How GTT Builds Networking And Security As A Service For The AI Era

AI is everywhere right now, but the hard part is turning dozens of pilots into durable systems that actually run a business. We sit down with Tom, SVP of Product Management at GTT, to unpack what changes when enterprises move from “testing AI” to deploying agentic AI at scale across thousands of sites, users, and applications.

We start with how GTT thinks about networking and security as a service, and why the promise is not just bandwidth or a product SKU but a simpler experience that helps customers connect, secure, and simplify. Tom explains the Envision platform and how it spans the edge, the core IP backbone, and public cloud so teams can deliver consistent connectivity, SD-WAN, and security outcomes while preparing for new AI workloads that increasingly want compute closer to the premises.

Then we get practical about what agentic AI requires: data readiness, trustworthy context, and APIs that let agents act safely without constant human validation. We talk frameworks versus one-time deployments, why vendor lock-in is riskier in a fast-changing AI cycle, and how an “AI factory” mindset brings manufacturing discipline to data pipelines, orchestration, validation, deployment, and continuous improvement.

We also share real internal examples, including a cash application agent that helps match remittances to invoices across messy real-world variations, plus how GPU infrastructure supports operational intelligence and proactive network issue detection. If you care about enterprise AI, SASE and SSE, edge computing, and building a scalable agentic architecture, this conversation is built for you. Subscribe, share with a teammate, and leave a review, what part of your AI foundation needs the most work right now?

Transcript

Welcome And Guest Introduction

SPEAKER_00 0:01

Hey everybody. Today we're talking with the leading networking security as a service provider, GTT, about all things AI and networking. Tom, how are you?

SPEAKER_01 0:11

I'm very good. Thank you, Evan. How are you?

SPEAKER_00 0:14

I'm well. Thanks for being here. You're Senior Vice President of Product Management, and you've been with GTT for uh nearly a decade. Fantastic journey. Tell us a little bit about your role and that journey at uh in tech is uh is an Eon.

SPEAKER_01 0:31

Yes, 10 years has been good fun. Thanks, Evan. So I'm responsible for all product management at GTT globally. And so that's my the team are responsible for thinking about what products we sell, how we build them, how we uh sell those services to our customers. Uh, and we're also responsible for some of the enterprise systems that support those products and making sure that that the changes and innovations that we have within the company are are product led and supported, supporting our product strategy. As you say, I've been at GTT for 10 years. I've been in leading the product function for the last two. Before that, I was uh leading part of our solutions consulting or sales engineering team. So I spent a lot of a lot of those 10 years talking to customers about how they uh how they use our technology and some of the exciting projects which they have in their businesses.

What GTT Does Today

SPEAKER_00 1:22

Fantastic. And before we dive into a lot of topics around AI, uh the practical deployment topics, uh, tell us about where GTT is today. How do you describe the company these days? You've been through so many uh uh changes, you've invested so much. Uh, where are you as a company at the moment?

SPEAKER_01 1:42

Well, GTT is is a global networking and security as a service company. And our our mission is to connect people and agents to data and applications anywhere in the world. And one of the things that I like about that mission is that that that core purpose has has not really changed in in a long time. And so as you say, GTT has has been through a few phases in its life. But that that core mission of connecting people and machines, agents to data and applications anywhere in the world has been a constant. And what's changed is how we deliver on that mission, how we deliver our products and services to our to our customers. We um we talk about our our the business outcomes we deliver as connect, secure, and simplify. And for a networking and security as a service company, connect and and secure are fairly self-explanatory. But for me, Simplify is where we are delivering a lot of value for our enterprise and wholesale customers because there we're recognizing that what they are doing is a uh is complex, that the environment that businesses are operating in today is complicated. And a core part of GTT's offer as a managed services provider is to simplify that through simple products which are easy for our customers and partners to interact with and you know make their life, their day-to-day easier. Yeah, we're also we're also still global and increasingly global in a in a in a market where some organizations are retrenching, GTT is investing. We are we are expanding our our network, which is which is now the third largest IP backbone in the world. We're opening new in Latin America, in Southeast Asia, in uh increasing our density in Europe and North America. So we are we are still expanding, we're still adding new products and services, and it's uh it's a you know it's exciting, exciting time. Last year we we talked a bit about Envision, GTT Envision, you and I. And uh Envision was a new platform that we we launched to our customers in 2025, and it was it was bringing together technologies and intellectual property which GTT had been investing in and building for 10 years or more. And yeah, it it really focuses on the experience of delivering those products and services to our customers. Many of my colleagues have heard me say many times that you know that the product that we offer at GTT shouldn't be a 100 meg DIA or whatever, it should be the experience of buying that that product from GT. And uh the Envision platform is how we deliver that experience.

