Smarter Decisions Start With Smarter AI

Hey, I'd like to welcome you to another episode of Mission Matters. My name is Adam Torres, and if you'd like to apply to be a guest on the show, just head on over to missionmatters.com and click on Be Our Guest to Apply. All right, so today I have Sankar Krishnan on the line, and he's Managing Director over at Transient.

ai, and this episode is part of our iConnections New York series, our Global Alt Series, where we bring you the best of those that attended and participated in the conference and the event that just took place in New York. So first thing first, Sankar, welcome to the show. Adam, pleasure to be here. All right, so we got a lot to talk about today.

Of course, I want to get into what you're doing over at Transient.ai and really the overall angle, how you can deliver, how AI can deliver better alpha to the buy side and sell side. So we'll get into that and more. But first, iConnections New York, the global alt event.

Now, have you been a part of iConnections in the past? Is this the first time? Like, give me some of the backstory. Yeah, this is our first time being a sponsor.

We have attended that in Miami, and we had a lot of positive outcomes from that. So we said, let's go in and be a sponsor and really happy with the energy that was all around. And, you know, I've told Ron that we'll be back with him in Miami. That's awesome.

I love that. You mentioned you came first as obviously attending, and then you decided to up the game and become a sponsor. Super interesting there to me. What led to you kind of like doubling down, so to speak, and saying like, you know, a lot of different ways you could spend your time, but you're like, no, iConnect is doing something right.

What led to that? Yeah, great question. So we saw Miami was, you know, over 5,000 hedge funds and private equity and such. And we went there more as a client.

One of our clients was a speaker and they invited us to the panel. And we saw the power of a network and power of like-minded people coming together. And we then said, hey, when is this going to happen again? And then obviously it was going to be in New York and which was home.

So we said, hey, this is a natural thing to do. Obviously, the New York one is more from all over the world, but also a lot of New York hedge funds that may or may not go to Miami. So, yeah, I thought it was a good upgrade for us to do that. That's awesome.

Great to hear there. Before we get into Transient.ai, how did you get into this business overall? Give me some of your backstory personally.

Oh, yeah. Like a recovering banker like everybody else. That's a good one. 15 years at Citi, eight years at Standard Chartered Bank, originally from Pricewaterhouse, spent some time with Carlyle Portco.

And really this was two years ago when AI was coming from every angle and then call it midlife crisis. But a bunch of us that have done this forever in large regulated institutions around the world saw a white space that, okay, the kids from the West Coast are amazing, love them, but look, they haven't had the benefit of working inside these 200, 300-year-old institutions would do things in a certain way. And we thought, how do we use our experience coming from Goldman City, Kiritu's UBS and such to help technology and AI kind of work in a certain way at these places?

And we saw a white space and a bunch of us said, hey, are we sure we want to give up our high-paying jobs and go do this? And that was unanimous yes, and that's how we were bombed. Wow. And I want you to go further into that thought process, of, and I'll use your words, the white space, or for me, I'm just saying like there was an obvious opportunity.

You got a lot of like really smart people that, as you mentioned, have a certain pedigree and job and lifestyle status. And like, let's go launch something like that. Pool must have been pretty significant. Like where were you seeing the opportunity or what are some of the ideas around the opportunity?

Yeah, I think first and foremost, this is like a new industrial revolution that has already taken off. So the good thing is we didn't have to go and start it. The fire was already on, borrowing from Billy Joel. I like that.

The fight was already on. Go ahead. Yeah So we said that what is it that we bring to the party We bring veteran experience in capital markets coming from all those big brands around the world And really the big brands also were too scared to consume the way the AI was being served to them by the frontier models or whatever LLMs they wanted to do. The problem was no one really knew what were the new risks they were getting into.

Obviously there was a wide agreement that the technology works, But the last thing you want to do is put your secret confidential algo that's making you a lot of alpha into some frontier model that gets reported somewhere into the Internet and everybody else kind of gets that, you know. And there's a lot of leakage that exists and people may or may not want to acknowledge that. So I think what everyone was saying was, okay, guys, with all the experience, why don't you tell us how can this AI work in regulated institutions that are highly auditable by the SEC and the Fed, highly verifiable for a lot of data, and a lot of AI has to be very explainable with a lot of governance around the reasoning and obviously boxing out the hallucination whenever that happens.

