Why Trust Is Agentic AI’s Biggest Bottleneck

Hey, everyone. Welcome back here to Techstrong TV. Let me introduce you to our next guest.

It's his first time on here. I'm excited to have him on. His name is Anudeep Parhar.

Anudeep is the chief operating officer for Digital at Entrust. Let's welcome him. Anudeep, welcome to Techstrong TV.

It's great to have you here. Al, it's wonderful to be here, and it's a pleasure to be on. Thank you.

Thank you. Anudeep, as I mentioned, you're the COO for Digital Entrust, but give people a sense of who they're watching here. What's your background?

So, in terms of background, hey, I've been a technologist for the last two and a half, three decades. This is pretty much all I know how to do. I've been on both ends of the industry.

I've built and sold software and technology, and I've bought and implemented. I've been with Entrust about 10 years, and it's been my privilege to lead several functions here. And as you can see, from a company perspective, I'm responsible for our digital operations, our both internal and external digital operations, and I'm also responsible for launching our AI initiative as we go forward.

I love it. Everyone has AI initiatives today, right? Anudeep, and Entrust has been around a while.

You're there years already, but there might be people listening, watching today who aren't familiar with Entrust, let alone Digital Entrust. How would you describe Entrust to them? So, the way I would describe, Al, to your audience is, Entrust is, as you mentioned, have been around for almost six decades.

So, we have been in the cybersecurity business for a very long time. But we are an identity-centric security company, so we address both physical and digital security, both fighting fraud as well as cybersecurity from both a physical infrastructure as well as digital infrastructure perspective. So, we provide software products, hardware products, which help you secure your physical assets and digital assets.

And we keep identity as a center of our attention as we go forward. I love it. Thank you.

That helps a lot. Six decades. Wow, 60 years.

Can you imagine? When we say Digital Entrust, I assume then that that means your focus is on the digital security aspect of things. That is correct.

Yes. A part of our portfolio also addresses physical security, and- Right ... it's a very critical and important part of our portfolio, but I'm responsible for the digital part of the portfolio.

I would imagine the physical part probably came first because six decades ago, we didn't have much digital security, did we? But today, digital security has taken on, obviously, a much higher profile, much higher visibility as every day it seems there's a new breach, there's a new vulnerability, there's a new security issue, et cetera. Anudeep, if it's okay, though, I'd like to pivot a little bit and now that we've got your background and understand your role at Digital and Entrust, what Entrust is about, I want to talk about why trust is AI's biggest bottleneck today.

Well, I'm going to say that it's the second-biggest bottleneck. The biggest bottleneck are people who are starting to get the bills for all this token maxing that's going on and realizing maybe it's cheaper to use people. But beyond that, trust is a real issue with AI.

Is it hallucinating? Is it lying? Is it just flat out wrong?

Can we trust it to do the jobs it claims it could do? And is it a security risk? Go ahead.

Definitely. So, very good question, Al. So, the way we have all experiencing AI is, obviously, we see the enormous power it brings to us.

Mm-hmm. The ability of democratizing intelligence, automating tasks that were not automatable before, and providing, in the end, a very simple conversational interface to be able to do complex tasks. That's a pretty powerful thing.

You've already mentioned the cost is becoming an issue, but from our perspective, the way we see it is we take a longer view of where some of this transformation is going to happen in the enterprise, as well as for the consumer. So, we believe that eventually where we are going to end up is an autonomous enterprise where AI agents will do a lot of tasks autonomously. They will not just inform you, they will not be just a way to retrieve information from existing systems, but they will also take actions.

And we also believe that it's going to take us a while to get there, and the human definitely will have to be in the loop in all of these transactions and automated things that happen. So essentially, agents working twenty four seven, hundreds and thousands and millions of agents doing work for an enterprise, creating personal productivity, functional productivity, and enterprise productivity, ultimately leading to both productivity as well as growth for the organization. So, that's sort of the view we take where the agents are going to go.

But we also know that the facts on the ground, other than the fact that you mentioned, which is the cost and the token maxing, is that anywhere from 70% to 90% of agents that are being built today do not make it into production. They are experimentational, they are more in pilot or POC phase, and they don't make it into production. And if you dig a little bit further into the numbers, about 20% of those 70% to 80%, or 70% to 90% Of agents don't make it into production because of trust issues.

There is not enough security, there is not enough belief in terms of saying what the intent of the agent was, did the agent actually do what it was supposed to do? Was the action that was taken within the policies of the organization? So there is a broader capital T trust issue from those agents moving into production.

