Abhay Jajoo on CustomerInsights.AI, Agentic AI, and the Future of Life Sciences Commercial Analytics

Russ Fordyce 🏆 (00:01.248) Welcome back to the winner circle, everybody. I'm here today with Abhay from CustomerInsights.ai. They are revolutionizing how we kind of interact in these kind of customer experiences. So Abhay, welcome to the winner circle and congratulations on the AI Excellence Award. Abhay Jajoo (00:16.856) Thank you so much, Russ. Thanks for having me on this podcast. Looking forward to kind of explaining what we do. Okay. Russ Fordyce 🏆 (00:22.688) Yeah, so step back for a second and give us that kind high level elevator pitch on what you guys do. Abhay Jajoo (00:29.366) Yeah, so our company is called customerinsights.ai and even though it's kind of a very generic name, but I think we help pharma companies, life sciences companies, which also includes biotech, find patients, help patients get better in life, And customers for pharma companies is patients and doctors. So we actually help. companies, identify which doctors to go after to recruit patients, and then ultimately get the drugs prescribed and get better utilization for those drugs. Russ Fordyce 🏆 (01:03.734) Well, this is such a ripe industry for AI and I'd love to get your take on all of that, but you've spent, I think, more than 25 years inside of Life Sciences Consulting before starting this company. What did you see from the inside that made you say like, this whole model's wrong? Abhay Jajoo (01:20.846) That's a great question. So I started my journey in about 86, that's more like 40 years ago. My first job was with a consulting company which has become the premier consulting company for life sciences and more on the commercial side of the business, which is sales and marketing. And that was, and I spent a good 14 years learning kind of the ropes, so to speak. And then after that, I started kind of went on an entrepreneurial journey to build some technology companies and then came back to this in around 2016. What I found was the tech, the questions that the life sciences companies needed to answer were still answered by people heavy processes, people heavy consulting companies and, And when I was on the technology side of the business, I saw that most of the other industries were actually deploying technology heavily. to kind of answer the questions related to how to recruit customers, how to maintain those customers, how to keep selling more to customers. And the reason was very simple. The adoption of technology is very slow in life sciences. And there are entrenched players like my previous employer who basically have a vested interest in keeping that model going of billing their people. So we, me and my partner Robin, we basically said we need to disrupt this industry with going with more of a technology led approach where we use technology to answer questions. And the three core skills that we brought to the table was our understanding of the domain, which is deep, and our understanding of technology, which is very deep, and our understanding of Abhay Jajoo (03:02.018) the newer technology trends that we are seeing, which is automation of data management and all the analytics processes that go behind getting insights and also applying machine learning and AI wherever we could do that. And so that's kind of the reason we came into this industry. And so far we've had good tractions and I know you recognized us for that. So that's where we are right now. Russ Fordyce 🏆 (03:25.782) So walk me through a little bit. If I'm at a pharma company and I'm in a commercial team, I'm looking for doctors and patients. What's kind of the real impact of the software that you're offering kind of on those downstream costs? Abhay Jajoo (03:31.309) Yeah. Abhay Jajoo (03:40.601) Correct, so three big metrics, right? So we call it, we leverage technology to do things better, faster, and cost effectively. We don't use the word cheaper, cost effectively. And the idea here is, if you look at a traditional kind of sales and marketing support organization, which is what we call as commercial operations in life sciences, they would have... to do this analytics, to just even identify which patients to go after or which doctors who are basically associated with those patients to go after, to with their messages, with the right marketing programs or with their salespeople. They would have usually a system where they would have data ingestion itself, like just to collect all the data from different sources. They would have a variety of systems. And then there would be like, a second separate system just to integrate this data. run models on them to identify who are the new patients, what messages we should be sending, then to visualize their system, right? And think about this as four, five, six silos. And there are functions in sales and marketing operations which are sales oriented, which support the sales teams, marketing oriented, which supports the brand