So I would say that a lot of us have done this in the traditional sense with a lot of trading technologies, got a lot of things right, but equally got a lot of things wrong. So that wealth of that experience was kind of directed into managing all these new AI models through a layer that works for anybody and everybody at very low cost. The second thing out there was everyone gets excited, including me, and we all subscribe to all the models out there. But before we know it, your membership went from $200 to $2,000 to $20,000.

And now you have burnt all these tokens and everything is in POC. Now your big boss comes in and switches off those tokens because everyone is saying, okay, you have spent a million dollars on these smart ideas, but where is the ROI? So what happened was even the POCs that were actually good couldn't go into production because they failed the ROI test for tokenomics. So we said, man, there is a better mousetrap out there that we can design, go configure, and load balance it a lot better for these institutions to actually make some ROI first, maybe save some costs in the way legacy systems operate, and then redirect those savings into the new AI model so that from a cost perspective, you're net neutral.

So whatever else you gain from new revenue is net new alpha. So that was the thinking as we designed this. Just maybe going a step further in that. And thank you.

That's a great overview. Maybe going a step further into thinking about like maybe pros and cons of what it looks like. You mentioned engaging in different types of models. So maybe pros and cons of directly engaging with, let's say, frontier models.

Yeah, I think the pros are, A, that technology works. I think the second thing is that the speed and compute of those models is amazing. and three, if you are somebody that knows how to prompt them the right way, they deliver good results. Now, if you are a very, very large bank that has a $10 million budget to spend, you're fine.

Otherwise, it's a little bit like using, I'll just use a kind of bad analogy, but it kind of explains it like a $2 million weapon chasing a $200 drone. And that doesn't make any sense. So the same thing is very true of these amazing frontier models. The real truth is a lot of institutions of various sizes, actually a small language model would suffice to get to where they want.

And then you graduate into a large language model. If the computer is more, the challenge is more. And obviously, for sure, frontier models, when you have to do years of activity, for example you go into a capital market trading floor and you want to analyze 10 years of swap activity the small model may not be enough and then you have to use a frontier model So for a given amount of money I think everyone has to deliver to their shareholders and it's a lot better if you're able to have a load balancer that actually mixes the small language with the large language and the frontier manage.

And that's what our AI layer tries to do. When it is launched either as an SDK or a download from a marketplace, you're able to use it to configure to whatever data you have, be it an Excel spreadsheet on one side or a quantum computer on the other or anything in between, and make more sense of the clean data that comes out and then layer it with some good AI that gives you the outcomes you want for better sales and trading, better investment decisions, better real-time decisions, and so on and so forth.

And so if we look at the landscape or just the ingredients here that we're talking about, like the result, at least when we're talking about the fund space? Like how can this, I won't say how it does, but how can this deliver better, like using AI, deliver better alpha on the buy side or the sell side? Yeah, that's a great question. So first and foremost, I think the time taken to do these things are shrunk by a huge number.

So we did this analysis that what would be months and months of work would now reduce to weeks of work at the maximum. And that itself gets you savings in cost. that gets you faster time to market. So that is just for starters.

The second thing is that we are able to take whatever the data sets exist within the enterprise and then mix them up with real-time feeds from all over the place. So let's say that I'm trading two commodities, copper and oil, and I'm about to hit the copper buy button. Before that, I kick a transient tap that kind of says, hey, you know what? The president's just tweeted something and that's actually going to impact the oil prices.

And already it is trending in a certain direction. So depending on the type of trade you want to do, you're better off trading oil rather than copper for the next three months. And given the same time period, given prices that are moving in a certain direction and given all the Greeks are moving in a certain direction, it enables you to make the better trade and therefore more alpha for you. So it's not the trade itself.

Obviously, it is a recommendation engine to the human in the loop and the human in the loop then decides what is better for him, which commodity to go after. And that real-time information acts as a deterrent to a bad trade and acts as a promoter to a good trade. So one is able to do all of these things with the speed of AI. Amazing.

