What's Up with Tech?
Tech Transformation with Evan Kirstel: A podcast exploring the latest trends and innovations in the tech industry, and how businesses can leverage them for growth, diving into the world of B2B, discussing strategies, trends, and sharing insights from industry leaders!
With over three decades in telecom and IT, I've mastered the art of transforming social media into a dynamic platform for audience engagement, community building, and establishing thought leadership. My approach isn't about personal brand promotion but about delivering educational and informative content to cultivate a sustainable, long-term business presence. I am the leading content creator in areas like Enterprise AI, UCaaS, CPaaS, CCaaS, Cloud, Telecom, 5G and more!
What's Up with Tech?
Build Fast, Govern Faster: The Real Path To Agentic Success
Interested in being a guest? Email us at admin@evankirstel.com
What if the secret to scaling AI agents has nothing to do with better prompts and everything to do with smarter integration? We sit down with Rich Waldron Co-founder & CEO of Tray dot ai to unpack how an integration-first architecture turns agent hype into measurable business outcomes. From cloud-native orchestration to resilient API handling, Rich explains why the hardest part of agents isn’t connectivity—it’s everything behind it: concurrency, retries, governance, logging, and security.
We dig into the real reasons enterprise pilots stall and the pattern he sees among programs that succeed: IT-led, department-partnered builds with clear ROI, tight scopes, and fast iteration cycles. Rich shares a standout customer story migrating hundreds of integrations off a legacy vendor while launching agentic workflows on a single governed platform—delivering 60% lower integration costs and three times faster builds. Along the way, we explore how engineering teams use AI coding tools to prototype faster and offload grunt work, freeing time for architecture and testing without threatening roles.
If you’re choosing where to start, accelerators for ITSM, HR, support, and knowledge make time-to-value tangible and create a structure teams can adapt to proprietary data and processes. We also cover the growing priority of enterprise governance: controlling which tools agents can access, how data moves, and how identities map across systems. Rich outlines Tray’s Agent Gateway for MCP—adding authentication and permissioning and exposing Trey-built tools to other services in a controlled way—so CIOs can move quickly without giving away the keys to the kingdom.
Ready to move from pilot to production with confidence? Follow, share, and leave a review to tell us where your organization is on the journey—and what’s blocking your next agent from going live.
More at https://linktr.ee/EvanKirstel
Hey everybody. Fascinating topic today as we talk about why integration first architecture is becoming the most flexible way to turn AI agents into real business performance with a real innovator in the space at Trey.ai. Rich, how are you?
SPEAKER_01:I'm doing great, thank you. How are you, Evan?
SPEAKER_00:I'm doing well. Thanks so much for joining. Super excited for this conversation. We cut through some of the hype and talk about real enterprise opportunities. Before that, maybe introduce yourself and the journey to Trey.ai. What was the big idea when you started?
SPEAKER_01:Sure thing. Yeah, I'm Rich Waldron, one of the co-founders of uh Trey. The big idea or how we started is sort of the way I think most uh founders uh operate, in that um we were extremely frustrated with the challenges that came with um automating and working with lots of different software solutions. So a kind of age-old problem that has seen all the various stages of computing to date. But our approach was how do we turn smart business people within organizations into capable programmers? How do you sort of take the paradigms of programming and make them more broadly available to a technical audience within a company? And so that that journey started 10 years ago, and we've been able to build up this huge kind of orchestration engine that allows people to interconnect lots of different pieces of software, databases, data services in a really kind of seamless uh visual way. And that's really led itself really nicely to some of the challenges that have emerged in successfully deploying agentic workflows and agents.
SPEAKER_00:Fantastic. We'll dive into that for sure. And how does your approach differ from you know traditional middleware environments that are out there today, been around for a long time, and um people are building on?
SPEAKER_01:Yeah, so I mean we we started out, frankly, trying to use those solutions. And what we found at the time is that um, in most cases, they were built for the sort of on-prem to cloud era. So they weren't as as comfortable with cloud workflows or workloads, they weren't as comfortable with um uh you know more native cloud service APIs. And we found that the effort that was required to build something successfully was not far off what it took just to go and you know build and deploy it yourself anyway. And so for us, it was all around um how do you uh sort of abstract away all the hard parts of integration, which is not really the connectivity itself, but everything that happens behind the scenes. How do you deal with concurrency and high throughput and um APIs failing and everything that builds kind of resilience behind these things that you create? And at the same time, how do you take away the barriers in what happens if a connector doesn't exist? What happens if that service isn't supported natively? How do you make it so that people are able to build capably on top of that? And so the sort of combination of the two has meant that everything occurs within a single platform on tray, you're not managing logs somewhere else and building in a separate um uh area. It means that you're you're sort of much faster in the in the work that you create and you're not thinking so much about um how to handle the scale on the on the back end.
SPEAKER_00:Fantastic. And we've seen lots of news, media reports, and studies, the MIT study of uh a few weeks back, that many enterprises are struggling to turn AI pilots into real scalable uh you know services that can provide business impact. Why is that? What are some of the top reasons that you see?