Inside The Envision Platform

SPEAKER_00 4:25

Fantastic. Tell us about the GTT Envision platform. You really tripled down on development. It's it's quite novel. Um, where is it today? What's some of the feedback you're getting from real customers, and and where is it heading as you advance the platform over the course of the year?

SPEAKER_01 4:43

Yeah, the vision platform has been really well received. As I as I mentioned, it's uh it's it's building on technology which we've had in the uh in the in the market for some time. So you know the while the way that we talk about it is it's new and the many of our investments are are new, the underlying technology is mature. And so you know the customers like the fact that it is it is uh it is a a stable and a and uh well advanced platform into which we are investing lots for to build new and exciting things. Yeah, that that platform extends from our customers' location on the on the edge through our core network, as I mentioned, now the third largest IP backbone in the world, out to public cloud. And we use our expertise in in virtualization and providing networking and security services to deliver a cohesive experience across the edge, the Envision Edge platform, through our Envision Core and out into public cloud. And we've made lots of investments in the last few years, and this year we are uh focusing on enabling more of our of our customers to deploy their own workloads onto some of those uh that Envision infrastructure because for lots of reasons, some of which we might talk about later when we talk about AI, you know, customers are looking to bring their compute closer and closer to their to their premises, and that's something that Invision is really able to enable for them.

rom AI Pilots To Real Use Cases

SPEAKER_00 6:12

Fantastic. So I was at the Gartner Symposium the end of last year, and um AI experimentation was everywhere. There was one insurance company on stage talking about 74 pilots of different AI applications that are underway, kind of amazing. You have a unique perspective on enterprise customers and their journey to adopt AI. What are you seeing and hearing from those customers that must be across the spectrum in terms of readiness?

SPEAKER_01 6:44

Absolutely. It's certainly, I would say it's it is a it's changing. You know, a few, I'd say 12, 18 months ago, there was an awful lot of experimentation, a lot of science projects, a lot of testing, uh, a lot of throwing spaghetti at the wall to see what sticks. And I think we are seeing more and more of our customers really identifying the use cases that work for them. And uh many of our customers moving from kind of traditional AI to much more agentic workflows. And what they're finding as they as they go through that process is that for really true agency operation, you know, that that is that is outcome-based rather than than coming back and asking for for human interaction, that they they need to think very carefully about about their infrastructure in the in the broadest sense, making sure that they have the right physical infrastructure, the right network infrastructure, but also that they're they have the their data structure in the correct way, that their systems architectures are able to support AI, genetic AI workloads, that that all of the data and systems which they need in order for an agent to act autonomously are accessible via API. And so it's on it's uncovering a lot of uh traditional challenges in in large distributed global enterprises that our customers are working to fix so that they can take advantage of some of these new technologies.

Why Data Readiness Matters

SPEAKER_00 8:10

Brilliant. And you talk a lot about data readiness and why it's so crucial, not just for you know uh AI today, but for agentic services we're building a foundation for in the future. Why is that? Why is data readiness so critical?

SPEAKER_01 8:27

Yeah, agentic AI only works if it can if it can trust the data it's acting on. A smart colleague of mine likes to say that uh that good wine starts with good grapes. And uh in this context, the uh a true agentic workflow can only can only be successful if it has access to reliable data that that it can trust. You know, that the data has to be well structured, it has to be contextualized, current, it has to be available in real time. Um, because you know if uh if if every if every decision, every every step that the agent needs to take has to come back to for validation, then you know people aren't realizing the the true benefits. So it's it's fundamental that that customers go through and evaluate their data structure to make sure that it is accessible and that it is current, and so that the the tools can work um can work correctly. As with data, the same with processes. You know, we we've seen lots of customers who have have taken a flawed existing process and and tried to use uh antic workflow to just do that faster. And that that that isn't that isn't necessarily the right way to approach it. The customers that we work with who've been more successful have have taken a more holistic approach to to a blank sheet of paper to to not just do what they do today faster, but think about new ways of working with the the tools and technology available to them.

