From IVR To AI: How Teneo 8 Scales Multilingual Contact Centers And Revenue

What if your contact center could speak 46 languages, automate millions of calls, and still hand tricky moments to the right human in seconds? We dig into the launch of Teneo 8 with CEO Per Ottosson and reveal how a hybrid AI approach—combining LLM-powered conversation with deterministic logic—delivers scale, governance, and measurable results without sacrificing warmth or control.

We walk through the architecture that makes this possible: long-form, human-style dialogue guided by LLMs, paired with rules and business processes that capture facts, validate entities, and keep flows on track. You’ll hear how a major global rollout built in English, then used language objects to expand into dozens of markets over a weekend, and why public APIs let providers embed high-scale automation under their own brands. Real outcomes anchor the story—like Medtronic’s virtual assistant in healthcare, which lifted customer satisfaction by eight percentage points and drove revenue gains while meeting strict compliance needs.

Beyond replacing legacy IVR, we explore omnichannel reality and persistent memory that unifies voice with channels like WhatsApp and iMessage. We share a practical stance on human-in-the-loop: AI first for routine steps, expert agents for edge cases, leading to happier teams and better resolutions. Looking ahead, we map the shift to voice-to-voice conversations supervised by a control layer that enforces compliance and intent understanding. If you’re evaluating Dialogflow, Genesys, or custom stacks, you’ll get a clear view of how to scale, how to avoid the “easy demo” trap, and how to measure value through CSAT and revenue—not just cost cuts.

If this resonates, follow the show, share it with your CX and engineering teams, and leave a review with one question you want us to tackle next.

Transcript

Setting The Stage And Intros

SPEAKER_00 0:01

Hey everybody. Fascinating discussion today on pushing the boundaries of conversational and hybrid AI with a true innovator in this space. Tineo. Pear, how are you?

SPEAKER_01 0:13

Excellent. Thank you for having us, Evan. I'm very happy to be here.

The Big Idea: Hybrid AI

SPEAKER_00 0:16

Well, it's good to have you back. You've been busy uh since we've last spoken, to say the least. Um before that, let's do some introductions for listeners, viewers new to Tineo.ai. Um, what's the big idea?

Teneo 8 Breakthroughs And Scale

SPEAKER_01 0:30

Well, the big idea is that the landscape is changing in two ways. One is there's LLMs now, but you can't just use LLMs. You need to use LLMs and deterministic or probabilistic and deterministic tech, which is of course what we're we've been doing. And then we've broken apart Tineo in Taneo 8 with lots of new functionality, but also public APIs. So you can essentially integrate TNIO into your contact center solution. And uh, as we've seen, the contact center providers are fighting it out right now. Everybody's AI here, AI there. And well, they all need some working AI there as well, I think. So that's uh that's a big idea we have right now to also be able to integrate Tineo into other solutions.

SPEAKER_00 1:11

And you've had a huge release, uh, Tineo 8. Maybe you can give us the snapshot of what's new, what's changed, um, maybe a little bit of what's ahead. Yeah, definitely.

Multilingual Rollouts And Global Reach

SPEAKER_01 1:22

So I mean the the biggest part of Tineo 8 is uh we are now delivering for a Mag 7. They have 10 million inbound phone calls. We also make 9.1 million of them. Uh so brilliant work from their side in the setup of this, right? We don't implement, but Teneo, of course, enables this, and that's across 46 languages. And the way that this worked was they built it all in English, and then with Taneo 8, there's a new functionality where you essentially use our um language objects, so our our uh pre pre-developed languages, which we have roughly 90 of, I think. Wow, and you use that together with LMS, and then boom, you're just deploying another language. And it takes about two weeks to really fine-tune it and get it really well done in other languages. So they they went from English to 40 odd languages over a uh a weekend. So uh very happy with that. And just imagine, I mean, if you're a US-based company and you're going into another country, well, you need to get an office, you need to get a contact center, you need to get employees, you need to train them. No, you don't anymore. Now you since we ought to make 91%, the other 9%, you could just simply route back to the US. So um, yeah, I'm really, really proud of that. There's other stuff like the AI agent builder, there's stuff like the third annotations. We don't have to go immediately to the model, which is quite expensive and also takes a lot of time, right? So that's important as well. So uh the other thing, we're blending now uh the LLM and the probabilistic and the deterministic. So think of it this way: I'm an insurance company, and you're calling in and you want to get a policy. Now, there's some facts I need to get out of you, but there's also a conversation to be had, which you usually would have with the human, which would be, you know, chatty. Well, now you can have that chatty conversation within LLM, but at the same time, Taneo running the business process and picking out the intents and the deterministic content to make sure that we understand, yeah, okay, you divorced, that doesn't really impact this. And you know, I just moved houses. Oh, that's something that impacts. We need to know that. So you can have a long-form conversation in all these different business process steps in an LLM way, which people love. And you could use voices like 11 Labs and all this new stuff that's coming out, that's really, really good. And you can get a really nice interaction with the customer. So um Teneo really, but then the big thing is the public API. I still think that that's the the biggest uh the biggest new thing for us is that you can now use Taneo in the background, right? So you you're building your application, you can build Taneo engine into it, that engine that can power millions of phone calls and automate them.

