The Real AI Bottleneck

Hey everyone. Welcome back here to Techstrong TV. My next guest is Ashwin Ramachandran.

Ramachandran. We're going to try to make sure we get that right. But Ashwin is SVP of product management with a company called Precisely.

You might have heard of them. If not, we'll tell you. But let's first welcome Ashwin here to Techstrong TV.

Ashwin, it's great to have you on. How are things? Thanks so much, Alan.

Really happy to be here. Things are great. It's a busy, exciting time in the market, so there's never a dull moment.

That's for sure. Never a dull moment. Every weekend.

And it's not like I take the weekends off. Yeah. I'm checking all my email, I'm reading all the news, I'm staying on top of everything.

But when you look at the cumulative from when you leave your office Friday to when you come in Monday or you go through your Monday morning, it's amazing because not only has all the news come out, but you've had a chance for people to kind of weigh in. Yeah. And you get all different views.

It's just nuts. But it is an exciting time. Ashwin, I don't know how long you've been SVP of product management over at Precisely, but I'm sure you've had a run getting there.

Why don't you share with our audience a little bit about that? Yeah. So I'm currently SVP of product at Precisely, focused on our integration domain of data products.

My path came up through the engineering side. So I started out in the industry kind of working in a tech customer support type role. Mm-hmm.

I was the person that you'd phone up back then or if- Sure ... you bought our products and you needed a license. If only we could still phone people, Ashwin.

I know, I know. It's all digital now, right? Yeah.

m. production down calls, right? Where, "Hey- Yep ...

product's not working. It's costing me millions of dollars every minute," right? That you can't get it back up and running.

I kind of worked out from that domain into pre-sales for a bit, and then really found my calling, if you will, on the product side. It brings together a lot of capabilities around understanding customers' problems. Not just what are they asking for, but what's behind the question or what's behind the query that they're giving you, and how do you find a solution that not only solves their problems but also solves the market's problem beyond one customer, right?

And so, yeah, it's been a great run. Precisely, we are the data integrity leader. We're focused on ensuring that data is trustworthy for your operations, your analytics, and increasingly now, your AI agents that you're building to, say, comb through that bevy of emails that you might get over the course of the weekend and understand what needs your attention and what doesn't, right?

It all starts with trusted data, so we serve over 12,000 customers, including most of the Fortune 100. So yeah, it's a really exciting time for sure. It is, it is, and you guys sound like you're right in the mix of it.

You know what? For people who may want to dig in on Precisely, what's their best path? Couple different places.

com. There you'll kind of see solution stories, customer stories, what we offer. You can obviously follow us on LinkedIn.

You can find me on LinkedIn as well should any of you want to connect offline. I love chatting with folks about data integration and where those things come together to unlock value. Absolutely.

You know, Ashwin, I've been involved in IT and tech, more than IT, tech. I've been involved in the tech industry almost 35 years, man. Yeah.

Mostly founded, co-founded a bunch of different companies. Been on the media side of things here about 12 years. Mm-hmm.

I've seen this pendulum swing from closed to open, maybe back a little bit the other way, back to closed again. Yep. It's constantly swinging.

As we enter this AI era, right, and AI increasingly determines so much of what and why and how we're doing things, this whole open versus closed decision, once again, it rises right to the forefront, and I don't want to say it's life and death, but it's critical, right? You could go down-- We're at a fork in the road. " Yeah, yeah.

So as enterprises are coming up now and making decisions, first of all, let's talk about the decision itself exactly and define it, but then let's talk about some of the things they should be thinking about. Yeah. I think if I look at just from a definitional layer, the way closed versus open as a decision point comes up in conversations with my customers is a closed platform is one that might be more monolithic, doesn't necessarily give an end user or the choice as to, hey, what controls do they want?

How do they want to compose different point products together to deliver an end solution? Whereas an open platform is one that's focused on things like interoperability, customer choice, and control. And if you look at the enterprises that we serve, typically Our customers are asking for openness for fears of things like vendor lock-in, for choosing the best solution for the task at hand, and getting that one plus one equals three composition.

So that's how, yeah, I hear the open versus closed decision- Yeah ... get voiced. Look, the thing about a closed platform, and I'm not knocking anyone, but Apple might be the best example.

Right? You are kind of locked in. It is a black box.

Yeah. On the other hand, if you believe them, and I'm not saying it's true or not, you get better governance, better security, better predictability. Mm-hmm.

Yeah. These are not things to be scoffed at. So, there is an argument to be made for that if you want to go down that path.

But like you said, to me, the biggest neg, the biggest con on the-- you always make your list of pros and cons, right? That's how we were taught to make business decisions. Yeah.

The biggest con is locking. Right? You're locked into that vendor.

Can you ever move off of that? And that brings me to my next point, Ashwin, which is, are we too early in the game- I think it's a really- ... to make that decision?

Yeah, I think so. Right? If you look at January to April alone and just look at Anthropic and all of the product innovations they've done on Claude and the harness of tools with things like Claude Code and Cowork, et cetera, now Design, right?

