Inside UJET's CX AI Evolution: Spiral, BYO-AI, and the Dawn of Agentic Orchestration

Live from the floor of CCW Las Vegas, the mood has noticeably shifted. The industry is exhausted by the endless promises of AI replacing human agents and the daunting prospect of ripping out legacy systems. Leaders are facing severe system fatigue, and the mandate from the boardroom is clear: show me the ROI, and show it to me now.  

To make sense of this tension, I sat down with Matt Clare, VP of Product Marketing at UJET, for a candid conversation about where contact center technology is actually heading.  

In this exclusive interview, Matt reveals UJET's strategic pivot from a traditional CCaaS provider to a CX AI company. We explore the myth that AI is here to replace human connection—because at the end of the day, people still buy from people. Instead, the real opportunity lies in eliminating unnecessary, frustrating interactions and using AI to augment the high-value conversations that drive revenue and loyalty. 

To help leaders navigate this shift, Matt and I unpacked the specific technologies UJET is deploying to accelerate CX transformation without the traditional operational chaos. Here is a look at the core innovations we discussed: 

Maximizing Existing Investments with CX AI Accelerators 

The days of ripping out your entire legacy infrastructure are fading. Matt explains how UJET is now leading with CX AI Accelerators—intelligent layers like Virtual Agents, Agent Assist, Screenshare/Cobrowse , and Conversational Analytics that plug directly into your current setup. The goal is rapid transformation, allowing brands to maximize their existing tech investments and see time-to-value in days or weeks, not years. 

Zero-Friction Analytics via Spiral by UJET 

A massive part of this rapid ROI comes from Spiral by UJET, their conversational analytics platform that turns unstructured data into actionable intelligence. Matt breaks down how Spiral is delivering what he calls "zero-friction analytics": 

• Interactive Dashboards: Instead of waiting weeks for data scientists to build complex reports, anyone in the business can use natural language prompts. You can literally ask your data, "Why did CSAT drop yesterday?" and Spiral instantly generates the dashboard and the answer. 
• Automated QM Scorecarding: Quality management at scale is finally here. Leaders can now upload their ideal scorecard and ask the AI to automatically grade interactions, build reports, and generate training modules, turning raw insights into immediate action. 

Developer-First Flexibility and Reducing Agent Load 

We also discuss the growing demand for flexibility and the rejection of vendor lock-in. Matt shares insights on UJET's BYO-AI (Bring Your Own AI) integration, highlighting their inaugural partnership with conversational AI platform Parloa as proof of their commitment to customer choice.  

Coupled with a Headless SDK—which lets brands build custom, native service experiences into their own apps while UJET powers the back-end—and a centralized Agent Assist Hub, UJET is drastically reducing the cognitive load on frontline agents. By aggregating knowledge bases and next-best-action recommendations, agents no longer have to "swivel-chair" between five or more different applications to solve a single customer issue. 

What's Next: Agentic Experience Orchestration (AXO) 

Finally, we look ahead to September's highly anticipated launch of Agentic Experience Orchestration (AXO). AXO acts as a persistent AI layer across the entire CX toolchain, using AI agents that can navigate software like a human to automate complex, multi-step workflows across disparate legacy systems. Matt explains how UJET is currently running its own internal support operations on AXO and co-developing with several of UJET’s leading customers to refine the product before General Availability—preparing to offer a solution that promises true ROI through system elimination, scaled operations and revenues, and continual improvement. 

If you are trying to figure out how to layer AI into your operations without a massive disruption, this is a conversation you cannot afford to miss. 

Watch the full video interview to discover how to accelerate your CX transformation today. 