Public APIs And Contact Center Integrations

SPEAKER_00 4:01

There's so much to unpack, and we're gonna dive into that. But at a high level, how does the release move the needle for enterprises already using some form of conversational AI, maybe legacy, maybe new, maybe good, maybe bad. Um, how can they work with you?

Replacing Legacy IVR At Scale

SPEAKER_01 4:18

Well, well, most customers, as I would say all customers use something already. But surprisingly, many still have an DTMF, a IVR where you go press one, two, three, four, five, right? Which just puts the cognitive load on you. And that's that's the easier one to replace. And that's nuance, which is end of life anyway, right? So we work a lot with Genesis, it's a great contact center, really, really powerful and scalable. And with Genesis, they used to have nuance in most of their customers, either they sold it or somebody else sold it to the customer. So that's the easiest one. But then, of course, there's others you've implemented dialogue flow, you have some problems scaling that, etc. So we can sit on top of that, as we do, for example, at uh at HelloFresh, for example, where we're doing that on top of Google Dialogue Flow. So we can do that too. And we also have a very large implementation now. We're working on where Dialogflow is doing parts of the job, then we're doing the other parts of the job. So that works too. The ones we probably don't work that much with is the uh the cognitives and the course, etc. But uh, we do have a lot of integrations to Microsoft, Amazon, uh, Google, those large ones, which we always see in the enterprise space.

SPEAKER_00 5:30

We have so many amazing customers. I'm reading here on your website, one in particular, Medtronic, a super impressive company, very impressive story. Walk us through what you built with them, for them, and the kind of uh impact it had.

Working Alongside Dialogflow And Others

SPEAKER_01 5:43

Yeah, they've been building themselves, right? So Mike there is an impressive individual, has been able to build a team around him and also deliver some really, really good uh good outcomes. They have a uh Virtue Earl, so Earl is the founder of the pacemaker. So they have Virtue Earl to be their uh their uh virtual assistant across many different channels. And of course, they had ketone, touchpad, IVR, they had all the stuff, all the good stuff before. And they've been able to, on a nice uh contact center, build out uh this uh assistant across many different things. Anything from diabetes to to uh to pacemakers, you can now be serviced with a virtual agent 24-7 instead of having to wait to go to your doctor to talk about, you know, if you have a if you have a question about your device and so forth. And the most interesting thing is that they really measured impact, and of course they have cost savings, but they also have revenue impact here. So the they're seeing increased revenue. And most importantly, I think is that they measured CSAT and the customer satisfaction. And this is not transaction MPS, which we have had customers measure before, right? Immediately after the call. This is actual CSAT taking into account this new virtual assistant, and it's up eight percentage points. It's not up eight percent, it's up eight percentage points, which is quite strong for a something that actually saves you money. So yeah, they're uh they're they're a very, very, very good user of the of the platform. But they yeah, they're building it all themselves, as do most of our customers. I mean, most customers, that Mag 7 customer, they're saving upwards of more than 20 million dollars a month on this uh project. Wow. But it's really being built by three people. Uh so it's uh a very small project team that is having massive impact on the bottom line there.

Medtronic Case Study And Measured Impact

SPEAKER_00 7:28

Fantastic. And speaking of bottom lines and industries under pressure, healthcare, uh particularly here in the US, but elsewhere under a lot of pressure to deliver uh increased patient satisfaction, reduced cost, also one of the most regulated industries out there. How do you innovate responsibly in that kind of environment?

