Parijat Banerjee, Global Head of FS at Latent View Analytics: Evolution and Impact of APIs and AI in BFSI Sectors

Speaker A: This is the scalepoint Podcast, hosted by me, Martin Koderish. On this show, I sit down with founders and experts in the payments and fintech industry to explore key trends, opportunities and the realities of building a sustainable and profitable business in this next phase of the industry's growth, defined by tighter budgets and more challenging access to external funding. Right, let's get started with this week's episode. Okay, great. So we are live with another episode of the scalepoint podcast and today I'm delighted to have my guest with me, Parijat Banerjee, who is global head of Financial Services at, uh, Latent View Analytics. So welcome Parajit, how are you doing today?

Speaker B: Thank you, Martin, doing great. Appreciate you having me in this podcast. Quick background, as you mentioned, I lead financial services globally for Latent View analytics for your viewers. Latent View analytics is a data, uh, analytics consulting organization that's highly data first and data native. Been in the space for more than two decades and we collaborate with across the left shift and right shift spectrum of AI, machine learning, data science, all the way up to data engineering and platforms across Fortune organizations and love to be here today.

Speaker A: Great, that's fantastic. Just to frame the question, really what we're looking to discuss today really is around APIs. So, um, why don't you maybe just give me your view on how the role of APIs have evolved over the years.

Speaker B: Sure. This is a great starting point. Modern, I think APIs have been or uh, have transitioned from being these solely low level technical building blocks to becoming this integral front end of commercial products in their own right. This evolution is rooted in several factors, right? It's the rise of the API economy, the change in the business strategy, the broader digital transformation, and specifically for the BFSI industries, uh, it's transforming how these sectors operate, how they innovate, how they interact with customers. And I'll take one or two examples to bring this to life. One of the areas where there's a lot of focus is open banking and regulatory compliance. Now if you look at open banking initiatives, and you know this even better, the whole open banking regulations like the PSD2 in Europe, right, that requires banks to provide secure API access to customer data. Now this has transparency, it drives competency, enables third party developers to create innovate financial services and that's great, uh, there's also this whole compliance section that allows compliance measures to meet regulatory standards, right? And you look at access controls, you look at monitoring, you look at auditing capabilities, all of these are sensitive financial data. Now what's the implication the implication is there's better customer control so consumers have more control over their financial data. That enables them to use fintech apps and integrate accounts across multiple banks. Most of us or the Gen Z population is probably much into that. The second great thing that comes out of that is market competition. So banks can now partner with fintech companies and offer these digital experiences that themselves they might not have been able to. Or it also helps them to maintain regulatory compliance. There are many others, but I think the other one I want to focus is customer experience. Right. And I think that has tremendously led this front end digital transformation with APIs. Right. Which leads with three key areas in my mind. One, it helps with seamless integration. So you're looking at banking services into your mobile apps, into wearables, into online platforms, providing customers with intuitive accessible financial tools. That is amazing. Right. So this is what API brings in for. And then there are two other angles. One is the personalization and the innovation that APIs help with everybody in that broad spectrum of customer behavior and data analytics and the omnichannel experience. This is amazing where you have unified customer experience across multiple touch points. And this wouldn't have been possible with this, the whole API thing coming to life. So in my view, API is not just a tool for integration. Right. It's becoming a strategic asset in the BFSI sector. So they are driving regulatory compliance, they are enhancing customer experiences, they're opening new revenue streams. Right. Other than operational efficiency and there are multiple competitive advantages. I think it's a rapidly evolving space and we just need to keep our eye out in this segment.

Speaker A: Yeah, absolutely agree. And um, I think the API side of things has developed to the extent that previously banks might have not, you know, had like those customer facing APIs. Right. So they've had the internal plumbing and that's the view. And now obviously APIs are coming forward as commercial products like you said. And I suppose there's one hand just having the being compliant. We're going back to your point about open banking and APIs being being developed from, by banks for compliance purposes and being presented just basically ticks box exercise, which might be just a uh, raw API, but there's still quite a journey for them to make from uh, a compliant API to, to something that's to be commercially viable and the products in the same. Right, yeah.

