Full Transcript:

Shruti Mehrotra: The process of insurance benefits verification is a significant bottleneck to functioning efficiently as a healthcare business leading to delays, denials, as well as frustrated patients and providers. In a world where instant information is the norm, we cannot afford outdated processes that hinder our ability to provide effective care. Today, we are diving deep into how generative AI, especially AI agents, are revolutionizing this critical aspect of healthcare revenue operations.

We'll explore how these intelligent tools are moving beyond simple automation to truly understand and navigate the complexities of insurance verification. We will uncover how they fit seamlessly into your existing workflows. Empowering your practice to streamline operations and maximize reimbursements. Joining us today are Kashyap Purani, co-founder of Aarogram and Priyank Thakker, co-founder and CTO of Aarogram . Both have an extensive background in healthcare tech and practice operations. The floor is all yours Kashyap.

Priyank Thakkar: Hey everyone, welcome to another episode of LinkedIn live event by Aarogram and looks like we have a very interesting topic. There are a lot of people I see who have joined who have signed up for this event. We're very excited to cover this topic with you. Okay, so let's get started. So, I'm co-founder and CEO of Aarogram and this topic is really interesting and so I look forward to sharing whatever we have gathered and what insights we have with all of you. Part of the interest of why so many healthcare providers are showing an interest in using AI agents for operational needs is that everyone has been hearing about AI agents and the generative AI and specifically its application in healthcare because healthcare is very important touches the lives of a lot of people. That is part of the reason why this topic is very popular, and AI agents are something you are probably hearing a lot about. And first it was generative AI and now it's AI agents.

So as we start maybe it would be a good idea to refresh our knowledge a little bit about what an AI agent or an agent is. So if you look at the you know I would give an analogy of you know like at an organization or company you know when a person joins it could be you know when there is a joining new joining they could be an intern or trainee and then they become the assistant then become like a experienced employee and so that's the kind of journey I would give as an analogy how AI is also evolving. So earlier there used to be bots that would do the jobs that you just ask them to do and they don't use their mind- it's just like you define the rules, you define the process and the bots will do it right so that was the early wave of automation we got like this bots got little bit smarter.

And then from an AI assistant, they graduate to become an AI co-pilot. So, co-pilot is something you know that will work with you alongside with you. in that case AI may not be the main pilot but can also act as a co-pilot. But now we are just at the you know at the frontier of the beginning of new era. I would say that in this context AI agent means that they are independent. It is a kind of agent that can take action, interact with the data with the software and get the outcome output that you want. So that's where you know we got new intelligence that has become powerful and with that intelligence now we are able to automate healthcare operational processes end to end - as it acts independently as well.

Although you could argue what could be the use cases. So, in our context let's start with the use case here in for example the benefits verification and we are comparing contrasting with the patient care aspect with the more human aspect. So we don't want to go into that deep debate right now like what AI agent can do and what humans are supposed to do but let's just focus on something like benefits verification, which is highly manual, highly repetitive, highly tedious. If you could automate it with AI agent, then I think no human should actually spend time on it because there are a lot of important things you have to do. You don't work in healthcare just to spend time on paperwork and phone calls and be put on wait by the insurance company. You work in healthcare to provide patient care and that's what you are supposed to do and this is where AI agents can support you.

So let's take a step back . I already kind of touched upon a lot of things about benefits verification but first let's see why is it important okay and why you know as a company we made a mission at least our first kind of a project to create this automation AI agent that takes care of this process. So this is something we have gathered by speaking with hundreds of healthcare providers, administrators if you were to define the pre-visit patient journey especially in outpatient context. So we have gathered a lot of problems that occur- some of the providers are dealing with certain issues more than the others but overall these are like most frequent things we have gathered so far from our conversations. And the very first thing  you have to do when patient comes to you either via website or referrals or walk-in like you have to verify benefits because as you know healthcare is mostly funded by insurance and patients would expect their insurance to cover a lot of things right but then that's where the problem arises because the insurance benefit terms are so complicated so forget about the patient but provider staff also don't have tools to actually make a sense of it a lot of times.

And that's the reason you know it's kind of first very first step a lot of things kind of break down okay because if you're not able to and here we're talking about the benefits verification at the procedure level, but at CPT level because without that you don't get clarity about the pricing responsibility and how much is insurance going to pay. So that's different from just checking whether insurance is active or not. So that's a big difference and a very important nuance that a lot of people miss especially the decision makers don't know this nuance if they are not involved into day-to-day activities of the practice.

So that's the reason you need that clarity at the procedure level and unfortunately there is no other system right now that can actually give you with that accuracy and may tell you the plan deductible or other things but to get the verification like you have to sift through all the information and if you are working in that role for 15-20 years then maybe yeah you can figure out but otherwise you end up calling the insurance company and that is the first step and if you don't get the benefits right in the beginning there's a chain reaction and it affects the price estimates, it affects how as a provider you collect from the patient up front, it creates problem for the billing later on, creates surprise bills.