SPEAKER_01 7:49

Well, I think the the it's really you have to have that connection between the AI and the human. So as soon as the AI detects something that a human should be picking up, that's incredibly important to be able to be attuned to. So you can't have any, you really need to understand everything that the customer is calling about. Not every word, right? We're not talking transcription. You need the content of it, so the actual intent, the entities, etc., of what they're calling about. And if it is something that needs human attention, you need to have humans on the line very quickly, right? Without any without any delays.

SPEAKER_00 8:23

Yeah, that that's super important. Uh you mentioned collaboration with Genesis, but but also others. You you've got strong partnerships with with so many players in the industry, and you know, the API presents a tremendous opportunity for additional collaboration. How do they shape what's possible for customers with with Taneo Age? What does it mean for um your go-to-market?

Responsible Innovation In Regulated Industries

Partnerships And Go-To-Market Expansion

SPEAKER_01 8:48

Well, I think the uh the way Taneo works, it's it really scales very high. So that means we've been focused on the very largest customers, right? So all our customers are million plus months per phone call. So uh that that's just the nature of it because that's where we excel the most. So what we see now is we can now then put this in, and we're having discussions with contact center providers. They might have hundreds or tens of thousands of customers, but each customer just has five or six seats in the in the contact center. And this way we can now consolidate all those smaller customers with one vendor. So that vendor will take the Neo as a white label, they'll stick it in their product, and now you can give the same effect to somebody who has like five people plucking up the phone. They can now get 24-7 support, which they probably don't have, and they they can yeah, they can add all the good stuff that the large companies have, but at a much better price because you develop it once and deploy too many. So that's that's what we really see with with this uh this uh public API. Um that's that's gonna that's gonna change it. The other thing that we see now with agents is that we also use the agent builder in Taneo as an integration engine. And that's also quite important because if you are a small company today and you have five five seats picking up the phone, you're gonna have agents calling you. Yeah, AI agents calling you, right? So we now set up so that if you don't have integration to all your back ends, so let's say you work with UPS, they were in the news now, right? That you saw they they were up 12% based on a good great quarterly report. But if you're UPS and you don't have an API access to them, which you don't if you're a small company, you can have your AI agent in Tone call UPS for the customer about their package, get back to the customer and call the customer, what's at the customer, with yeah, your package is in the truck, it's gonna be delivered today. So you're gonna be able to provide the service instead of saying to the customer, hey, get in queue with UPS, you can now let the AI agent do it. And that's something I really think is gonna change a lot for smaller companies. Uh and you need an AI agent to pick up the phone too, because Evan is now gonna have five agents working for him all day. Uh so you better have somebody pick those up.

Agent Builder As Integration Engine

SPEAKER_00 11:00

What a brave new world. Really shaking up the traditional contact center industry to save at least. How is AI changing you know, the role for human agents? What does human in the loop mean to you? What does your philosophy kind of look like in practice?

The New Human Role And AI First

SPEAKER_01 11:16

Yeah, so we we we have two types of human in the loop. We have the reinforcement learning where Taneo sits on top of any LLM conversation that I spoke about, right? But then of course we also have the humans. And what you see is that this the satisfaction for those working where the agent, the AI agent, does all the, you know, finds out who are you, what's your invoices, what's your customer status, all that stuff, gets all that all of the system and then serves that to you. So those few percentage calls that actually are not automated have a much better uh say that and a much better employment satisfaction. But obviously there's fewer people working in that space, which means if you but if you look at the churn, customers have pretty much 30-35% churn in the contact center. So it just takes a year or two, and you've already reduced the contact center quite a lot. And the people are staying are the experienced ones that you can lift now to L3 and so forth. So we we we view AI first and then human. We don't view it as AI set serving the human in the contact center, right? So it's slightly different than agent assist is to us that's a bit like a hybrid uh car instead of an electric car. You're all going to go electric at one point. Why start with a lugging around a petrol engine and a battery at the same time? That makes no sense. So we don't think that that's the that's just an unnecessary step.

SPEAKER_00 12:41

Well said. And you have so many success stories, and yet in AI, in conversational AI, there are a lot of failed implementations, failed trials. Um, what do most enterprises sort of underestimate when they start their AI kind of transformation journey?