Frameworks For Scaling AI Agents

SPEAKER_00 10:00

Fantastic. So Agentic AI is on everyone's mind, uh including mine. I have five AI agents doing work for me every day. And it's really unbelievable through MANIS and OpenClaud and um Claud co-work is just unbelievable for my little business, which is me. I can't imagine as you get into an enterprise and you have hundreds of thousands of these agents. Um, what sort of frameworks are required to support that level of autonomy? It's one person, me overseeing five agents, but how do you build a framework that supports hundreds of agents with all the dynamic multi-step reasoning and unpredictable workflows that go with that? Uh, how do you think about building the foundation for an energetic architecture?

SPEAKER_01 10:51

You're you're quite right to use the word framework. Framework is absolutely the right approach. I think the mistake that we that we see is quite often is enterprises are not building a framework, they're they're building a technology, they're going and buying this thing or this service and trying to trying to make it work. The where we're seeing more success with the organizations we work with is where they have built a true framework, recognizing that this isn't a one-time deployment, it's something that's constantly evolving. You say you have you have five different agents working in your organization, and you will look you will see that it's constantly evolving. It's you know those those agents are constantly being enhanced, able to do more things, but there are also more entrants, new technologies coming available. And so you know the this this pace of change, this pace of development and this pace of investment is very different from the sort of three to five year buying cycle that many enterprises have been using for their technology. And if you if you're if you're uh tying yourself into one technology, locking yourself into one vendor or one model for three to five years, you know, it is very likely that over the course of even three to five months, you know, something else comes along that that could work in a more advantageous way. So frameworks is key. We like to work with our customers to help them build frameworks that where you know the different components of the of the architecture can be replaced, that they're not tied into single technologies, that there is there is agility in the infrastructure and the design so that as new workloads emerge, new um new models, new components, new capabilities, they are able to take advantage of them without losing the investments that they have made up till that point. It's really sort of fast, low-friction innovation is what is what we we hope to enable with with the organizations we work with.

What An AI Factory Means

SPEAKER_00 12:56

Brilliant. Um the other topic that's getting bandied about a lot is AI factories. Um have you know a mental model of what a factory is. I don't quite understand how that fits into autentic AI. Maybe give us some definition on who you know what the term and why it's so relevant as we we plan to roll out uh uh agents into real-world enterprise environments.

SPEAKER_01 13:20

Absolutely. So lots of people define AI factories in in different ways. Uh, but you critically you when you you said in your question that you have a mental model of what a what a factory is, and and an AI factory is is similar. It's uh you know you you need to have a uh a factory-style architecture that that can provide data pipelines, orchestration, you know, that can that can um a scalable way to to build, validate, deploy, and improve agents continuously. You know, that that is exact your your mental image of a of a of a production line is exactly how um enterprises should think about an AI factory. You know, you have to you have to think like a manufacturer. You don't instead of handcrafting every part, set up workflows, automate feedback and and match the right infrastructure to the task. You know, not every um not every piece of a of a workflow needs to be done on a on a local GPU. Some can be pushed out to different services, some can use third-party services. You know, that really thinking about taking advantage of technology that already exists and and linking it together rather than having to try to develop everything in-house is is a critical part of that approach.

GTT’s Cash Application Agent

SPEAKER_00 14:41

Fantastic. So GTT is uh you know a major consumer of Agentic AI. Um you have thousands of customers, thousands of sites, uh, perfect use case there. How are you using Agentic and sort of drinking your own champagne, as they say?

SPEAKER_01 14:59

Absolutely. Well, we we're building a flexible shared foundation that supports multiple teams uh without creating redundant tools or or new data silos. Um and a key part of that was restructuring our data to support the use cases that we were building. We talked earlier about how you know the importance of having data in the right structure to um in order to really take advantage of these technologies. And so you know we we have gone across all of the data struct the data that we hold in our in our organization and made sure that we we have the right architecture in place, that we have the right um that things are marked in the right way, categorized in the right way, have the right permissions to enable us to take full advantage. Uh yeah, I'm trying to think of a of a recent example. But so one one particular use case that we we uh we have deployed is uh it's it's it's quite simple. Uh but you if you've uh if you've operated in uh in an enterprise for any length of time, you'll be familiar with cash application, which is kind of a critical step in order to cash cycle where you where you match money coming into a bank account with uh with invoices. Yeah, this is this is a hugely time consuming and and manual process because um not everybody pays pays their bills neatly in uh in the correct currency in one in one payment with the right uh with the right references. Quite often there's a there's a lot of um a lot of of manual handling involved to to track um yeah, to match payments to invoices. That was huge, that was a that was a very time-consuming uh process, particularly at uh at this kind of scale that GTT operates at. But we um yeah, we built a little cash app agent that's able able to analyze remittance data and and match payments to invoices using contextual reasoning that has been enhanced with by our accounting team, by our finance team, to make sure that we're we're approaching it correctly. But it's how to it you know handles variations in in currencies and and banking rules, it it learns from collections history to to to learn when it from from decisions and matches that it's made in the past to speed up the process in future, and it uh it's able to escalate exceptions proactively to our finance team. So, but not just say here's something that I can't work out. It will the the agent will say, you know, here's a suggested resolution, this is what I think we need to do. Can you confirm? And so it's it's it's it's uh it's been a real game changer for our finance team. We're thrilled to have it so they can focus on on more impactful business initiatives.