The space is just storming like crazy. I think making those bets now is probably going to be a shortsighted decision. What I've seen and even what we do at Precisely, right, is we're focused on ensuring that we can try different solutions out, understand, hey, which one is right for the use case at hand?

In a year's time, from trying to predict the future here, in a year's time, we may determine, hey, tool set or vendor A, taking that closed ecosystem approach meets the needs of our business. There's nothing like it, right? That might be- Right ...

completely suitable. On the other hand, we may determine tool set A is great for these use cases, tool set B is great for these other use cases, and now we need to go ensure we got the right governance layer around that, the right set of controls around that. Mm-hmm.

I think it's just a little too early. Even within our lines of business, just speaking as a customer and user of AI technology. As a consumer.

Right? Yeah, exactly. I'm not ready to put my eggs in one basket yet.

No way. I agree with you. I think this is why we hear a lot this term harness thrown around.

Yeah. Right? I want a harness that is going to let me use a whole bunch of different agents from a whole bunch of different people.

Yep. And I'll harness them and pick which ones I want to move here, there. And maybe one is open.

And if I go with the closed, I lose functionality and- Yeah ... I know AWS had a thing in New York last week where they talked about- Yep ... some functionality like this.

I think we're still over there a little bit. I read another interesting article. We were talking about the weekend trying to say.

Yeah. I read an interesting article in Medium this weekend where, with each successive release, and certainly with Anthropic, but I don't play favorites. Do you want to know the truth?

I use them all the- Yeah ... Look, every release of OpenAI, of Chat, and Codex and all that is also equally as good. And- Yep ...

I personally use a lot of Perplexity computer, which uses- Yep ... Gemini in the back end, which, well, it uses all of them, including Gemini. But Gemini is also making strides, and- Yeah ...

it's hard to say I'm going to bet against Google. Yeah. I don't know what to make of the Microsoft strategy, but whatever.

There are a lot of people who use the Microsoft Copilot system. Sure. The thing is that the level they keep improving, the premise of this article is, how many more iterations, how many more next releases do we have before- Yeah ...

not that the AI development hockey stick stops hockey sticking. Yeah. But before it gets, quite frankly, then it'll be like enriched uranium, right?

We can't give it to everyone. Yeah. And so, we'll plateau off at maybe a level or two beyond Fable.

Right. And then you could start making these kinds of decisions, right? Yeah.

I certainly think that that is the point in which you decide, okay, what's our approach? Are we going closed, out-of-the-box solution that simplifies adoption, or do we go open platform where we can combine the different things that we care about, and we're sophisticated enough to then go put those governance layers on top of it, right? Mm-hmm.

I think where I focus on oftentimes is the data level of all of that, right? Because you can get to the Fable plus two era of models, let's say. Right.

Right? You're still going to have a fundamental data problem, and that's going to result in bad stuff. You can have the greatest model in the world.

If it's operating on bad data, it's the garbage in, garbage out thing. I think the real question if I look at the enterprise scale is, hey, how do we get our data estate under control, fit for use, and well-governed so that as we're making our choices around open model or open AI stack infrastructure versus closed AI stack, we're ready to go. That's one of those big questions I think enterprises are wrangling with, and I feel that pain every day.

I use my- Yeah. No ... and it's like, hey, I don't have this right piece of data context, so whatever the model is confidently telling me is totally wrong.

Right? So yeah, it's an interesting problem, for sure. It really is.

But now I got to ask you to take off your Ashwin hat and put on your SVP of product hat there precisely, right? You got to make decisions around product- Mm-hmm ... that, as you said, most of the Fortune 100 are using.

Yeah. Or a good chunk of them, anyway. That's a pretty awesome responsibility, right?

Yeah. You can't afford to be wrong on this. Don't mess it up, Ashwin.

No. But what do you do? How do you evaluate and what do you do?

Hedge your bets? Make a bold bet? I don't know.

Look, so just for a little context, Precisely has been around, or the founding of the company goes back to the late 1960s. Wow. In fact, I believe most enterprises bought a piece of software for their IBM mainframe environments from, so back then it was called Syncsort or Whitlow Computer Systems.

Uh-huh. We were one of the first places they bought software from that wasn't IBM for their Z environments. We operate in that domain today still.

A really great portfolio of solutions in that space today. And then we've got a bunch of great solutions in the data space, in the location data space, and in the governance arena as well. So we span a pretty broad array of solutions.

What that means is we also serve a pretty broad array of buyers and users. Everyone from ZOS systems programmers to GIS analysts, data engineers, data stewards, and the CDO office. And the reason I say that is when we have to make these decisions, we have to take that broad customer base into account, right?

Everybody's at a different stage of maturity, and so it really just comes from conversations, listening, asking the right questions, and we have a couple mechanisms we use to understand where customers sit. We leverage forums like our customer advisory boards, where we invite our top close customers, understand what their plans are, share with them what our plans are, and use those forums to have them shoot holes into what we're doing, what resonates, what doesn't resonate, and we use that to adjust. And we definitely take risks and we take bets.