SPEAKER_01 12:59

The biggest thing I can say from my experience, and I've been in this industry now for 14 years, uh if it looks simple in an hour, it's gonna fail in a month. Wow.

SPEAKER_00 13:09

So if you if you like when it looks simple in an hour, it's gonna fail in a month. I haven't heard that one before.

Why Simple Demos Fail To Scale

SPEAKER_01 13:16

So if you if you take like all these like AI Studio in Azure or something, that's something you can open up and it's very workable or lovable, you know, all these things that look so easy. So the CFO can now build a solution and just get rid of the contact center, that's just not gonna happen, right? You need competent people, you need linguists, you need AI people and programmers. Um, but if it looks really, really simple to build, that is not a product that's gonna scale with you. And uh we see this quite a lot now where people are trying their, you know, building on I call them uh I call it lipstick on a pig. You take the LMM and you build a prompt, and then you put a graphical interface on the prompt and you sell this as a greatest solution since sliced bread. But the thing is it doesn't scale because there's so many edge cases, there's so much, so many enterprise great things. You need Hippie, you need SOC, you need ISO, you need confidential compute, i.e., encryption also and memory. I mean, it's just virtually impossible to do that with any lipstick on an LLM model. You really need a solid uh background. And it's gonna look more complicated. So in in that first instance, it's gonna look a lot more complicated than drag and drop nice bubbles and fade stuff. But it's gonna pay off in the long run.

Proving ROI Beyond Cost Savings

SPEAKER_00 14:30

Yeah, and speaking of payoff, uh, I mean, the payoffs are sort of obvious, cost savings, faster resolutions. But you know, when you have to get the CFO to sign off on an investment, how do you measure the real value of AI and CX? Uh, is it more than just the obvious savings?

Omnichannel And Persistent Memory

SPEAKER_01 14:48

Well, I think that CSAT number is incredibly important. So if you can increase the revenue 1%, that's much more impactful than reducing the cost of the contact center, right? Because the contact center is still a small percentage of your total cost. So leaving those difficult calls to humans, but having all that information flow, having persistent memory in the AI agent, etc., that just helps humans feel appreciated because they like the fact that a simple answer, like let's just say, hey, oh, I had that problem today. I was at a hotel last week and I still didn't get the invoice on my uh uh to my email. And obviously, they misspelled my name, of course, right? So I I had to call them and you know, recall them several times, very difficult to get a hold of, etc., etc. That's the type of stuff that you just want to bother to pick up the phone, I give them my mail address, I say which room it was, when I checked out, and they just send the bill, right? Very, very simple stuff. That's the type of stuff you humans want to do there. And then they're very satisfied as a customer because they get that resolution quickly. And that's that's how you increase your revenue. I mean, who are you gonna choose? Let's let's take health insurance in the in the US. There's five, six really large players. How much are they spending on Super Bowl versus getting rid of those phone queues and you know those problems that are sitting there in the queues? I think uh Super Bowl ads are great, but you know, a small project like this is gonna save you money and give you a really good CSAT. So I think it's a it's the revenue side really. And then you can't ignore the fact that if you have 10 million calls today, you're gonna have 40 million calls tomorrow. There was an article in The Economist today about how AI agents are now taking out the what they call the scam economy, i.e., meaning, you know, if you're gonna buy a car, the AI agent's gonna find the best price for you. There's no there's no room for people that that think, you know, Evan doesn't understand the car business. I'm just gonna I'm just gonna hike the price for him. There's no room for that in the AI agent world. So there's gonna be AI agents everywhere. I mean, I don't know if you've tried the latest open AI browser.

SPEAKER_00 16:49

I use it every day. I'm using it now as we speak.

SPEAKER_01 16:52

I mean, I I let it log into pages where it can't buy stuff for me, right? But stuff where I research, for example, stuff. Absolutely fantastic. It's just brilliant because it does the work for me. I just prompt it and tell it, you know, go to this site, log in, find me this.

SPEAKER_00 17:09

It's it's an amazing time to be in tech for sure. And there are also new channels emerging um uh for for service support and more, from WhatsApp to iMessage and beyond. Uh, but they're often siloed, you know. My voice conversations are completely separate from any messaging SMS conversations. How do you think about blending these together for a more unified experience? I totally I I personally hate chat, right?

