Ravi Krishna

So right now I'm working on my startup, and we're working on a product in the space of modeling business functions, which is really not just one workflow or one domain, but model a lot of workflows together and then enable the business to publish the same and continue to use the same. So that's what I've just started to work on.

So I've been in the technology space. Been an engineer to start with, and then did my management studies as well. In the long term, in the technology space with IBM, American Express, and then I have been associated with all the big names out there: Bank of America, Merrill Lynch, Commonwealth Bank of Australia, across all the geographies, in Germany and so on. So early in my career I realized that you can just keep changing and learning new languages all the time, and where do you really see a pattern? And then I got into architecture. Got into integration architecture. So it was, you know, the first attempt at making the business speak technology or the technology speak business. Now, from there, enterprise architecture was a natural progression, and so I've been in the enterprise architecture space for the last 15 years or so, or 20, and then said, okay, now, patterns of the problems are the same. It appears to me that every three to five years we apply new technology but solve the same problem. Okay, and what is that problem? Let's say if you look at when the internet first hit, then we said, okay, let's have all the business functions accessible through a common user interface. And then we said browsers are the problem, because there are different screens, different layouts and so on and so forth. The core is the same. Let's say you are buying something. Your transaction was to basically pay the merchant. And then you just wanted to see how it is, and then you started interacting with it through the browser. And then mobile came and said, hey, I can do everything over the mobile. What are you doing? You are still paying the same. You just changed the form factor, but your ability to pay, or your capability to pay, is still the same. So the core has not changed. Now we are saying, hey, we got AI. What do you do? Instead of you going and swiping, you're basically delegating. And now the complaints with AI are going to be, hey, I'm trying to talk to it and then it is throwing a whole lot of irrelevant things at me. But I just want to buy one thing. I just want to tell it, you go buy. Right? So the experience of buying is basically changing, but the problem of buying has not really gone away.

To me, the way I look at these things is from a technological lens. If you look at it, it's just another new technology that can enable a few things. Every time a new technology comes, it comes with its capabilities and limitations. And then the biggest limitation, or constraint rather, not limitation, because the constraint for any new technology that has come in is that it has to integrate with everything else that was already in place. So AI is a new technology. But then what is it fundamentally bringing to the table? For the first time, what AI is bringing to the table is the ability for the machine to have the promise of speaking human. Is it speaking human? Can it understand? Can it have a conversation like we are having, and not steer away from the context and tangentially push off? These are all still things yet to be seen, because in certain businesses especially, you cannot have this non-deterministic approach to it.

And that is not just for the AI. That will also mean that we will have to recalibrate ourselves as to how we deal with this, because it's one thing talking to a person here, and then if you have a machine sitting here and you're just staring into it and talking. And so the human brain has to recalibrate its ability to interact. And that has fundamentally brought a capability to the technology platform. Now, how are we going to leverage this? Where are we going to leverage this? What additional risks is it going to bring? That is a whole different thing, and the industry is still not ready for it.

So what we are trying to do is be able to replicate, from between zero and a hundred, as close as we can come to a hundred. I mean, if we start off with 10 percent, 20 percent of a particular role in the organization, if we are able to perform those functions and have your delegate. So, you see, in this case the delegate is still technology. AI just happens to be another piece of technology, is how I look at it. It's just that you would talk to your technical delegate now just as you would talk to your team member, which you could not earlier. If you had to do reports and so on and so forth, people had to write queries, wait for six months, get all the data from hundreds of sources, churn it and then prepare it for reports and so on. Now you don't have to do that. You can just say, hey, can you do this? I'll make an attempt, bring it back. And so in this platform that we are building, the basic technology, as we were writing and coding and architecting and structuring data and deploying and all of that, is still there. But what we are trying to bring to the table is have this continue to observe and learn a particular role in the organization, and then see if it can be delegated that work, beginning in bits and pieces. Because here we say AI is probabilistic. People are also equally probabilistic, in my opinion. You ask a question, you don't know what answer is going to come back, right? So AI is in the same mode. Let's say you take requirements. There's a life cycle right from the requirement. Now, what happens? Why do projects become costly? Because the difference between the content of the requirement and the intent of the requirement is the gap, which is really the understanding of the individual, right? And that changes, and then that drift you will see, and the projects are failing, or they are off target, or they are able to satisfy 80% of the requirements. And so that challenge will continue with AI. Content and intent: how do we close the gap? Because if we keep on saying AI is probabilistic, you don't want probabilistic in certain cases, right? I mean, if you want to pay $50, you have to pay $50. You can't say, is it 50, or probably 48, or probably 52? 50 is 50.

No. So the product that I'm talking about is basically a product on which you can model your business functions, with roles and accesses, communication and so on and so forth. Plugging in AI is an aspect of the technology. But the problem that we are trying to solve for is, if you look at business functions, right, most of the products out there, like all these Salesforce, HCM, ERP, CRM, they're all specific to a particular domain. So what we are building is a capability on which you can model any domain, and then you start using it. Because typically workflows and things like that have always been between two people, two roles. Like, if you raise a ticket and you want a laptop to be fixed, there's just somebody else that is there on the side, and it's always just between you and him. But in complex business functions, you will need multiple roles coming in, performing their activities, like the conveyor belt of a car being manufactured, right? Everybody has to come in and then do their work. It's not just two people. So on one platform you can have multiple skills, and these multiple skills need not be in the same organization. They could be from different organizations. Like purchasing a house: you will have the seller and buyer involved, you'll have banks involved, you'll have the city involved. But what is the activity happening? The activity is, Josh is purchasing a house. So if you want to use the platform, then you will say, hey, bank, you come in between these three to four steps and then you do your work. The city comes in, your house inspection, those guys come in. But you know the entire life cycle. But it's more than automation. Automation is just one part of it, because...