Transcript

SPEAKER_01 0:17

Is this the pressure to repas the AI and actually extract animal value to a contact checker investment that you already have? So if you like having just into the key or exploring how to get AI in your operation without a convert to approach, but what's like that quite easy on product marketing as you said, well that's really uh see there, I'll take the best. Uh really funny. I mean understand both that on. I think it's a lot of experience. I mean, you try to do something other pandemic for all of the we are gonna exactly have it. If you have lots of conversations with really the new lines around the world, you have that object really nice carefully, the more we got it, the more of the background, because you can see we can make a thing company because we have a right complaint. Well, I think the other thing is locally right now we want to say, I think the world really realized that the all the areas and all the version of place and that's really adding the goal of uh you know replacing cost and reducing the cost of the product center. We try to realize that one, we also realize that because it can contain a lot of our calls, then it's actually all supporting the human and preferred to before you have a channel scratch on because people from people, people like to people, um, some competition required of them for the PC, but I'm not gonna question whether or not AI is done, but I just don't highlight all. But as far as I'm concerned, a lot of people are still preferring documents as an line at the airport a couple weeks ago, and everybody in a row had a five detail, everybody, thinking, even I decided a lot of them as AI just felt that they're simple uh to talk in between them at that point. So I think broadly speaking of there, it's really thinking, um, I do have two activities to how you're trending and keeping it from the list where the block press. And rather than each AI as a triple to the flash computer anymore, use AI for simple outload strings that they can find out, doing more mode and keep more virtualized as well. Yeah, I think that keeping it for is all you to a tool like you can AI, training AI to other automatic or whatever has to have to perform the capital stuff and kind of. But I think that we automatically the big common screen that they have in context and kind of modern is a very necessary. Because if you're trying to have a topic called that office, if you have a customer calling, because a full web experiment or a full access experience, and you can just in the first place, use your AI more than rather than trying to control the interact well together, and it's very hands and highest value graphs to look at more information.

unknown 4:05

I think that you should have a lot of things we have.

SPEAKER_01 4:11

So I can actually bring it in active, it's not I'm thinking to kind of explore that. So when you begin looking at proofs like the head of FPA, PY, code, AY, uh, we've partnered by Hubba. So what's great is our stream customers that give out a bunch of things. Yeah, I thought we were really quite around, we have a PK a lot of customers to really control the finding of the model and try to get into website using that kind of out of mock UI and blind cover. We're really good customers to do our basic power like data experiences on the web mobile app experience that we have today. On the virtual AI side, we're obviously a very interesting part of the Google, the only celebrity, uh, but we do like that as time goes on, more and more customers will have to draw it. I think other times within a same separate organization that can contain idea, decline, one for different cases of this organization. So we did want to bring your AI framework or different partners that I can have to try out. I will continue to go on there going forward, so that's a team and parameters topic is that they won't contact the logic year and if they are. So I think I think that's in what is more important. Yeah, I think what um an interacting point is I don't have a idea, but you can think of that cost, and why not just kind of listen to something wrong with the new approach and do anything on it and people can just everything very much about that one and find that money for the world? Okay, so what did you often think about the next year? Yeah, so eventually you have the FAI platform. And rather than thinking that AI can look a lot here, we're gonna try out one group in your work collection, a point group number, and one booth for your other, these developers either copy in the platform, and really can contact virtualization keeper browser in your own hardware using a structure. We need the AI, control AI to what a virtual category is just type functionality, but I just can just change AI that's hard in virtualization. So it's not like you're getting a dedicated encrowed type structure. Um I think the biggest opportunity here is really hard via optimized performance. If you're telling you how virtual agent performance, it can probably type a limit with unstable.

unknown 7:37

But that's something you think it has to be. So if we closes it out by the encryption engine, stuff that you have to say, or connect that into Admin, or future agent changes, virtual agent operation, stuff you're putting.

SPEAKER_01 7:56

Uh, they can run the same text in every application, we can make two through the process, and then all right, we'll do my easy, but if you're on a dumbass, or if you have a playbook to the play, then you can edit and do to get there, and we've got the other. All of the icon, but it's absolutely about what I did, you know, after the limit.