Voice-To-Voice Futures And Compliance

SPEAKER_01 17:38

I if there's a chat button aside, I'll more often than not just close it down. I just I don't want to deal with chat, right? Because they're always they're always limited in their functionality. So uh I don't believe that chat is a good interface. But all the others that you mentioned, uh we obviously do chat too, but omnichannel is incredibly important. And we use now a lot uh WhatsApp um specifically for you know some parts of the US or some some uh what do you call it uh language zones of the US, like the Spanish-speaking people in the US, they like WhatsApp. Uh so we then have WhatsApp as a confirmation. So you speak, but then you can send a summary to WhatsApp, but you can also then respond back to the WhatsApp because we have the persistent memory in Taneo. So that that resides in Taneo, or if the customer chooses to reside somewhere else, but Taneo can always get to that conversation that we just had. So I think that's very important. Uh we see iMessage WhatsApp uh quite a lot. Don't see traditional SMS that much. Maybe that's also because we're mostly in the US, right? We're we're uh very, very large proportion of our revenue is in the US. And in the US, there's not that many Android phones. It's not a big uh big thing, right? People have a big problem.

Defining Hybrid AI And Market Reality

SPEAKER_00 18:55

I think RCS is opening things up as well. So interesting opportunities for more conversational dialogues. So you're leaning in so far into the future, I dare not ask you like what's next to NO9. Uh, but what's what what's on your radar in terms of where you see Toneo and conversational AI going in the next year or two?

Where To Meet The Team Next

SPEAKER_01 19:15

So the big thing that we're uh that we're just about, so let's say that that's in the roadmap for the beginning of next year, is where you no longer go through text, right? So you go voice voice. So uh essentially letting the LLM have the conversation and the text transcription and all the systems of record get updated and the APIs are text-based still, but you have a voice-voice conversation with Teneo as a sort of the parent in the room, making sure that you're saying the right thing going both ways. Uh, we believe that that's something that's gonna come at the tail end of next year. Uh, obviously the tech is there. The question is how mature are enterprises to go that route. It also has compliance issues. You know, you're not gonna have transcription, but you're gonna have you're gonna have the um so what Tenio provides is always the understanding of the call, right? So you have that in the database, so you could go back to that. But in some industries, you need the actual transcription. So let's say that you have a business transaction with a bank, you're gonna need the actual transcription for compliance purposes. Uh so I'm not sure where that's gonna hit. I don't think healthcare and banking, but I do think the telcos, which is also our biggest uh biggest group we have, software and telcos are biggest customer groups. Uh, I think they're gonna go voice-voice during next year. So I think I I see that coming quite strong. No longer going through the the text mode.

SPEAKER_00 20:36

Interesting. I think you call that kind of hybrid AI. Is that a buzzword, or what does it really mean?

SPEAKER_01 20:42

Uh we started using hybrid AI, but I've seen it pop up everywhere in all sorts of places. I actually uh I saw it on all in the podcast. Uh Shamat was talking about how you can't do probabilistic AI in the enterprise, right? So the LLM-based, transformer-based is always probabilistic. You have to have deterministic as well. And he started saying hybrid AI as well, but I don't think we invented it, but we we've been using it since uh beginning of the summer. We think that's uh that's definitely a trend, right? That people now realize that oh, those LLM probabilistic things didn't quite work out the way we want to. Um there's so many failed projects.

SPEAKER_00 21:22

I mean yes, it's fine for making fun Sora videos, but beyond that, uh mission critical, not so much. So, where can folks uh meet you, see you, uh say hello, or otherwise over the next couple months?

SPEAKER_01 21:36

Well, Genesis in Paris, the experience in Paris is in uh three weeks from now, and that's a great place to meet up with us. Uh, we were at the Genesis in Nashville at the beginning of this year. Uh always, always good to be together with Genesis. It's like I said, from a contact center perspective, there's many contact centers that do not scale to millions of phone calls. Uh, but certainly Genesis have been in that industry for a long time. So that's one of them. Uh that's probably the best place in the in the near term now to get to meet up with us.

SPEAKER_00 22:05

Fantastic. Well, congratulations on the launch and uh of Teneo 8 and uh all the success onwards and upwards, pair. Fantastic, Evan. Always good to be here. Thank you, and thanks everyone for listening, watching, uh, and sharing. And also check out our companion TV show, techimpact.tv, now in Bloomberg and Fox Business. Thanks, everyone. Thanks, Bear. Thank you, Evan.