Yes, elimination of friction, of no ability to predictably conduct your business. Businesses like predictability. I mean, why does Wall Street respond one way or the other? They like predictability. Without predictability, you don't really know what's going in and what's going out. So automation is just one small part. Predictability of your function, that it is going to perform this end goal and get to that end goal by the time you do it, is what a business is really all about, right?

So that is just, yeah, then you just break down, or take a top-down approach. Break your ability to predict down into pieces that you can predict. Predictability is not about just predicting. I don't say predicting means the future. I'm saying you have a certain task at your hand, and then how would predictability increase? The probability of it hitting a hundred percent will increase if you identify all your external dependencies.

Yes. And for the probabilistic part of it, you give a tolerance limit to the goal, and then you say, this is my tolerance, and then this is...

So we are going after a market segment that is not addressed by the big platforms, because the big platforms, they've all started at a certain point in time, and the big platforms cannot go there, because the platforms themselves have already boxed themselves in. Salesforce, HCM and all these things, they've boxed themselves into it. Now, this market segment that we have identified either does not have the ability to go build itself, or even if it has, it is not its core competency, and so it's not doing it. And then, if it has all of this, it can still go to the big guys, and the licensing and all of that is pretty expensive. And you can just go buy, you know, all these big ones, right? You still need all these costly consultants to come and configure. I mean, SAP for example: we've all known SAP for a long time, and they take ages and ages. Generations come and go, but the projects are still going on, right? So what we have built here is the ability for the business that is going to subscribe to this platform to get in itself and then be able to model themselves without having to write any code.

Yes, something like that.

Yes. And then with that, what happens is you don't need all the bells and whistles that come with these big platforms. You build what capabilities you need on this platform, and then there's no bloat in there.

So that goes by the country that you're going to operate in, and every country has its own regulations, right? So, for example, in India they said data has to be local within the country, and then they have their own laws. Here in the US we have different ones; in Europe we have different ones. So it's first a legal problem to solve for, and not a technology problem as much. Like I said, every new technology has to work with every other piece of technology in there, right? So we will expose the ability either through MCPs, or through APIs if they still want to do the old world. And then once that is there, it doesn't really matter what model they are picking. Because we keep talking about models, but then what exactly is the model doing at the end of the day? All that it is saying is, hey, give me a simple layman's language description of what this API is, or what your software is, and then I will figure out whether that's the right one that I need to use for my work. Right? So they can plug in their own if they want to.

Yeah, it's not about future-proofing your technology. When I go, and then people say, who are your competitors? I always say my competitor is the problem that I'm trying to solve. It's not others that are trying to solve the problem. It's my immediate problem that I'm trying to solve for. So future-proofing is observing how this problem is going to change its nature over the course of time, and not necessarily what technology is coming and what is going. Because regulations will come in, new technologies will come in, new capabilities will come in. Sometimes industries just get wiped out. For example, BlackBerry. Vanished.

So my approach to that is, it's not the technology that you should future-proof. It's the problem that you're trying to solve for. Because what is technology all by itself? AI by itself, what can it do?

Just a few.

The org might scale. But what I see is, as the org scales, having worked with large-scale multinational organizations across countries and regulations and so on and so forth, I can basically bring that to this building, and then ensure that all the inefficiencies that we had there are not creeping in here. So to me, large is not really, okay, I have a hundred thousand people or so. To me, large is really, okay, is it impactful across a particular market segment? And so eventually it will.

Yes, I imagine. Yeah, like so many search engines were there before Google, right? And now Google kind of became the default search engine. But at the end of the day, now with AI being able to answer a question, as opposed to just search through, you know, Control-F, find-and-replace kind of words, now it's a little bit more meaningful in terms of AI going and searching, and you know...

Yeah. So in that itself, you have just seen what has changed: the nature of the problem. Right? Initially, Google is also a search engine. Here also, if you simplify that, in both the cases you are asking, telling it to go search and bring the results back to you. So the nature of the problem was there. People all wanted questions to be answered earlier also. It's just the technology was not there yet. I mean, they've tried building it with things and so on and so forth. But now the technology has come wherein it says, I can answer your question. Right? I had these things way back. I come from the time when Google was just coming and Yahoo was the big thing, you know, AltaVista, Yahoo and all these things. And I used to wonder, it would be nice if it actually answered my question, right? Instead of just saying, here are the words in these thousands of articles.

Okay, that's the next challenge.

But that was there with Google search results as well.

I would love to see the experience of the daily life of a human being when many of these things come together and then be able to give you a connected experience. I think that is possible in the agentic world, but provided the enterprises open themselves so that they can participate in it. What I mean by that is, simply, in the finance world there's a lot of this agentic commerce discussion going on, right? And what does it mean? It means, as you're walking, if all your devices, all the technology that you're connected into, can interact with each other and help you, not necessarily wait for you to ask it a question, for it to be able to come up front and say, hey, you know what? You are driving there, or you're buying this, or you're not looking at your finances, and there is this that is on the horizon. Do you want me to do anything about it? Connected experience is what I really... It's in the mix of IoT plus AI plus the core technologies.

Yeah, but humans have been evolving. They will evolve.