correct until downstream evidence proves it otherwise downstream evidence proves it otherwise downstream evidence proves it otherwise uh uh uh to be incorrect.

So, just a little bit more on that. So, So, just a little bit more on that. So, and any of the X12, any of the and any of the X12, any of the and any of the X12, any of the information coming from the insurance information coming from the insurance information coming from the insurance company, any time can be wrong. It can company, any time can be wrong. It can company, any time can be wrong. It can be updated later. be updated later. be updated later. So, have fun. This is just an example of what it would This is just an example of what it would look like if you're um look like if you're um look like if you're um in that patient journey. You're you're in that patient journey. You're you're in that patient journey. You're you're trying to figure out how much you would trying to figure out how much you would trying to figure out how much you would pay as a patient up front, and it's very pay as a patient up front, and it's very pay as a patient up front, and it's very important for your experience going to important for your experience going to important for your experience going to the doctor. And then the different treatments that And then the different treatments that you have in that clinic can oop. Yeah. The different treatments that you Yeah. The different treatments that you have in that clinic can be the evidence have in that clinic can be the evidence have in that clinic can be the evidence that you need. Like you might get a that you need. Like you might get a that you need. Like you might get a CBCT. Well, that those images and slices CBCT. Well, that those images and slices CBCT. Well, that those images and slices of those images might be the evidence of those images might be the evidence of those images might be the evidence that they're asking for. So, ultimately, if you're delivering So, ultimately, if you're delivering that treatment, you're sending that that treatment, you're sending that that treatment, you're sending that claim, that claim is like a receipt of claim, that claim is like a receipt of claim, that claim is like a receipt of what you did. I did this, like here's what you did. I did this, like here's what you did. I did this, like here's the invoice. Right? Like you send it to the invoice. Right? Like you send it to the invoice. Right? Like you send it to the insurance company this invoice, and the insurance company this invoice, and the insurance company this invoice, and they would pay you back. So, that is they would pay you back. So, that is they would pay you back. So, that is your ultimate like contract of you're your ultimate like contract of you're your ultimate like contract of you're saying you did this work, it's sealed saying you did this work, it's sealed saying you did this work, it's sealed um and now the ball is in insurance um and now the ball is in insurance um and now the ball is in insurance company's court.

And just a little bit about this And just a little bit about this progression of the claim from you're progression of the claim from you're progression of the claim from you're sending it, you're getting some sending it, you're getting some sending it, you're getting some acknowledgement, has like the syntax is acknowledgement, has like the syntax is acknowledgement, has like the syntax is right with that 999. right with that 999. right with that 999. The status has been updated. Hey, cool, The status has been updated. Hey, cool, The status has been updated. Hey, cool, we received it. we received it. we received it. Um maybe you call them and you verify Um maybe you call them and you verify Um maybe you call them and you verify the status didn't come in. the status didn't come in. the status didn't come in. Then you have this EOBs 835 receipt of Then you have this EOBs 835 receipt of Then you have this EOBs 835 receipt of payment. And then And then we're getting to the end of this we're getting to the end of this we're getting to the end of this uh this talk here, but uh this talk here, but uh this talk here, but I think that, you know, LLMs, I'm fully I think that, you know, LLMs, I'm fully I think that, you know, LLMs, I'm fully AI pilled, right? But we want to make AI pilled, right? But we want to make AI pilled, right? But we want to make sure that we introduce sure that we introduce sure that we introduce la- la- la- language models, small tiny models in a language models, small tiny models in a language models, small tiny models in a very skeptical conservative way. So, very skeptical conservative way. So, very skeptical conservative way. So, being AI filled is great, being AI filled is great, being AI filled is great, but you should also be very AI but you should also be very AI but you should also be very AI skeptical. Like these things, they make skeptical. Like these things, they make skeptical. Like these things, they make mistakes and it's not even you can't mistakes and it's not even you can't mistakes and it's not even you can't even say they make mistakes. Like we even say they make mistakes. Like we even say they make mistakes. Like we make mistakes designing them. We might make mistakes designing them. We might make mistakes designing them. We might set them up to fail. So, we have to be set them up to fail. So, we have to be set them up to fail. So, we have to be very skeptical of them and we have to very skeptical of them and we have to very skeptical of them and we have to use them in a way that's also use them in a way that's also use them in a way that's also cost-effective. You can't throw I mean cost-effective. You can't throw I mean cost-effective. You can't throw I mean you don't you don't want to use an you don't you don't want to use an you don't you don't want to use an overpowered over expensive model cuz overpowered over expensive model cuz overpowered over expensive model cuz then if you're going back to if you're then if you're going back to if you're then if you're going back to if you're reducing costs or not. Let's say it's reducing costs or not. Let's say it's reducing costs or not. Let's say it's ends up being super super expensive to ends up being super super expensive to ends up being super super expensive to deliver one of these routine things that deliver one of these routine things that deliver one of these routine things that need to be done a thousand times a day.

need to be done a thousand times a day. need to be done a thousand times a day. You definitely don't want that.