teams, and then some access, market access is a very special thing, which do the pricing and contracting with with the pairs. So all of them are sitting in different silos. They all have different needs. And they are all developing their own what we call as siloed systems, which are all split. And that's where the inefficiency is. A, the data gets moved around from place to place. And there are people who are doing this. They are repeating the work. Abhay Jajoo (05:23.154) And the result of that is A, things move slowly. Things are usually lost in translation as you go from each step. And there are errors that drop out. The second thing that happens is because you have these redundant systems, the costs are higher. The third thing is that time to insights is longer. Sometimes when you get a refresh of a data set from whoever is the inbound provider. the output, by the time you get to the output, it could take weeks for someone to get insight as to what happened. We basically came in with exactly an approach where we had an integrated platform, which is our offering called CI Parthenon, which integrates all these data flows. It automates the ingestion of data, it automates the transformation of data, automates the modeling on the data, it automates the visualization of the data all in one place. And so as a result of that, you're not moving data around. You're actually analyzing data. And we have demonstrated to our customers that. If usually, let's say, the update comes in at 3 AM on a Monday morning from a vendor, these guys used to take three to four weeks to get insights. We were able to actually distribute that insights at 7 AM. And this is an art of the industry. Now, what that led to is not only speed, but faster decision making, which has demonstrated that, and also reduced costs. We've demonstrated about 50%, 60 % reduction in costs. And you can see exponential increase in just time to insight. And those are resulted, we have seen in some pockets we have seen actually, where we have done controlled experiments of rollout of this system, about 5 to 6 % increase in revenue also. So we hit all those three buckets on these metrics, just by automating that. So that's going to be, and this has happened across multiple customers, not just one or two. Russ Fordyce 🏆 (07:17.952) Well, what you're doing, and what I've seen across the AI space is these specialized vertical-based niche AIs that are a lot harder than people think. On the surface, it sounds like you're just training it on some select data and then saying, ask a question and get an answer. But life sciences, I imagine, is very different, especially in the commercial analytics space. Walk us through why this is different than just that. Abhay Jajoo (07:32.845) Yeah. Abhay Jajoo (07:39.081) Yeah. Abhay Jajoo (07:43.502) Hmm. Yeah. Russ Fordyce 🏆 (07:46.006) traditional Gen.ai, know, LLM use case. Abhay Jajoo (07:49.547) That's a great question, Russ. And I think this is where, and I don't know if you look at the trend of what AI is happening. And we are talking about generic data infrastructure providers. And then we have horizontal agents providers. And what we have done is we have actually built an agentic AI platform, which is an evolution from our path, which is Athena, which I can talk about that later. specific to life sciences, okay, in each of those steps that I just described to you, okay, the data. So not look at the data, just ingesting the data, right? very specifically how it gets used, okay, is if you look at other industries, especially if you look at consumer products, there is a high volume of data, but the variety is not that much. I mean, we go to the three V's or four V's of big data, right? Where people focus on velocity, volume, veracity, and variety, right? What happens is in pharma, variety is different because the whole data distribution flow of products and the various, the number of Stakeholders that are involved in the process are totally different. The volume is not that high. There are like a few hundred thousand doctors at the end of the day. For that 100,000 records, there's nothing in a consumer product company, right? So the variety is high. So that's where the challenge is. Now what that means, the other thing is the data comes in through multiple sources, okay? And so you've got to make sure that they all connect properly, they get cleaned up, okay? So those are the challenges. Now the second step is then how you use the data. And it depends upon the use case at the end. Are you trying to address an operational problem? Are you trying to address a strategic problem? Because sometimes there will be holes in data, which you need to fill to make that decision. Example is, let's say you're trying to do a sales forecast. Let's say some data reporting messed up over that particular time period. Now you're trying to tell the customer, you lost sales. But actually, might not have lost sales because of some reporting. Abhay Jajoo (09:46.821) issues. And this is the insight that you have from having done this in the past, okay, that's having the domain. Now also you integrate the data and you talk about integrations, right. There are various joins or etc. If you get too technical, you can miss some of them, but we know some of them are usual, some of them are not unusual. So we need to have that domain knowledge. Same thing with business rules, applying business rules. How do you take these data, aggregate them up, what is supposed to be included, what's supposed to be not included? Again, that's domain knowledge. Okay, that's our pharma-specific knowledge. And that changes not only by... within pharmacology, within therapeutic classes, for example, an oncology product which has a specific distribution channel has a totally different approach doing that versus a mass market product like cholesterol reduction or diabetes. So these are kind of the nuances which become very important because ultimately the insights are going to be consumed by business users to make decisions and they all need to make sense and they have to have a certain pattern. They have to be explainable as kind of the people discuss. So that's our differentiation, is, and that's the specific domain, specific expertise that we bring. And we call it, call ourselves hyper vertical. So it's like not even vertical, vertical would be people who say healthcare is vertical. No, actually we have pharma. That's number one. Number two, we are commercial. We don't do manufacturing. We don't do R and D. Some, some, some overlap is there, but it's all about commercial CSM. So that's the hyper vertical agent applications that we are building, which are very specific to solving the business problems. Abhay Jajoo (11:22.478) And we understand the business users very, well. Russ Fordyce 🏆 (11:25.836) Yeah, and you have these two different channels, the CI Athena and CI Parthenon. First, I want to know what CI stands for, just because I'm a naming nerd. And then go ahead with that. What is the CI? Abhay Jajoo (11:30.519) Yeah, correct. So yeah, go ahead. Sorry, I didn't mean to interrupt. go ahead. So let me, I think there's a kind of method to this madness. So CI Parthenon, CI is the customer insights version of Parthenon, customer insights version of Athena. Now Parthenon is actually that is named by my co-founder's wife because she's an architect and she goes like, if you go back to history, the best architected building, even that current day architects look at in their train is actually Parthenon because it is a foundationally very strong architecture. So our architecture, Parthenon, is strong. It has the longevity and the strength of a Parthenon, which we believe will last long time. And then Athena is the goddess of wisdom that sits in the temple of Parthenon. And we are generating insights over customers using the infrastructure of Parthenon. which are relevant and providing wisdom to our customers. So that's kind of the motivation behind those two names. Russ Fordyce 🏆 (12:41.814) Well, this is marketing guy's dream job with the visualization we can do here. Abhay Jajoo (12:44.974) Yeah, exactly. So, but I'll, so I think I talked to you more about Parthenon, which was the foundational piece, okay? Athena is, if I can jump to that, okay? Is, yeah. Russ Fordyce 🏆 (12:56.406) Well, yeah, in the nomination, I think you guys mentioned that you signed a multi-year kind of large contract with a client. You don't have to name names or anything. But I'd really like to kind of understand what did the validation process look like with that big client and what did they have to see before it got signed? Abhay Jajoo (13:01.122) Yeah. Yeah. Abhay Jajoo (13:09.838) Cut it. So basically, first problem you have to solve is can you solve, because we sell to business audiences, can you solve a business problem? And they always come to us and say, listen, here's my problem. Can you even solve this? Are you able to address my business issues? It could be like, well, I'm about to have a discussion with a plan. I need to negotiate a contract and figure out what's my contract going to look like. Okay. Have you done in the past where you have helped companies negotiate their deals? Okay. That's the example. That's where you have to have a clear business understanding. So that's where they, so the trust and understanding that you're solve this problem, which is what, then the second thing, which is, and by the way, other companies can do this, which is our competitors, but then the question is like, okay, how is you, how is it that you're different? So we say, well, we already have it kind of a, technology led approach to doing this, which includes platform plus our agents, which will allow you to not only make do this efficiently, because we already have built this so we can deploy this fast for you. You're not taking a six month, nine month process to do requirements gathering, blah, blah, blah, and all kinds of