Maybe sometimes you know you don't get paid by the insurance company because you didn't get clarity or there were upfront issues. So all kind of issues happen. That's why today we are going to focus on mainly benefits verification but in our other episodes we have touched upon the price estimations for patient services as well. In fact, we have the next upcoming LinkedIn live event next week, where we're going to talk about the price transparency aspect in more detail but for the scope of this presentation and this session let's focus on insurance benefits because it's a huge bottleneck.

There are other things also has to happen after benefit verification but let's focus on benefit verification that's where the first step you know if you don't get right it kind of breaks down okay so let's see you know because on the surface it looks very simple but as you peel different layers you will discover there a lot of intricacies a lot of complications so you have to figure out to get that right and how do you what's the current state current state either you go to portal go to your system everything you know will throw all the information at you okay you have to figure out and sometimes staff gets overwhelmed and if you get overwhelmed lot of confusion then you have to call to insurance company and you're put on hold . So those are the problems occur and you know I have actually tried calling insurance company and I was put on hold for hour hours that's my record time like I was once put on hold for two and a half hours. Right, so now think about it when you have hundreds of patients especially for a larger practice coming every day and then staff is stretched thin and stressed out- then you know how do you deal with it?

And this is actually a real life example. I got a screenshot where you know you go to a payor website and like this is all I described to you and then you have to figure out you know what applies what doesn't apply what is a deductible co pay, co insurance, out of pocket maximum and this is also actually one of the better examples . I've seen other examples when there is no structured data, there is missing information and there are different options and you are on your own to figure that out. And so that's not a good situation.

So what if automation could help, and if it could help then why hasn't it been automated yet right? My answer to this is like there were definitely technical challenges and we have to deal with however a system works We can always imagine the ideal world scenario when the system is very seamless, your data you get right away from the insurance company but unfortunately we're not there yet. So, in the meantime what do we do, then that's where we dive deeper into and started automating these things and our goal was always been to make it as simple as going to restaurant and getting a menu where you get the upfront pricing offer to you even before setting up appointment. But you can't get there if you don't have the accurate benefits and that's where  we try to solve and we have been solving it we are on it for last one and a half years and we just even today, discover lot of nuances, lot of complications and these are some of the examples for preview like for the same patient same plan. If the patient goes from one provider to another provider, benefits will change, the fees will change, the estimate will change, right? why? Because it depends on a lot of factors.

For example, you know, if you take the provider's profile that is a service categorization, how are you categorized as institutional, professional, facility, what is your network status? And there could be some variations, tier one, tier two network status, place of service. Some insurance companies have other rules. Some insurance companies have different rules. Sometimes you know there is a different place of service for verification. There is a different place of service for billing. So how do you deal with all these things?

And you know remember when you get like out of the box solution from practice management system, they don't do any customization for your profile as a provider's profile. Okay. So then how do you get even accurate benefits, right? If they they don't do any customization, they don't do any finetuning for your profile, then how are we going to get that accuracy? And that's where it's very important to actually do that. We actually have a very robust onboarding process that goes on for one to three weeks depending on the complexity. And we also look at the historical claims but also we have our own rules and library and the knowledge and the data training plus on top of it we all can also use provider's own knowledge because provider is in the business for 10-15 years and this they have gathered a vast amount of knowledge that is very difficult for any human assistant to you know gather and apply on the fly but machine can do it.

And how can it do it? Well, we  have built an AI agent and for any AI you know get to level benefits like you need to consider all different. So these are the only some of the key parameters you have to get it right you know to get to the exact accurate benefits. Okay. So and even within this if you go deeper into it like there could be other things you to discover, right? and so how do you deal with all this? well, now luckily, you know, there are AI agents like us, ours who can actually handle this. Okay. So, so we have built an AI agent. And when I say when we say agent, that means we're talking about the human level intelligence or at least in the context of this benefit verification where it can take care of the whole process end to end. And how does it take care because we have the whole library of benefit mapping based on different variations in the plan and the different plan type but then we also gathered the subject matter expertise that we feed into AI and AI is trained on you know we have process like thousands of patients and estimates and verified thousands of patients so that all goes into it and then we find for particular provider's profile apply specific rules and as I said before we also utilize the provider's knowledge own knowledge base as well to figure those things out and our system is so sophisticated that it is actually now possible to handle all the variations limitations exceptions and when we got started it was we didn't we were not sure or whether we're going to be able to do it automatically but you know luckily like there's so much break so many breakthroughs in AI models and now you have all the software technology that are actually making it possible that was not possible earlier so that is why it's very exciting that you know we are actually doing it and here's the agenting workflow that I want to share quickly. And when say agentic workflow that means you know as I said the agent means like it is taking care of the whole process end to end independently. You could call it one agent doing multiple task or there are multiple agents working separately and they're orchestrating their action and getting the input output and to get to the final conclusion. So as you know like the getting their benefits from insurance is not the hardest step but There are a lot of other steps like determine the network status, service coverage, benefit attributes, apply fee schedule, calculate estimate and that's how you get to the final point, right? And now we are able to do it within seconds and we have a little bit of preview as well as we go.