Speaker B: And that, that, that's a wonderful point, Martin. I think being a commercial viable product is still some way to go and it's also dependent on the geography. Right. For that matter, for example Europe or APAC has a leg ahead in the process in I live in the us, I live in Virginia and in US for that matter. It's a lot about market competition and market requirement on how the APIs are getting leveraged. Right. Which is slightly different when you look at uh, objectives like PSD2 in Europe or if you look at in APAC, how they are working across the board. So I think your point about how can they be leveraged as commercial products that still has some way to go on based on party regulations, how it comes out and the convenience of the respective geography to start leveraging that data. Right. Or that particular piece to run with it. For example, the UPI system that runs in India is amazing on how they have leveraged in many ways, uh, the usage of open banking or payments in that segment to make it much more viable for citizens across. Right. So it's going to take its different pieces on how it runs across the board. In my view the geography is playing it differently, but it's a matter of time when you would see new revenue streams coming out of this area. So as you look at digital transformation, the role of API in banking and financial services and insurance will continue to expand. It's just a matter of time how the respective geography and regulation plays this out.

Speaker A: Another way of looking at it I find is that just to view this whole API and as part of an unbundling almost of the sector of certainty of payments, we've seen an unbundling continue to see it also in banking and then it's been rebundled into other areas either into a, into a platform so that you have those APIs deeply integrated into a customer experience rather than being something that you actually have to search out and navigate your way to and consumers. A siloed banking app, for example, is actually embedded into your experience as a consumer. And it's also perhaps also rebundled in perhaps apps like Revolut have done an extremely good job in rebundling those various different services to create a super app type environment. So this unbundling and then rebundling is I think quite common. And yeah, and certainly in Europe it started with open banking, we're now shifting already into open finance. So it's extending beyond access to payment accounts into other forms of accounts, into insurance. We come onto that later on. And then also I think even beyond that it's not too far off that we'll see more broadly speaking other data based ecosystems being brought into that open environment that uh, might be a telco or utility type data Environment. I mean we have in the UK something called the Smart Data Initiative which is a follow on to open banking in the uk. So it's already there. It's uh, pretty advanced in other markets like you mentioned, like Brazil. There's a greater focus on it in the Gulf region. Dubai, Bahrain, very front leading I think Saudi also. And then in the US it's a different situation, isn't it? Completely. It's not driven by regulation, it's entirely market led. But I think it's also finding its own um, it's like bilateral agreements between plaid for example and a bank, et cetera. But that's also proven to be viable. It doesn't need to be regulatory driven it seems to me. What's your view?

Speaker B: And ah, you're spot on. Right. I think it's looking at it from two different lens. Right. From a lens of regulation on how you're going to get this across us you're very right. Is market led. So they try to look at how this plays in the market competition space. So banks can partner with fintech companies. They enhance digital experiences based on how it's uh, how the market is really looking at that and that's a different angle to look at it. It's also playing across different areas. So likes of stripe and plaid as you mentioned and others. Right. They are working with some of the larger giants of the Wells Fargos and JP Morgans of the world. How do you. I loved how you put it of bundling and rebundling. Right. So that's constantly happening at that space to see what the market really needs and what the customer really needs. I think the bottom line is the customer needs to have improved engagement. Uh, the customer enjoys faster, more reliable services with consistent user experiences across platforms. Right. Platforms needs to be seamless and that's where I think, at least in us that the focus is and I hope we would see a lot of these changes over the next few years.

Speaker A: I totally think it's feasible in the U.S. right. What's missing? There is just a level of standardization perhaps and some regulations around access, et cetera. But. And it may lead to some level of fragmentation. Some banks will open up and uh, provide uh, APIs more than others. And so it's not ideal in that respect. So you might be better off with some foundation, layer of regulation and then build on top of that. But I think in all circumstances you will eventually need to get into a market led environment where you might have regulations building some level of foundation. But then the market itself, building the house as it were. So, so, so the outcome, the direction of travel is similar and almost the same but it might be slightly different route in my view. But you mentioned some of the big banks, right. Partnering with the fintechs and that's always very interesting and I think m in our prep call you mentioned some of the challenges, some of the larger bank when they start digging into their tech stack where you might obviously you have this outer layer of APIs which is customer facing, enabling all the embedded experiences that you mentioned. But to really create value you probably need to go quite deep into the tech stack of a bank and to obtain that data. And that's an important source of any service that is then provided by the API. But that's an immense challenge for some banks with legacy or outdated technology.