things. are like literally our first deployments are literally three to four weeks. And that usually is because we have to figure out where the stuff is sitting in the organization as opposed to us kind of spending time figuring out what to build, right? So that's the second thing. And the third is, can you operationalize this? Because if you are asking a business question today, that question probably is going to come back again, okay? And you don't want to have a process where that second question, the second time that question gets asked, okay, you start all over again, right? Okay, because that's what consulting companies do, okay, because it allows them to build people. Abhay Jajoo (15:07.374) We already have everything set up, right? So the idea is what we call is like the word is like we operationalize through technology our approach to solving questions. If you're asking me a business question, my technology solution and my processes are equipped to answer that question or similar questions again and again and again, okay? Not every six months, not every month, okay? And that's where you get the efficiency, okay? And the effectiveness. So that's our differentiation. Russ Fordyce 🏆 (15:39.158) So on the Athena product, you signed this big client. What did they have to see? I imagine your sales teams were in there and they were responding to that voice need. What did they want to see before they kind of got that level of trust? Abhay Jajoo (15:47.544) Correct. Yeah. Good. Let me step back. So sorry, I think you asked me that question. didn't get to tell you how. So you trust ability to show that we can deploy technology and then ultimately kind of operationalize that. Now, the Athena part of our business is actually we have taken the entire Parthenon approach and converted that into agents. So the data ingestion is done through intelligent agents. The integration is done through intelligent agents. And then visualization is done through intelligent agents. Those agents, and get orchestrated through a workflow, okay? And through use cases, depending upon the questions that people are asking, that a workflow is determined through the use cases. So those are, our IP is built around those intelligent agents as well as workflows. Now, what you, our interface for that is, in Athena, is a conversational interface, right? And by the way, we can deploy this using any LLM on any customer stack because there are some kind of check boxes that you have to sign. and check off before you can do this in an enterprise version. So the way we sold it is very simple. Look, we can solve this problem. Here's our conversational interface. Start asking questions. So in a meeting, you would just not even do a demo. And there is no slideware, nothing. You just go with it. It's like, log into this. Here's an instance. Obviously, it doesn't have your data, but similar data. Start asking questions. Abhay Jajoo (17:20.43) And just like you go to chat GPD, it starts giving you narrative on those questions, supporting data shows up, reports can't show up. And they click on it, they ask follow up questions, et cetera, et It doesn't hallucinate. It gives the right answers, gives the same insights. And the demo sells itself at that point. And it's like, OK, aha, can you do this? Can you implement this fast? Sure, we can do that. So that's kind of the sales process. Now, obviously, Then there is a whole MSA and contracting process, which in big pharma is a long process. And there are all kinds of people who need to sign off starting from IT to law and legal. So that takes some time. Russ Fordyce 🏆 (17:56.375) Well, and I like the way you describe it as kind of this Athena goddess at the top and the Parthenon as the foundation. Your Parthenon product, the CI Parthenon has been in market for almost, I think, six years. So it's interesting because you've been collecting all that data on all these different deployments for that long. What surprised you in that data and what didn't you expect to see or vice versa? Abhay Jajoo (17:57.805) Mm-hmm. Mm. Mm. Yeah. Abhay Jajoo (18:18.67) I don't know if you got any surprises from a data perspective, right? I think that's your question because we kind of understand the industry. So Parthenon is not a data collection, just to correct you a little bit. Okay, it's actually, it takes third party data or customers own data and integrate it and converts them into these applications which are ultimately used by customers, okay? The learning, I think, from our perspective has been that as customers use these systems, right? Okay. And they get faster insights. Their demand for faster and faster analytics goes up. So what you're seeing is like, well, there is almost a... If it is possible for you to provide insights so fast and so in such an easily consumable manner, they tend to ask more more questions. And that's where they start seeing the power of this integrated non-server platform. And now with Athena just being