We have a little surprise for you. We're going to share something that will make this session interactive. So that you can get firsthand experience as well.

We seamlessly integrate our AI agent with your EHR system and you know first thing you manage entire patient workflow in your EHR. So we fetch patients from your EHR put it into Aarogram revenue cycle management and through that we connect with the payers from which we get a lot of raw benefits which you're supposed to get by calling them or logging into portal or using third party tools.

So this is our one of the customer facing interface where you can come in choose your insurance company put insurance plan. The plan is kind of optional but yeah let's go ahead choose the state first name last name date of birth. So this is where a patient could come in and put all their information and say okay I'm the insurance holder. It's HIPPA compliant. and we click accept and continue. We're connecting with the insurance companies and getting the benefits. One of the key information we get here is insurance is active. Second key information we get here is is the patient's insurance in network or out of network for your practice. Third is what is the plan type? Is it PO? Is it HMO? What is it? And what are the services that you provide are available to the patient? That could be five services, that could be 50 services. It's all highly customizable. Then patient can click on one of these services and see what would the benefits and estimates are. You could see I have selected a service which is consult and these are all like programmable based on your provider profile and you can see the benefits are verified automatically at the procedure level and then this estimates are also calculated and shown here. Right? So all these benefits came in real time from insurance company within seconds and we calculated the estimate using the appropriate fee schedule for those services.

So, if you have a fee schedule for Cigna that is different from Aetna or Medicare, we account for all those different fee schedules. If you have a fee schedule for in office or online consultation, we also account for that. You have a fee schedule for in network and out of network, we also account for that. And if there are any benefits that are individual level or the family level, primary insurance level or the secondary insurance level, we marry all those benefits and bring it up for the patient in real time within seconds.

So they can see what is the cost that's paid by the insurance company for their service and what they owe to the practice for this service. So in this way this AI agent within seconds gives you all the answers about the cost per appointment or per service for that patient regardless of the insurance company regardless of the insurance plan regardless of the services covered.

In this case AI agent is kind of behind working behind the scenes.

So it's may not be visible. So this is just one use case which is patient facing. But then we have other variations as well because it's just in simple terms input output. So we have provider interface, we have API interface. So AI agent you know does it work behind the scenes and then uh you know you can interlay any kind of UX on top of it for your requirement, right? So, we can build any workflow with this system. The workflow could be patient, you know, uh just going on your website and checking the benefits or checking the estimates if you want to share the estimates with them or your staff going into EHR companies and clicking a button. Get benefits from our program and our agent will give all the benefits in your EHR within seconds. And we connect with all major EHR companies such as Epic uh you know Cerner, Athena, E-clinical works uh and so on. We have also integrated with Salesforce um and some other companies. Any questions?

Questions:

Is it hard to train healthcare staff on AI tools and AI agents?

Answer: It's a great question, right. The good thing is these AI tools are your own tools. Your EHR or your website? So, it's actually AI that is learning from your staff and AI is being trained by your staff. It's not the staff that needs to be trained. It's a simple click of a button within your EHR or a simple click of a button on your website. That's why we are calling them agents because as na agent, they work alongside your staff.

What is the biggest blocker for adopting AI agents at healthcare businesses?

Great question. Yeah, the biggest thing is that the staff aligns with the leadership or the outcomes of the project. For example, one of our customers, their staff was sensitive about losing the job to the agent, which is not the case when they had the right alignment within the leadership and a good understanding of what AI is and isn't capable of. Leadership guided the staff  by using this AI tools, their per patient effort reduces from minutes to seconds and they can reutilize their workflow in enabling new use cases such as collecting upfront patient, reducing the overcollections, reducing the under collections, reminding patients and adjusting their appointments at the right time with the right provider. So, there are a lot of use cases that open that enables provider practices to experience more price transparency and profitability.

One of our clients, with a large patient volume, previously dedicated significant staff to benefit verification and estimates. After implementing Arog, they saw an 80% reduction in this workload. This allowed their staff to address pending prior authorizations, which significantly decreased claim denials. Consequently, their prior authorization rate increased due to the freed-up staff. This demonstrates the real-world opportunities AI agents can provide. This is a recent example I wanted to share. Thank you.