Building GTT’s Internal AI Factory

SPEAKER_00 17:52

Brilliant. And alongside that, you've also built your own internal AI factory. Um tell us more about that and and uh how that's unfolding.

What’s Coming In Cloud Edge SASE

SPEAKER_01 18:03

Absolutely. We um we we're quite proud of the uh the framework and the architecture that we've built in in GTT. We've we've procured uh the hardware, the uh our own AI factory, so we have GPU infrastructure uh uh in in multiple locations around the world, and we're finding lots of different ways in which we can use that infrastructure both internally and externally. So we we you know a good example is our operations team using that for kind of proactive um identification of of network issues, and clearly that's a benefit for us internally, but it's also a benefit we can share with our our customers. And it's it's it's it's something that that uh will improve the experience of of being a network and security customer from GTT. And we we can share the outcomes of of the things that we have proactively detected and resolved with customers retrospectively, not to you know, here here are the here are the things that we fixed before you even knew about them, which is which is a great situation for our customers and uh and you know a really good example of a of a way that we're the industry is being improved by this technology. Um we're you'll you'll see more coming in in coming in the coming months and about some of the other technologies that we're we're using to develop internally and then ultimately make available to our customers. There's some really exciting uh some really exciting things coming through 2026.

SPEAKER_00 19:36

Yeah, tell us about some of those initiatives. I know you have so much underway with SASI, uh with your cloud and edge solutions, as well as you know, investing in people and expertise. Give us the preview, if you would.

SPEAKER_01 19:51

Absolutely. I mean, we mentioned earlier some of the work we're doing about our cloud platforms. So GTT has operated its own cloud platform for for many years. Uh, it we're we're very proud of that infrastructure. But for lots of reasons, customers are increasingly coming to us and saying they want us to do more. So organizations are moving workloads out of public cloud, or they are moving workloads around geographically for data sovereignty reasons, or lots of lots of different use cases. And they they're coming to GTC and saying that we already have compute infrastructure that we we build and manage on our customers' premises as part of the Envision Edge, in our core where we provide SD-WAN gateways and so on, and uh we also manage infrastructure in public clouds, and so our customers are saying, you can we can we do more in that world? So that's a major area of investment for us to allow customers to be able to run some of their own workloads on those Envision Edge devices, which is very exciting for the conversations we just had about AI for customers who want to do inference or on or some some very specific AI use cases on their on their locations. Other things we're doing, you mentioned that we're you know we've been a leader in in SASE and SSE for some time. We uh we're adding some new capabilities that this year with uh with uh you know expanding um our offerings with some of our existing partners. So we we already offer uh a couple of different full stack SASE offerings. We're going to be adding a another one this year. And uh lastly, you mentioned actually the the investing in our people and expertise. Something that is an important part of that GTT experience and and in vision is that it's not just the technology, it's not it's not just the digital experiences, it's it's a key part of that experience of buying of service from GTT is the people and the expertise that exists within our organization. So we we uh uh continue to invest in in training and upskilling of our existing teams. So that we we can offer that expertise to our to our customers as part of the offer. It's a key part of the Envision experience.

SPEAKER_00 22:11

Brilliant. Well, on that high note, the people, all the talk of AI and agents, it's nice to end on the people side. And speaking with thank you. It's really been an insightful, i i interesting conversation as always. You guys are killing it and um always enjoy learning from the best. Thanks for thanks for sharing. Thanks, Evan. Great, great chat. And thanks everyone for listening, watching, and sharing this episode. Take care, everyone. Bye bye.