able to deploy and ask questions. See, the benefit of something like Athena with the conversational interface is you're just asking questions. You don't even need to train them, right? In Athonon, we had to still tell them, OK, follow your workflow. To look for this data, you've got to be trained and go here, et cetera. Whereas in conversational interface, you just ask questions, you get answers, and then you can do deep diving, et cetera. So it's very executive audience friendly from that perspective. Russ Fordyce 🏆 (19:50.807) Well, now when you're inside of these customers, these customer years are possibly the most regulated rightfully so industries on earth. They're also probably the most skeptical of kind of building trust around these AI generated systems. I mean, we still are getting hallucinations. We're still getting problems with AI and at least the LLMs I interact. Abhay Jajoo (19:57.753) Cut it. Cut it. Cut it. Abhay Jajoo (20:12.11) Correct. Russ Fordyce 🏆 (20:14.934) How did you kind of work on that when your decision kind of sits in between this compliance and regulatory framework? Abhay Jajoo (20:21.686) Yeah, so I think that of. Three things which are very, which are table stakes when you implement something like this. One is secure, compliant. And what third is, and this is very specific to Pharma, is guardrails. Because you cannot, you, and I mean, there are, it cannot be an open system where you can just go and willy-nilly access any data and, upload any data and just start analyzing. So there are really clear rules. And by the way, the rules are evolving. They have these AI consoles and stuff like that, that organizations are building. And then that emerging problem which is becoming very, very recent, like as people are, as there is like almost a mandate to adopt. AI, the use of tokens has just gone crazy. What that means is like instead of, so companies are just saying, okay, we're going to cut our workforce and open this whole thing to people and then they're using it and there's just no way to control the tokens. So there is almost like a token management system, token alert system that needs to be deployed as a part of these things. And that's just on the, what I call as the compliance or infrastructure side of things. So what we have done is like inbuilt into our platform, okay, and by the way, hallucinations is very important also. So we basically deploy all our, we call it as our container. Athena is essentially from a technology perspective, a container which gets deployed at our customer in situ, on-prem, something we call it on-prem, but it is in their environment. Abhay Jajoo (21:55.083) It connects to their existing data sources. It connects to all their security, single sign-on, everything around it. So it's compliant with that. It leverages their existing LLMs. So we don't have to use because our IP is the agents, which is our container code, and Workflows, SOVs, which can be deployed anywhere. So by definition, it connects on that. And then we have what we call as a good administration module which basically allows you to check who's accessing what kind of token usage is there because ultimately they don't pay us for tokens. leverage whatever their ultimate provider is for their tokens. So we are monitoring the tokens so we can send alerts out okay and so that they don't get a surprise bill of some kind. So those are the four things okay amongst many which are the more important ones that we can adhere to okay and that's what separates a enterprise implementation usually from A, what I call as a bespoke implementation, know, somebody going in GitHub and getting and doing one of five coded implementation, right? So that's the big difference between how we implement versus how kind of trials happen. Yeah. Russ Fordyce 🏆 (23:07.35) Well, your model is really interesting to me because I think a lot of the AI companies that I'm talking to are moving towards this model of kind of having this agent system rather than kind of the single LLM. So what does that buy you that the alternative doesn't? Abhay Jajoo (23:14.307) Yeah. Abhay Jajoo (23:18.104) Correct. Got it. Abhay Jajoo (23:25.493) As I said, control, right? so think of it as like, well, the normal LLM to me is just a tool, okay, at the end of the day, right? I mean, it's just looking at, it is interpreting text, okay, and then basically executing certain workflows, okay, right? So. That is to a tool. My friend uses a very good analogy for saying that. You can go to Home Depot and buy a bunch of tools, but you can't build a house. So you have to have the proper architecture. You have to have a very good idea of how the room is going to look like. So to me, all these data enablement tools, agents. depending upon how you can surface the information. are LLMs, these are all tools. How you stitch it all together to build the right house, which looks nice, which has all the functionality, is where the kind of the, my partner calls it, is