Speaker B: Yeah and you are spot on with that. I think when we look at the uh, broader scope there are banks where if you go through a few layers and specifically legacy banks. Right. That's where they, they have the issue. If you go down a few levels you are probably in archaic technology. So it needs that upliftment over time. Right. I think what happened in, if you look at, you mentioned Saudi, you mentioned the whole reason of apac there was this whole skip from leveraging payments of the credit card. Right. And they directly went into some of the uh, company countries directly went into leveraging systems like upi. Right. So the movement was faster. Right. Us was heavy in credit cards and usage of cards across the board. So the technology across the board is still archaic in so many senses. So there is a lot of investment that needs for that to go up and that's where you have the likes of Stripe and plaid and others that are collaborating. Eventually there was a statement and not to quote it but I think read somewhere where Jamie Dimon had made the comment that JP uh Morgan moves around $10 trillion every day so we are the ones who has to change the payment system for us to some extent. Right. Because that's where the movement is happening and I think the fintechs are playing a part to come together and run with that. If you see the entire BNPL story that played with the likes of a firm and others also was leveraging some of this segment to, to run with. And so you're right, the market led part also equally comes with the technology that is there and the underground technology that needs to change, which is changing right now by the way.

Speaker A: Right. Should we move on to the next topic of our discussion around a use case that you wanted to bring to light around um, healthcare insurance, I believe.

Speaker B: Sure. So let me do this. The healthcare. I'll take a step back and just lay the ground here. The healthcare insurance ecosystem is conceptualized as a triangle, right? And that consists of three main stakeholders. So you have the payer, which is the insurance companies, you have the provider, which is the hospitals, clinics and the other care facilities. And then you have the patient. That's you and me and everybody else, right? Now, as they say it in the us it's both a, uh, vicious and a virtuous triangle. Each point in the triangle generates immense amount of data, right, Other than the services that are there. And this is highly fragmented. So when you look at the payer. Let's start with the payer. If you look at the insurance companies, right, Just to set stage, the payers generate and manage claims data, uh, risk profile, member health histories, financial transactions, right? So they rely on this data to assess risk, to set premiums, to manage cost controls, all of this. And this data, the challenge is it's completely fragmented data sets, right? They're leveraging APIs in many ways, and I'll come to that. But along with the fragmented system, there is delayed insights. Most of the insight is post hoc, right? So it's not helping the broader ecosystem. Now let's look at the provider. The provider is the healthcare facilities and these are the ones who are maintaining the detailed clinical data. You have these electronic health records or EHRs, the diagnostic reports, the treatment plans, patient outcomes, all of this is here. Now what's the challenge here? The challenge is interoperability. EHR systems and provider data platforms rarely talk to payer systems. And this is leading to delays in data exchange, right? This is creating data silos. And the third is if you look at the patient now, the patient is the center of all this, though I call it as a triangle, the patient is still the center of all this. And they have their personal health records, the wearable devices, the patient reported outcomes. All of this is in different segments. So the incomplete data sharing, that's the biggest challenge. Patients may have their health information distributed across these multiple providers and insurers. And there's no single entity that's tying this up, right? So that's the stage just to set stage. And the current data flow challenges are huge. One is the whole. And I look at, when I look at this, right, Considering I've spent years with a company that's process focused like ge, I look at every problem through two lenses, right? You look at the process door and you look at the data door and if you look at the data door, the data flow between stakeholders here, claims processing, health records, that's all fragmented. Right. And if you look at the process door, there are ah, multiple manual processes that are running. They're reconciling claims with clinical notes and stuff and all of that. So both the doors are still has a lot of inefficiencies that run. Um, now how can API or in a broader sense AI. Right. APIs are still being used across multiple areas of claims and others. But how does AI add value? And in my sense, there are four big areas where AI adds value. First, the integration of data and standardization. That is amazingly important in this system. Right. Second and third is tied second is creating predictive analytics. How do I create AI can forecast risk, predict high cost claims, or flag potential fraud for that matter. Right. And also look at real time insights. Real time insights eventually will become a uh, norm out here that'll help payers and providers make timely decisions for care management and cost control. But all of this modern comes down to one key area and that is personalization of care or value of care. And AI models can analyze individual patient data, report to give treatment plans, to suggest preventive measures. Right. And this is where the whole story comes as the outcome for it and to, to bring it live. Right. We had, I was very closely working with one of the large nonprofit health plan providers in California. And over the last few years, right. We have been working with them to come to the cutting edge of leveraging data, uh, or benefit of health care and health care insurance. And when we started the program, the issue was on three segments. The first point was when we started the issue was data was not even at one particular segment. So they approached us with a problem that said that's true for most Fortune 500s, a problem of one of too many dashboards. Right. And I call it death by dashboards. So if you look at sales, finance, marketing, every team had their own dashboard. They were trying to convey their own insights. Different teams, different methodologies, calculate the same metrics. KPIs were different. So finally insights that were being circulated were incorrect. So we said, okay, fine, this is not gonna work, right? You need a single source of truth for you to run. And that's where we came in and helped out build cortex. And cortex is a single source of truth for all data that is there. And we moved from consciously from the use of dashboards to decision boards, which helps you define consistency, accuracy and simplicity of insights. So, and just to give numbers, There were around 100 plus dashboards but we have the unified template created the hub and spoke model. And the icing on the cake was Cortex won the prestigious 2023 Mission and Value in action award. So that is great advance.