So as we look, and I'll kind of timestamp this for us. So recording this June 18th, 2026, and not asking you to quite pull out your crystal ball, Sankar, but just thinking about when we think about, entrepreneurs that are listening to this, allocators, LPs, all the above, executives. And if we're looking out five years or whatever that may be, how do you see AI affecting the future of the capital markets just in general? We're talking a little grandiose here, but in general, I know that's a big question, but I'm doing a timestamp, okay?

Absolutely, absolutely. Look, I would say the following, right? So obviously, we did a lot of equity research in my earlier job. I mean, obviously, it was costing X hundred thousand dollars in the US.

And obviously, the model was a knowledge process offshoring company would kind of do it for a lot less, would be 80% right hand it off to the chief editor for a research report, and they would come in and do something. So a lot of these were about 40 to 50 person savings and a lot of offshore and all those things involved. Now straight of the bat on-prem, to agent could do that much better much faster much cheaper 24 without the need for all those edits because the editorial agent would already have edited it and sent So I talking about the lifetime of a research report shrinking when these research agents get involved and so on.

Second, I think hopefully it creates more jobs in the future. We all believe it will. But for those places that do not have the where with all to hire a lot of smart talent, I think the human in the loop is able to navigate the agents to do as much as what is needed. So they're able to manage their shrinking budgets much better and rapidly evolve into something with a lot more accuracy.

That is the second big thing as we see it. I think a lot of businesses are going to be much more revenue intensive in outcomes, but I think they are going to operate with a lot less people. But that doesn't mean they're inefficient. They're going to be more efficient.

So I think the whole of capital markets activity like we know it will change drastically. And just like there are characters in real life that are good actors and bad actors and good and bad always fight over and what happens, happens and good hopefully wins in the end. The same thing is happening of a lot of AI and there are bad actors and bad agents that are going after customers and so on. So one of the cons is that a fraud can be perpetrated also with better ease than what is the case today.

And then we have to program the good agent to identify the bad agent and therefore good prevails over the back. So I think the entire world of capital markets as we see it from research to origination to workflow to processing, the whole concept of discovering alpha, executing on alpha is changing very rapidly. And I would say that it's here and now, and it wouldn't take five years. I think the fact that the SaaS companies have taken a beating, while some of that must be overdone, I think there is a lot of AI that can actually help you control your SaaS spend with better AI.

So I think overall, as an industry, things are going to get more efficient. And I would say it's here and now. I would say in two, three years, if you have not embraced this change and moved on to a better state, I think you're going to get left out. So I think it's changing everything as we know and making it a lot better.

And I'm sure a lot of people, when the industrial revolution happened, I was not here. I'm not that old, but I was definitely here when the dot-com revolution, the internet revolution, the cloud, and all of that happened. So we all had these misgivings earlier, and then it all happened. And I think it's happening now in AI.

At some point, everything will become a commodity, like it usually happens. But for now, it's a great value addition mechanism to every business. Well said. Sekhar, this has been great having you on the show today and learning more about the work you're doing over at Transient.

ai. That being said, if someone wants to continue the conversation, follow up with you and your team and connect and follow the journey, how do they do that? Okay. You can go to Transient.

ai, which has all the details of the contacts. Or I'll even put myself out there because I believe in this and love what I do every day. Sankar K at Transient.ai.

Anyone wants to reach out, happy to return their call and give them a demo for what we are doing and share some thought leadership that they can use for themselves, making their business more efficient. Wonderful. And for everybody listening, just so you know, we'll definitely put some links in the show notes so you can just click on the links and head right on over and connect with Sankar and his team. And speaking of the audience, if this is your first time with Mission Matters and you haven't done it yet, hit that subscribe or follow button.

This is a daily show. You heard me correct. Each and every day, we're bringing you new content, new ideas, and hopefully new inspiration to help you along the way in your journey as well. So we don't want you to miss a thing.

Hit that subscribe button and say, Kar, thanks for coming on. Adam, pleasure. And nice to see you take this message to a global audience. And I've seen several of your podcasts.

Very interesting. Great being here with you, Adam.