kind of the secret sauce of the chef that was trying to build the right dish, so to speak. And that's where I think the domain knowledge comes in place. Russ Fordyce 🏆 (24:36.852) Now, you've described the platform, you mentioned it a second ago, that it's kind of data agnostic, cloud agnostic, almost use case agnostic. That's a pretty hard thing to build. Especially most of the specialized AI that I talked to are really kind of narrowing down on their choices because they're building for performance. Walk us through kind of your architectural stance and kind of why you built it this way. Abhay Jajoo (24:44.397) Yeah. Yeah. Yeah. Abhay Jajoo (24:52.237) Hmm. Good. Good. Yeah, yeah. So we are not data agnostic. We are infrastructure agnostic. And so think of it as the stack as infrastructure, and then the data, and then agents, and then use cases. Right? OK. And agents are horizontal agents, okay, which could be like for example and just if I'm allowed to use these in like example infrastructure could be AWS, Google Cloud, etc. Just providing metal plus something on top of it. Abhay Jajoo (25:36.055) and then the database on top of it. And then you have these data provisioner agents like somebody would have data sitting in Databricks or Snowflake, for example, and they would have agents which will allow you to consume that data. How do you consume the data and make decisions, is where our agents come in. It's at the word declines. And also the LLMs are used usually to interpret the data and orchestrate that workflow. Again, so that's why to us, these are horizontal layers. What our agents do, our hyper vertical agents do is they are containerized. They use any of these because let's say there is a patient. adherence problem, for example, that is our use case. It's like, well, a question that usually pharma marketers will ask, like, what is the adherence of my patients when they get on drugs? Right? We need to go and get the data from let's say the claims information that comes through a third party provider. That is probably sitting in data blitz instance. And now to kind of ingest that data, we need to communicate with the database. Now, if somebody else has a snowflake, yes, that's doable. If they are sitting in fact files, yes, our agents will go and figure it out. So that's kind of the agnostic part of it. Then the workflow is what our proprietary is. Like, well, what is needed to understand the other? And so we need to look at patterns. We need to look at certain cutoffs on how the other is Abhay Jajoo (27:09.936) determine all those stuff, right? That's kind of the downstream work. That's all built intelligently into our container, so to speak, right? Now, to interpret that, orchestrate that, now the LLM is going to go and ask those questions and surface it when somebody asks a question. That LLM can be anything, okay? It could be a small language model, large language model. All it is doing is taking our conversation, text conversation, and converting that into a workflow, and then interpreting the workflow and just presenting that information, okay? So that's what we mean by agnostic, agnostic of your infrastructure, agnostic of all the other tools that you have. And our IP is sitting in those agents and in the workflow. Russ Fordyce 🏆 (27:54.123) Now, one question that I wanted to ask you, we've seen this kind of cycle, or at least I have, in technology where we go into these models, I think we saw it first with mainframe computers, right, where you had network access to mainframe computers, and that was the first cloud. And now we've moved into the second wave of cloud computing. And I think we're starting to see, I saw it on the communication system side of people actually bringing these systems into their enterprise more often. Abhay Jajoo (27:58.04) Mm-hmm. Abhay Jajoo (28:09.902) Mm-hmm. Russ Fordyce 🏆 (28:22.732) instead of relying on the cloud because these, because it's frankly, it's a lot easier to run your own cloud, right? Your own network. How is that being impacted kind of these pharma execs as they kind of get, you know, I'm sure their boards are, you know, are AI mandate, bring it in house, we got to do this. You know, here's a ton of money from the board. I'm sure these CIOs are, are flush with cash. How are you kind of convincing them to kind of, you know, stay in that model? Abhay Jajoo (28:25.614) Mm. Abhay Jajoo (28:43.895) You Abhay Jajoo (28:49.07) Yeah, so first things first, we cannot do anything in pharma on a public cloud. Even if it's a public cloud, the data has to sit in kind of an on-prem version of some on-prem version of it, because the data cannot leave. So that's kind of like almost the firewall from a physical firewall that happens. Near physical, let's just call it that. But the question you're asking about the CIO is audience is a very good one. I think, as you said, every CIO out there is essentially just flooded with requests from the board, implement AI. And guess