SPEAKER_01:Yeah, I think it you know, my view was the the um the headlines perhaps don't tell the full story. So I've I'm fortunate enough to you know be on the front line. We uh Trey has thousands of customers, they're they're medium to large enterprises, many of the world's you know best known brands. And the the trend that I typically see is um the the um uh agents or the agentic implementations that are owned by IT, that are built in collaboration with the departments, that are well thought out and scoped and the ROI is is is kind of measurable and clear, they're being super successful and they're iterating quickly. Uh, they mimic in many ways how historically integration projects have been rolled out. Where I think the pain point is is this expectation that frontline employees are gonna be amazing at writing prompts. And if we build these agents that have great sort of capability, uh our employees are gonna be able to know exactly how to uh instruct them or work best with them. And then secondarily, where we're trying to put the creation capability into their hands, it feels a bit early to me. I think that a lot of those, you know, people are building agents in areas that perhaps don't really need them. And so I think it's a bit of a difference between are we thinking strategically about the problem we're trying to solve, and then are we deciding whether a deterministic or non-deterministic solution is the right way to go, versus the blanket, well, people are trying to do things with some of the uh major LLM providers and aren't getting a great deal of success, which was my interpretation of that reporting.
SPEAKER_00:Yeah, what really great insight there. I'm looking at your website. You mentioned thousands of customers. I see one uh you talk about X, um, well-known brand. How was that such an interesting uh case study to focus on?
SPEAKER_01:Yeah, YXT was a great case study because uh they were migrating off a legacy vendor. And so their challenge was um they had hundreds of integrations built out. And now for a publicly listed company of a of a good size, do you think that building the next set of automations on a platform that was built for a bygone era is the is the right place to go? Secondarily, they recognize that a lot of these processes are going to change, they are gonna infuse AI, they are gonna be agentic. So, how do we kind of blend the two in a in a single in a single place? And so their their journey with Trey was part um migrating everything over, and they're now in this position where they're launching agents internally, they're building them out on Trey, and they have this single kind of governance pane as a company whereby they're able to control which services these agents connect into, they're able to get additional depth because we connect to so many different solutions, and they're able to very quickly turn this around so that their employees can change the way that they interact with the agents that are built out. So that that's been a compelling case study because they've had great success with the agents they've built, but equally they've been able to modernize the uh integrations that they were previously running.
SPEAKER_00:Wow, pretty cool. And you even have some results here. Uh, 60% lower integration costs, three times faster builds. That's uh that's a pretty uh fascinating speed and efficiency gain there. Um is that what is that the kind of impact you're seeing across many different industries, customers?
SPEAKER_01:Yeah, I think the speed part is critical because um you know speed obviously matters from a cost component, but actually where it's most impactful is speed of iterating. You know, not many people build the perfect solution the first time around. And actually, what's been most impressive with some of the larger companies is how agile they've got in trying something out, um, uh taking the feedback in really quickly and then and then you know switching these things on. I think the the days of we're gonna build out a solution and it's gonna take 12 months are are behind us from a from an agenda standpoint. It it has to be quick deployment, fast iteration, figure out where the where the value lies and and kind of move toward that. So I think the the speed component kind of does double duty from that from that perspective.
SPEAKER_00:Fantastic. And there's there's a lot to talk about the effect of uh these tools and your tools, others on automating developer workflows, automating engineering teams' work and testing. Uh some are concerned what what that impact is on their jobs and careers. But it it seems that uh there is an upside as well in terms of freeing up engineers' time to work on more valuable work and interesting projects. What are you seeing there at in terms of how it affects engineering teams?
SPEAKER_01:Yeah, I think for us, um uh as much as we sort of provide uh uh AI and agentic solutions, we're also a consumer. And uh our engineering teams are using some of the the well-known um uh coding uh AI tools. And I think the value there is is pretty immense in that when you're um when you're tightly managing the work that's being done, when you when you you got to go through a period of experimentation to figure out how you extract the most value. Uh the coding tools are fantastic at prototyping, they're fantastic at um carrying out a lot of the kind of grunt work that historically um you know agents would uh sorry, uh uh engineers would have to do um uh manually. And I think what we found is uh where the uh where our engineers have got benefits is it it's kind of allowed them to move faster. So I don't think they're necessarily necessarily seeing it as a kind of a threat to their to their role or a threat to the jobs per se. It's it's sort of uh it's a productivity gain. And it's not to me, it's not dissimilar to you know, I I've grown up as a as a not great engineer being used to going and checking Stack Overflow for how do I do this? Somebody else has probably done it. There's got to be a way that I can get a get a gain here or move faster. And I think AI is kind of bringing that right to the front line of the creation piece and and and actually carrying out some of that, some of that work for people. So yeah, for me it's very much about an accelerant, and and we've been really impressed with some of the results that we've had in internally.
SPEAKER_00:Fantastic. And speaking of accelerant, do you have agent accelerators that are sort of pre-built uh agents for real-world use cases? I see you've chosen things like a support agent, knowledge agent, ITSM agent, that's a great one, HR agent. Is this the sort of low-hanging fruit that every enterprise can take advantage of today?