One of the things that we've launched in the past several months was our AI assistant, Geo with a bunch of data management agents as part of that. The bet that we took was, given our heavily regulated customer base, we said we're going to go down the control route. Is we're not going to go and prepackage a bunch of the capabilities for you because you're a heavily regulated enterprise inside of financial- Mm-hmm ...

services or insurance, et cetera. You're going to want to maintain control and the cost envelope of the models that are running and manage it the way you manage the rest of your business. Now, that's probably a conservative bet.

Now we're kind of looking at, hey, what opportunities do we have to prepackage more of these things for you so that it's more simple out of the box? But yeah, we have to take a bet, right? We have to go, given our understanding of our customer base, kind of have to make a decision.

And then we react to that and figure out, okay, what's the next thing we do from there? Part of working in product is you do a lot of signal sensing, fact-finding, feedback collection, and then you got to synthesize it, and then you make a call. And the key to it is then being able to react quickly to what you learn in the market.

Absolutely. And I think never more so than now. Yeah.

I bet you if you and I wrote, if we were here five years ago, we'd say that, too. Five years before that, we'd say the same thing, too. Whether it was cloud, especially a company like, as you said, Precisely.

You've been around since mainframe '60s, right? Yeah. And every new era seems to foreshadow the ones before it, but this just feels-- To me, this is the closest thing since electrification, right?

And that might be what it's going to take here as well, that kind of effort in terms of putting it in. But as we said, bringing it full circle. It's exciting.

Yeah. It's exciting to be doing it. It's exciting.

Let me ask you one other question, Ashwin. So what do you, if you don't mind sharing, what are the telltales, what are the leading indicators that you watch that help you make decisions around this? Ooh, it's a good question.

I think on the product strategy side, I think leading indicators, I'll give you a great example. I'll go back to the customer advisory board context. We hosted a session back in the fall with this close group of customers where we said, "Hey, where does AI sit in your operations?

Is it you got an application in production? " Six months later Tune has completely changed, where people are now empowering their teams to go and figure out how do you go and take existing business processes and rethink them. Right?

And so I look at that from a leadership level with large enterprises as one of those indicators for me that, okay, the appetite within this customer segment, which is our ideal customer profile really, is such that they're now looking at how do we actually operationalize these things, and they're probably going to be ready to leverage our products in a much more agent-first way in the next six months. Right? That's one of those markers that I use, which is just kind of, hey, what do our really strategic customers, what are they doing?

How are they thinking about the problems, and how are they prioritizing their investment? And that's one of those things. I think the other thing is just looking at, in terms of my team, how we are rethinking our operations top to bottom.

We are not doing product management in the same way today as we were even four months ago. Yeah. We are actually shifting where we're spending our time.

There was a lot of time spent in the past on things like requirements drafting and documentation. Yep. Gathering.

That's still an important part, but now our energy is spent on validation, refinement, and like- Yeah. The bottleneck, this theory of constraints stuff, right? The bottleneck, the cheese has moved, if you will.

Right. Someone moved the cheese to the next constraint, to the next bottleneck, or the next focus. Yeah.

I think we're all living that right now. We all are living-- That's exactly where we all are. Right?

Yeah. Whether it's in security, we're not hunting for vulnerabilities. We have more vulner-- They're being hunted.

We got to figure out how to remediate. Yes. In product, we don't have to do the data gathering, the background.

Yeah. We've got to look at the analysis, and even when it starts doing the analysis, we need the human in the loop. Yep.

And- Well, one thing I would say on that specifically is a really good point is, we know that vulnerabilities are going to exist. The value marker is not are you zero vulnerability, but how are you organizationally set up to go and respond to that, right? Because if- Yeah ...

you have to assume these things are there now. Similarly, we talk about moats a lot in the era of AI. Hey, what is the moat of a given company?

If it's cheap for now to go and build something, right? Then the value is not so much in building it, but it's really in having the judgment to determine what you should build, what you shouldn't build, and how quickly can you respond to feedback. Like if you think of lead time as a metric, from signal- Right ...

validated signal that this is what we should be doing, what we shouldn't be doing, through to deliverable. That's what you really care about optimizing now. And to the question of leading signal, we try to collect it everywhere.

Through our product analytics, through market research, through customer conversations, collating that quickly, identifying the signal from the noise, acting on it, and delivering, that's what we're focused on is really optimizing that pipeline now. Absolutely. You got to be reactive to it.

Ashwin, we're probably way over 15 minutes. I don't have a clock here, but it was a delight talking to you. Yeah.

I enjoyed it very much. I hope our audience liked it as much. com, I recommend.

Maybe come back on in a couple of months. Tell us how things are progressing, which way you're going. Yeah.

Do we continue to look open versus closed and all of this. Yeah. But for now, thanks for coming on.

Appreciate it. Thanks so much, Al. All right.

Ashwin Ramachandran, SVP of Product Management at Precisely here on Techstrong TV. We're going to take a break. We'll be back with more.