what? The next question there is which use cases? And the CIOs, traditionally, nothing against CIOs. The CIO's main job is to keep the lights on. Keep that in mind. If something goes wrong, the S hits the fan, they get accountable for that. So their biases, and I've worked with lot of CIOs, their bias is like, hey, I'm going to implement this, but you better do this right. And they're of averse to innovation from that perspective. So you kind of work with them, right? So they are all sitting with all this money. They have very good knowledge of technology, but they don't have good knowledge of the use cases. So what they do is they go after the business and say, listen, let's help you build these systems. And then the whole cycle of understanding the requirements, et cetera, starts, which is a long process. So our pitch to CIOs is, and they are slowly in the last, and it's funny, in the last three, four months we've seen this shift where they were kind of like, we can build it. Now they have come to the realization. Sometimes buying makes sense. And what we call is like, almost like assisted buyout is what we can buying off this thing is what we can do. So what we do is we accelerate your buying, okay, the buy approach, okay, by bringing in something which is 80 % to 90 % ready to be deployed. Abhay Jajoo (30:43.934) So you ultimately own the other agents, you own the whole instance, you control how it's going to be deployed, all that is there. It's just that we are bringing in something which is you don't need to start, for example, modeling and start them from zero. We already 80, 90 % there and we can fine tune the last 10, 20 % for you. And that will be custom to you. And by the over 80, 90 % over time will get better and better, so you get the benefit of that. And now you're not going through this whole requirement-guiding process because the entire workflows as well as the use of this is already pre-built in our approach. So that's the story we are going to going with the CIOs. Limited success, I'll be honest with you, because our audience is primarily business and they love us, okay, because we can just kind of solve the problem. They are pushing the CIOs to make the decisions. So they are still getting used to it. For that, we need to get to a certain scale, which we are trying to get to, but that's kind of the... But I think eventually as the mindset of the CIOs changes that look, we need to embrace these hyper vertical providers who we can leverage to kind of accelerate our deployment cycle. Because ultimately the board is also asking for like, you're sending all this money, are you showing the results? And that's where immediate ROI comes from for them from just having something which is ready to deploy. Russ Fordyce 🏆 (32:04.438) Well, now you guys have taken on this big industry, this consulting industry, right? even broader than life sciences, these were the experts in every industry. They were financial experts or analytics experts, tech experts, whatever, that were kind of on higher. And AI is taking a big piece out of that, I think, right now, and it's going to continue. What do you see it looking like, kind of consulting as a whole or the life sciences consulting niche look like? Abhay Jajoo (32:09.038) Cut it, Cut it. Abhay Jajoo (32:16.184) Correct. Correct. Correct. Correct. Abhay Jajoo (32:31.424) Yeah, I think there is a, yeah, so I don't know if you saw in the last two, three days, there is big announcements by Anthropic and OpenAI to build their own consulting arms. So I think there is a realization on those guys also part of that. Russ Fordyce 🏆 (32:33.792) five years. Abhay Jajoo (32:47.478) Ultimately, if you're going implement AI, you will need people who understand the enterprise, the domain, and all the challenges that are associated with that. You can't just develop a model and just throw it at an industry, right? It has to. So I don't think consulting is going away. OK, let's just say that first, because experts are needed to implement this. And the more domain-specific people are there, they're going to be there, OK? The business model is going to change, which is where it is important. So the traditional consulting businesses, which luckily we did not have there, is a pyramid-based middle model. You have a bunch of partners and engagement managers and then look at a huge bottom of the pyramid of analysts who are just moving stuff around, putting decks together, all that stuff. I think that the pyramid model is going to break very fast, right, because that layer is all agents, right. Now, not all agents, but mostly agents, right. They're going to just, so that goes away. So now where do you leverage? Where do you create that leverage in the consulting business? It has to come to technology, which means if you are a partnership model in consulting, where partners are getting highly compensated, A, your revenues are going to have right away, right? now, but you could still be more profitable. So