Speaker A: Awesome. So I suppose. So, uh, in summary then what we're saying is that ah, this healthcare triangle that you described, right, it's mentally fragmented. There is value in the data from a customer, patient's point of view, sharing that data. To have a better experience, you need to do certain level of transformation to make that data become available, to move around. APIs are enablers of that communication, I would say. And then really, that's when you can then layer on AI, right? To really accelerate and expand and really leverage what is possible.

Speaker B: Yeah, and you're right. So I think the first part of AI, the heart of AI is always the data, right? And we first helped them do, and we did together, is to have a consolidation of that data. So now when the data is unified, the insight starts getting more accurate and this opens up avenues for improvement and growth. Right. So typically this dovetails. Your question perfectly dovetails into the next part. Because when we unified the data then they said, okay, fine, let's start solving the problems we have. And one of their major problem was digital adoption. So they had a rapid 2.1 million households, right. Of which around 60%. So I think 1.2, 1.3 million were registered. Think how many were using it. Just 300k, 300,000 people of 1.3 million was using it. And what was happening is they were constantly going to the contact centers, the volumes of calls were high, there was huge amount of expense that was going on there. And they said, no, we want to get into more digital front end MEM adoption that we can run. So what we did was we jumped in to say, let's look at your key metrics now that you have the data in one place, right? And it's, you can rely on it. It's a single source of truth. Let's look at registrations, claims, payments, appeals, digital logins, all of it together in one segment to identify the right KPIs you want to look for it. Then we started running digital surveys. We did topic modeling. So if you look at topics, what is the problem that a member is having? Is it ease of use? Is it bill payment and tracking? Is it registration and logging? Where is the issue? Right? And with this information that we, that the data started speaking now and which was true insights, you know, the company got armed, um, to go back and start making mends on the campaigns they were running or the updates they needed to do on the website or the digital fronts and areas. And the impact was amazing. Right? The impact is in a span of a year they increased their registration by literally 20%. That is huge compared to where they were trying to do that over the last many years. Right. And the cost came down because the number of calls that were, and we had this is data for only 2, 3 months of work that had happened. I think the calls were around 200, 220k. That came down to less than 200k. So they were already seeing movement in that process as impact. Now that data was there and they were looking at impact. To your question, this is when AI can be leveraged, right? And what they started was, and this is, we are very excited to work with, we call it data for greater good also is to create this member DNA. So think of this, they are creating this, they want to create this entire data mart which has if as an example, if Martin is a patient, you have 600 plus attributes of demographics, medical history, claims, clinical conditions, hospital visits, you name it, that's all tied up with that one key customer. And once you start creating that and putting it together and eventually it's not 600, I think it went to thousand plus characteristics across each one. Now if I want to make this, paint this picture for you, think of the marketing or sales leader who wants to run a campaign on a demographic of people in California. Right. And who wants to check who are pre diabetic in a history or has cardiac issues. Now if the head of sales needs to do that or used to do that before just to collate the data, they would take around anywhere between two to four weeks and then take an action on that. Right. So questions like how many members are ah, prescribed diabetes medication? How many members in this county is diagnosed with hypertension. Right. Or how many members have total medication expenses over $100,000. Now this had come down, so this is where AI had started to play. This has come down from four weeks to 40 seconds. And that's the, that that's where you can look at. There are complex business logics, there is quality of data, there's recency of data. All of that is looked into. But what's important is these advancements open the door to a future where data, it transforms how we think about the patient provider and payer ecosystem.