SPEAKER_01:Yeah, that's exactly it. The the accelerators and and the idea behind them is these are these are the starting points. And you know, if you think the the sort of thing that differentiates companies is is the processes they have, the the data and where it's structured, and ultimately the sort of proprietary tooling that exists within an organization. So if you're gonna get started on this journey, you know, you probably already have some sort of um uh IT support solution in place. So starting with an ITSM agent, um, working with some of the tooling that you already have, getting something um uh sort of into production that starts to demonstrate how it can work, sets you on your path to then go and say, okay, how can we knit this deeper into the way that we already operate? How can we reach further into some of the platforms that we've already implemented? And it it's it's a nice way to get going because it kind of gives you a framework and it gives you a structure to begin with. And so the accelerators for us have been a really nice way to work with our customers and sort of get them on boarded, get them moving in the right direction. What we then see is they kind of take the ball and run with it and they start being able to sort of um fashion these around some of the challenges they have. And that and that's really where I think the value starts to accelerate.
SPEAKER_00:Fantastic. So as you get into large enterprise, of course, you know, tends to be dominated by big tech companies of of various sorts. And uh, you know, if you have a seat at the table with the CIO, you're you're probably uh you know, a company like an IBM or a NetApp or Cisco or others. And I see you have lots of integrations with with these kind of companies out there. Is that your go-to-market? So being sort of agnostic and uh partnering with everyone and anyone in the enterprise.
SPEAKER_01:And the the go-to-market for us is, you know, for me, um the the connectivity piece is table stakes. Like you have to provide great connectivity, you have to be kind of agnostic to the ecosystem. You you you've got to have breadth. But but to me, the value is that you can kind of work across these different vendors from a governance perspective. So it means that as a CIO or as an as an IT department, we're in control of where our data goes, which vendors we operate with, which actions we want to take, and which downstream vendors. And when you add that agent layer, you're then dealing with how do I make sure that people aren't throwing proprietary data into an LLM? How do I make sure that we're not exposing uh uh systems to people that we shouldn't? And so sort of being that pane of glass that sits between these different systems, or rather that that that glue that exists in the middle is is really the role that we play. And so from a go-to-market perspective, it's about how do we help you speed up your implement implementation? How do we give you the governance and control that you need? Uh but ultimately it's it's enabling these teams to be able to build faster and and be successful.
SPEAKER_00:Fantastic. And speaking of those large enterprises, I mean, many of them already have some kind of integration layer. You know, big companies they've worked with for sometimes decades. They've been around for a while. How do you start with create thatai and and you know embark on this journey when there's so much legacy out there?
SPEAKER_01:Well, that's kind of the beauty of it. I think you find that most uh enterprises have multiple vendors and and they're serving you know different problems. And I think it just because you are you have used a vendor historically doesn't mean that that's the right vendor to use for the next thing that you need to go and build out. So quite often we'll see a legacy vendor may stay contained with a with a um a certain set of uh automations or integrations, and that organization will say, hey, actually, we're trying to do something that's a little bit more cutting edge. We're trying to build an automation that is going to be able to go beyond the pale of what we're used to. And that's where you know Trey steps in and they're able to start getting something stood up with our team very quickly. We kind of go through a joint build process, and and that ultimately means that they can get demonstratable value fast, but it it doesn't matter that necessarily there's a there's a pre-existing vendor in the in the building. I think um, yeah, for for companies of the size that we're talking about, that that's a that's a pretty normal thing to expect.
SPEAKER_00:Brilliant. So almost at 2026, hard to believe, but uh, what are you excited about as we head into the new year? Uh, can you give us a peek into the the go to market or maybe your roadmap? What are you uh uh anticipating for next year?
SPEAKER_01:I think the the hot topic right now is is uh orchestration and and governance for companies. I think the first wave was we have to do something. We have to we have to experiment, we have to try different vendors, we've got to get things stood up. And I think we're now kind of moving into that stage of okay, we've we've sort of started to figure out what has worked and what hasn't. We now need to put some better checks and balances around what we roll up into production for for um uh the companies or the the employees that we that we work with. And so for us, it's uh it's a lot about how do we release capabilities that help companies go and do that. So just recently we launched our Agent Gateway that allows um uh companies to provide a um uh uh authentication and permissioning layer to MCP. It also allows uh companies to expose tools created on Trey via MCP to other services, and it does it all in a governed manner. And we know that that's a problem that a lot of companies are wrestling with right now. So the the sort of roadmap for us and the future releases are all about that drumbeat of how do we continue to allow you to take advantage of all this amazing technology, but harness it in a way where you're not kind of giving away the keys to your kingdom.
SPEAKER_00:Brilliant. Well, thanks so much for the update on the vision and strategy. Really uh looks tremendous, and uh good luck for the new year. Thank you very much. And thank you. Thanks everyone for listening, watching, sharing the episode. Also check out our TV show, techimpact.tv, now in Bloomberg and Fox Business. Thanks, Rich. Thanks, everyone. Thank you. Take care.