they have to change their model. So that's going to be a legacy companies. The bigger you are, they're going to have struggled with that more transition. And then there are software companies which have of butts on the seat model, so to speak. They are also going to struggle because agents are going to replace all their workflows which they have automated. So those are the two competitions which we are seeing. We kind of sit in the middle very well because our model is not to build by the by the seats, not built by the people, it is built by the outcomes. How much benefit are you getting? How much usage are you getting? And so that actually, and that's a recent transition we've done and our customers have embraced us. So that's where we are kind of pigeonholing ourselves. Russ Fordyce 🏆 (34:42.272) Well, these vertical AIs are going to be very successful in my mind too, because the way you've laid these out. If you could hand one principal or a founder that's looking to build one of these kind of very niche or vertical AIs, maybe in some other regulated industry, whether it be financial or healthcare or legal or whatever, what would you tell them? Abhay Jajoo (34:46.638) Good. Yeah. Abhay Jajoo (34:55.053) Yeah. Right. Yeah. think hyper vertical is the way to go, but you better have a known problem that you can solve. And not only solve. But you can, in enterprise, scale it. And obviously, you've to check all the boxes of just being compliant and all that stuff and not hallucinate. But then we have the ability to scale it. I mean, the narrowest of use cases have huge potential. But I think where people miss out, and that's kind of the smaller ones, is they just like to build something and then just deploy it. But you have to have the discipline. Because no enterprise. I mean, look, some ad hoc deployments can be done, but no enterprise is going to have an operational system which does not check the boxes of all the compliance, security, and guardrails, those issues. Russ Fordyce 🏆 (35:56.789) Yeah. Now for customer insights, AI, do you see yourselves going deeper the next five years or in a broader to other parts of the life science spectrum? Abhay Jajoo (36:03.342) No, think we are life sciences commercial analysis specialist. And as you said, as I told you, like you have to have vertical expertise for us to kind of get successful. We have deep vertical expertise here. We may go in commercial. There are certain adjacent use cases in real world evidence as well as recruitment of patients in clinical trials where the same data is used. We can go there. We have no. ambition to go to other industries, no ambition to go beyond that because I think this is a big enough industry for us. And there is a of opportunity to make the industry go and bring efficiency and effectiveness to this industry, to technology and AI. Russ Fordyce 🏆 (36:49.384) Any direction for your roadmap? Are you trying to focus in on one area or another? Abhay Jajoo (36:53.814) Yeah, so the use cases where we are focused, right? And what we are seeing in the industry, traditionally sales marketing used to revenue for farmers. Okay, I mean traditionally meaning I'm talking I'm dating ourselves in the 90s in the 2000s right but now it's become a very pair driven market Okay, which means not only how many reps you send in front of the doctor? How many messages emails are bringing them to speaker programs all that stuff that used to in the past drive? Prescriptions for pharma that's not happening anywhere. That's that still happens, but it doesn't take a bigger weight. The bigger weight is now do you have patients? Do you have access to those patients through negotiated contracts with payers like the pharmacy benefit managers, et cetera. And that piece is coming up with the right contracting strategy and understanding what. what are the trade-offs of those contracts, and then the downstream effects of that on sales and marketing, okay, is a kind of an underserved area, okay? And the traditional consulting companies, because of their legacy, are stuck in the sales and marketing side of things rather than the access. Pricing has become a very, very important part of the performance of some drugs, some therapy areas, and that's where we are gonna double, kind of, double down, so to speak, right? Because I think it's an out market, customers are hungry for answers on how to kind of not only integrate this data, it's much, more complex than your traditional data sources, you're the way getting it to insights. And we are seeing some significant success in that area. Russ Fordyce 🏆 (38:38.828) Well, this is fantastic. Abhay, Customer Insights AI, super interesting. As you guys get into this deeper, it only helps out the patients at the end. So congratulations on all the success and the award again. Abhay Jajoo (38:47.459) Perfect. Abhay Jajoo (38:51.257) Thank you so much, Russ. Appreciate you taking the time and spending and getting to know more about us.