Speaker A: There's an acceleration, a function of generative AI. Specifically, did you use an LLM?

Speaker B: You're right. So we, we initially it was machine learning models that Ran. Right. But on top of that we started using. We started changing and leveraging Genai and using something called slm, which is a uh, small language model which is specified to. Because this is all. Again, this data is very guarded, so you can't let it out. Right. So it is a small language model that ran, but that really helped us look at topic modeling that helped us look at different methods and the accuracy went up. It was around. I think when we started it was around 70 or 80% on machine learning models. With leveraging of SLM and Genai, it went to more than 93, 94%.

Speaker A: Let's move on to the closing section here. What are the. In terms of practical advice you would give to organizations who are thinking about attempting a transformation and using AI to do that?

Speaker B: Yeah, and that's an amazingly rich question and deep question, Martin, because it has quite a few angles. So I always feel right in this age of AI, when answers are cheap and abundant, good questions because the new scarcity. And recently uh, in my, in one of the articles in Phobia was writing called Personalization and Automation and Trust in Data, uh, I tried to explain part of this. So with AD AI transformation and you've heard me say a few times with you in this discussion, data is the heart of the whole operation. And this would be followed by model development and then you would have organizational transformation and other things that would follow. But in my quest over the last many months and discussions with industry stalwarts and seekers like you and others, I've realized that there are a few key areas that people are focused on. Right. And they should ask these questions. Questions being how do we redesign or design robust enterprise architectures that can handle these macroeconomic shifts? Right. That's happening because that plays a big part in it. How do we democratize intelligence? It's not democratizing data, it's democratizing intelligence in a way that alters the learning curve of how humans look at it today and forward. And how does that deliver material outcomes with a defined intentionality of how we do it. So I think what we. When I talk with business leaders, my suggestion is everybody. And that's the norm in the part of the hype cycle also is you're going all in AI and agent AI. You must champion the creation of a robust architected data repository, ensuring that the lines of business. Right. They're different lines of business. They are defining governance organizations. Almost like a two in a box model. Right. Which what that means is this requires much more than pure play, technical implementation it demands a uh, cultural shift towards data literacy and a crystal understanding of the domain. That's very important because data without domain will not make sense from an uh, insight and action and outcome perspective. So it starts with embedding this whole AI in the DNA of the organization or data thinking I use. And we in our uh, consulting framework use something called raise R A I S E raise that helps organizations plan from the first phase, which is retrieval of the data, to the final execution of the data. And it has various aspects of it. It has the first parts of retrieval, it has the aspects of generative AI, it has the aspects of when you're getting into agents, then you have the, where the agents are interacting amongst themselves. Right. As a symbiotic situation, we've not come there yet. Right. We have some way to go. And finally the implementation of it. Right. With the human in the loop as one cohesive structure. So with that said, I think over the last few years you must have seen, and I've seen this too, there's this corporate craze of uh, having the Chief Data Officer and organizations have defined this role and position. But I think it has to be much more purposeful. And that's my discussion with almost every CXO that I have. The CTO shouldn't be perceived as a support function lying to the cto. In my view, if the head of commercial can command attention because they bring in every revenue, the CDO must have a similar influence across the table.