And if you move specifically into the regulated industries, financial services, healthcare, governments, and in all of the industries that are regulated either regionally or because of the nature of the business they're in, that number goes up to 50% to 60%. So if you look at a very broad swath, what has happened in the last couple of years is the proliferation of agents and technologies has commoditized or made it easy, so to speak, creation of agents. But what's stopping it from getting those agents into production is the lack of trust plane or trust infrastructure, where enterprises, the CSOs, the line of businesses, the CIOs, and the CTOs can comfortably say the intent of the agent was in fact documented and understood.

The consent that was provided to the agent was in fact documented and understood. And if you're in a regulated industry, if an agent and its sub-agents in the agentic ecosystem does certain actions, those in fact can be proven after the fact. They are replayable when either your risk infrastructure or your risk officers come and ask you for it, if you have to defend yourself in an audit, or God forbid, in a court of law.

So the evidentiary proof of what the agents did, if it was within the policy, within the intent, and by consent of the said individual, it's extremely important. So that's why we see that trust is a big factor. Building agents is easy.

Putting them in production and actually getting the productivity and growth benefits, that's where the trust comes in. Yeah. And that's also, unfortunately not where the value comes in, right?

Because if I can't get that working, even if I trusted it, it's not enough, right? And we're finding it pains... And Anudeep, you've been around the block, I've been around the block.

It pains us to watch people who try to leave big pieces of this equation out of the equation. You can't get to the profitability, productivity piece without trust. Right.

But trust alone isn't necessarily enough either. You've got to move from the lab to production and make sure it works. And sometimes, even in this age of AI, there's no shortcut to doing that, right?

You got to do these things. So let's talk about the Trust Accelerator a little bit more, though. What exactly is it?

So we've taken a very Entrust approach to this. We acknowledge that AI is not a fashion, it's not a fad, it is here to stay. We believe it's going to have huge impact on organizations, and it's going to have a huge impact in the positive for our customers and prospects that we work with.

So with that, we also have vivid all or enough of a self-realization that a lot of work needs to be done here still. So which means is that infrastructure and how exactly some of this trust will manifest itself in enterprises needs to be worked on. So what we are doing is we are launching the Trust Accelerator, where we're going to work with select customers that we work with today, and we're going to invite others into the accelerator as well to help us define what that trust solution should be.

With a very clear outcome to be able to say, agentic enterprise is here to stay. It will deliver the value, like you said, of profitability and growth for the said enterprise, and trust is going to play a very, very critical role. And that trust plane is yet to be defined, and we are here to help our customers define that.

I love it. Anudeep, with time we have left, if you wouldn't mind, walk our audience through, they're saying, "I buy into this. Trust is important.

I want to find out more about this Trust Accelerator. I'd like to get that. " What's the on-ramp here?

The on-ramp, the best way to do it would be is we just launched the site. Please go visit us on LinkedIn or go to our website. Call us, talk to us.

If you have an existing Entrust representative, talk to them. All of our field staff is fully equipped with this. Otherwise, hey, I'm on one of the best shows on TV right now.

Call me. Absolutely. I'll be happy to help.

There you go. What is the website, though? What's the URL?

com/accelerator. /accelerator. I love it.

Yeah. And people can get... Is there any kind of trial or initial kind of thing they can get their hands on?

At this point, we are collecting information from our customers, which is we want our customers and prospects who want to try the accelerator or want to be part of the accelerator to have an on-ramp on either one. If you're experimenting, there is an on-ramp for that. If you have, for example, an executive sponsor, you have funding available, there's an on-ramp for that.

And then we have existing customers who are already working with us and partners and other tech providers who we are working with. So there are several ways of making sure you can benefit from the accelerator and help us help you build trust into your AI infrastructure. I love it.

Anudeep, thank you so much for coming on here today and educating the audience. Appreciate it. Continued success and best wishes with this agentic AI Trust Accelerator.

Everyone's playing with AI, as I think you mentioned. But if we can't build that trust, we're never going to get to where we think it can go. And it's such an important part of it.

100% agreed, Al, and I'm glad to be on the call, on your show, and it's lovely to talk to you and your audience. Thank you for your time. Thank you.

Anudeep Parhar, Chief Operating Officer for Digital at Entrust. com/accelerator. Get some trust for those agents you're using out there.

We're going to take a break here on Tectrum TV. We'll be right back.