Speaker A: Yeah, and I think you mentioned earlier on at the beginning about this left shift and right shift decision point. What do you, what's your view on which direction to go first?

Speaker B: This is a controversial question and everybody answers it with a split. But I'll give you my view. I personally think the right shift is where you're defining your plumbing and you're defining your engineering and you're defining your base as a structure. So over the next few years you would see and already banks and financial services, the larger ones are taking time because they have a lot more archaic technology to move with vis a vis. The smaller ones are the fintechs, as you mentioned, Revolut was a great example, or the likes of plate and stripe, they're moving faster because they don't have that. But that shift of merely implementing technology in a sense of doing the plumbing and the engineering is extremely important. And then you look at, in my mind the shift of moving to AI because you need to set your data right. And then you get into data science and then you get into machine learning and then you get into the eventual leverage of AI and agentic AI in the process. Right. In my view, again, the only way to stay ahead of the curve is to set your data foundations today for anybody to have an agent tick tomorrow.

Speaker A: And how's that controversial? Give me the alternative view.

Speaker B: Yeah, the alternative view is playing in the hype cycle where people would want to jump into and say I want to go ahead and build AI models and I want to make sure I can have the agents already performing across my various sections and the human would go ahead and have some part in the loop. What will happen as part of that is if the data is not cohesive and in one segment the hallucinations will go up. Because the advantage and disadvantage of AI is this one. It does things amazingly well when it does predict. Right. But it equally predicts something that is flawed and you would not know the difference of it only because you have no control on the governance and the definition of the data.

Speaker A: Okay, so that uh, m totally makes sense. Right. So, but I suppose people get caught in the hype cycle, right. And just trying to create value quickly and do something that's, that's demonstrable to the market and their customers, et cetera, so they get pulled in that direction. What's an appropriate timeframe for companies to plan with regards to AI strategy?

Speaker B: Normally it depends on where in the lifecycle of data they are in. Right. In some organizations where the base and structure of data is well aligned by that the data engineering is well done. I think the cycle is much faster because it becomes much easier for you to build your models. On top of that, start relying on your models, start training your models with the data that you have and eventually spit out outcomes or even uh, the leverage Agent Ki to do parts and functions of the workflow as you would. Right. But in many organizations I think the base structure of data is still to take time. So I think it's anywhere between a uh, three to five year timeline that would be put in place. And it's also important on the lines of business where you want to focus on. Right. If you want to focus on say a risk management versus a front end, there's difference if you're doing the front end customer experience management part of the story. And if there is some flaw as part of the. I, uh, wouldn't say flaw, but if that inaccuracy is there to some extent and it's not a 99% accurate outcome, it doesn't impact that strongly visibly when you're doing a risk Management or an operational outcome where if you have any inaccuracy it really harms the outcome or from a uh, regulatory standpoint. So I think it depends on which department you're going to run and it's going to have its own cycle. But there is an entire plan of setting your data, uh, making sure it's retrievable, actionable, and then you start working your analytics on top of that and build your AI on top of that.

Speaker A: Does that mean that you, you recommend that companies build or upgrade their existing technology to enable that access to the data, or are there different approaches they can potentially take? Perhaps more of a greenfield site where actually building something standalone and then transferring at uh, a later date.

Speaker B: It's a mix of both, right? Some companies have already invested in huge data platforms, right? So it's more about making sure that how do you connect the rest of them into this broader system? Because AI works best when it gets a broad sense of data, right? If you can give many parameters, the uh, outcome becomes much better. Now that equally means to your comments, some of them have to upgrade the technology, some of them have to literally move out and throw out the technology they have and get something completely, completely new as greenfield. Or uh, you can have incremental bits and pieces that keep working on that as a distributed system and you keep joining that as hub and spoke model to run. I think a lot of it comes back to intention and it comes back to availability of spend. Right? It's a mix of both. It wouldn't be fair to say that there's one size fits all. It does not. But it would be fair to say that you have to make sure you're looking at this and there is some spend here to get this right for a better tomorrow.

Speaker A: And uh, it seems to me that smaller companies have an advantage over more mature legacy companies, as we discussed earlier on those challenges that we mentioned. Although on the flip side, perhaps they have less data and less transaction to work with. What do you think is a sweet spot in terms of size of company can make this transformation particularly successful?

Speaker B: Let me answer that from a slightly different angle. I think everybody has to find a way to make this successful because if you don't do it, you're just left behind, right? There is no the road to. It's like when you come to a junction and you have a fork and you have to select which road to take. I don't think, um, there's a way of not taking this role, right? Everybody has to take the road or you're Going to just be left behind and you would not be in the competition anymore. The question is how fast you can do this. So it's almost like changing the wheels of the bus while the bus is moving. Right? So you have to do it in tandem. You cannot stop any of your work, but you have to invest in tandem to figure out how do you fix your data and start running your models on top of them faster and slowly transition to that. So if I have to take a guess, obviously the small size companies, even if they can do it the fastest, their impact to market will not be that high. Right? Only because they do not have that space in the market. The largest companies will not be that fast to make it only because they have much more legacy systems. But they have to figure out a way of working this structure in partnership because you don't need to build it. You can equally partner and get that outcome and equally structure that format. The mid companies are where they are best scheduled to get into a faster structure in the process. But then again as I said, there's no easy way to come out of that. If you ask me what is the optimal outcome? There are many theories but maybe it is to be able to eliminate these. The base level decisions that's happening right now. You target elevated reasoning, you nurture competing ideas. But most importantly I think people or uh, companies need to pursue or create a culture of dedicated curiosity that's hundred percent needed for success.

Speaker A: And my final question, although that was a great wrap up actually but it's back to the full circle to the API economy driving this forward. And you mentioned that uh, for a lot of companies an opportunity to really build through partnerships, is that a viable path for most companies the question of

Speaker B: build, buy or partner I would say again my view, build and buy will keep on happening in segments as the market demands. Right. And it will continue. But I think the future is part. There is no other way to survive in just building your own product or even buying product for that matter that you keep on adding to your ecosystem. I don't think there's going to be an ecosystem going forward which is completely self owned by one entity. Right. It's going to be a partnership of this hub and spoke model across the board that keeps running.

Speaker A: AI is inherently a partnership driven experience. Right. You're building on top of a uh, LLM and then creating a wraparound it. So it is.

Speaker B: You're spot on. You're spot on. It exactly is. And I was going to mention right, I, I'm reminded of this Austrian scientist called Hans Moravec, right. In the late 80s and he made this amazing comment. He said that humans were adept at very, or AI. Typically computers were adept at very complex things that humans did the chess, the math, but they struggled to match the perception of motor skills of a, uh, infant born with billions of years of human evolution. Right. And so this is called the Moravec's paradox and it's extensively used in robotics industries, organizations, how they can gain competitive edge. But now if you realize this whole concept of scientists in a cribinology is becoming true in the partnership space, right? So there is no simple ecosystem, just like the humans evolved as part of the ecosystem. There's no central segment out here. You have to realize analytics as a product and it has to work in partnership with critical future growth organizations to run. And I want to end with this line, Martin, that I really liked from Scott Fitzgerald. It says true intelligence is an ability to hold competing ideas in your mind and still work effectively. I think that's the future.

Speaker A: Perfect. Thank you for that, uh, closing remark. Very much agree. And thank you so much for your time today. And, um, let's have another call and chat in due course and find out the next wave of evolution in the data analytics and AI space. But, uh, till then, thank you very much.

Speaker B: Thank you, Martin. Appreciate it. Take care.

Speaker A: Thanks for tuning in to this episode of the Scalepoint podcast. Don't forget to subscribe, leave a review and follow us on social media. You can also find out more at our website, which is scalepointpartners.com that's it for today. See you next time and take care of.