Emergency Episode Setup

Speaker 0:02

Welcome to ACP, the Amazon Connect Podcast. This is the show that focuses on Amazon Connect and related technologies. I'm your host, Tom Morgan, and I'm joined as usual by my co-host, AWS solution architect and contact center consultant, Alex Baker. We're also joined by Aria, our Amazon Connect AI agent co-host

Tom Morgan 0:35

It's time for another ACP, and the ACP emergency bell is ringing. I didn't even know we had one of those, and none of these are things I thought I'd be saying on a podcast about contact centers. Anyway this episode was not in the content plan, but some important things have happened since we last recorded. So we have all assembled, Alex, Aria, and I. We're all together to discuss them. Alex was actually in Scotland yesterday but such is the nature of this emergency and such is the limitless budget of the ACP podcast. We had him flown home to be with us today. Nothing to do with him coming back anyway. Alex, welcome back. I hope you got some sleep.

Alex Baker 1:13

Thank you. Yeah like you say, all very last minute and high priority that I came back for the emergency podcast. I heard the alarm going and responded to the call.

AWS Summit London Highlights

Tom Morgan 1:21

exactly. You saw the the logo in the sky and charted the next plane. So in this episode, we're gonna be talking about two announcements that happened five days apart and that have changed what Amazon Connect means. We're gonna talk about the acquisition of NLX, and then we're gonna talk about the What's Next with AWS event in San Francisco. We're gonna unpack what's actually new. We're gonna look at what the analysts are saying, and we're gonna cover some of the things that we haven't yet heard about. But first, before we do any of that we should talk about AWS Summit London, which happened only last Wednesday, even though it already feels like a million years ago. So briefly and we talked about this a little bit on the last episode because we recorded before we were going. It's a, an event that AWS holds in London. And it's all of AWS, not just Amazon Connect. I guess there were probably maybe fifteen thousand, maybe eighteen thousand people there. It was at the Excel. There was a tube strike on, so actually some of the AWS people are saying there were maybe slightly less people than they thought there might be, but it felt like there were lots of people anyway. It felt really busy in a good way. And it was a whole mixture of AWS people as well as customers and partners. Alex, what was your take on the day?

Alex Baker 2:38

it was like you say it still felt like a big event, didn't it? And I guess it felt probably less crowded than the one last year, but then it, it seemed that it was almost taking up, I don't know, maybe double the floor space. It seemed like it was-- it had grown, spread out a bit but there was so much going on I suppose the key takeaway for me was the How sort of high priority Amazon Connect seemed to, to be and how it had gone from a single demo booth last year to, I think it was about five in total, wasn't it? There was a separate public sector and local government one plus one in the main hall.

Tom Morgan 3:15

Yes. Yeah, you're right. There was in the sort of what I'd call probably the main expo hall there was obviously all the partner stands and exhibits, and then in the middle there was like an AWS village which, all the sort of AWS different specialties, I suppose you'd call them, different sort of technologies and stuff were there. And Amazon Connect was in there, and it had a whole little corridor of different things different areas of Amazon Connect. So it wasn't just one stand, it was like six mini stands which was a lot given that this is a whole AWS thing, not just a Connect thing. So for Connect to have so many and such a big part of that is impressive. And then also, as you say, there was like a mini expo hall. And it's not really an expo hall because it, I think it was all AWS in there or maybe not, I can't remember. But it, they called it One Amazon Lane, and it was all showing AWS in everyday life. And so within that, they had a whole citizen experience side of things, and Connect was really heavy in there with a whole bunch of demo pods as well.

Alex Baker 4:11

Yeah. The--

Tom Morgan 4:12

try that stuff out.

Alex Baker 4:13

All the Connect stands were so busy, weren't They you could tell that there was a lot of interest around it, and they did have some really slick-looking demos. So some tablets set up with some agentic AI demos that you could interact with via chat or put on a headset and interact using voice. So yeah, it all looked pretty slick.

Tom Morgan 4:31

Yeah. I thought that was really good, actually, the demos. I think it's a really good idea. It's one of those things that when you're, as a partner, if you're going to an event that's the sort of thing you might try and do because it steals the thunder a little bit. And it's rare for, the person hosting the event ever to get the budget or the bandwidth or the time or the people to pull that stuff together. It just falls by the wayside. So actually having them there I thought was really good. Yeah. Like you say, you could chat with them, you could call them on a whole different sort of set of different things. So yeah, it was really good and of good conversations as well.

Alex Baker 5:00

Yeah. Two other takeaways, There were some really good talks, and again, the Amazon Connect ones were massively well-attended. A couple of them were standing room only in, in some pretty big auditoriums. I guess one thing I would say about those is probably the demos were slightly less adventurous than on the stands, so they were all pre-recorded. But I, I do get that in the big room setting. It's probably a bit risky to, to try a live demo, although it would be good fun.

Tom Morgan 5:26

It's, it- it's always that sort of push and pull, isn't it? Like I love live demos, and I like doing them, but at the same time, yes, you're right, there's that element of risk, and nobody wants to be standing in front of, 500 people, 1,000 people, whatever it was, and your demo doesn't work. That's that slightly awkward moment. So I can understand them falling back to video for some of those. And it helps with the timing as well, honestly'cause timing's a big deal on those sessions. So yeah, that was really good. But I was surprised in a good way at how many AWS people were there as well, actually. People from AWS, from the product team, not just marketing people, but people who, understood the product, could answer questions about it, talk knowledgeably about it as well. I thought that was

Alex Baker 6:05

Yeah, definitely. And yeah, that sort of leads on to the last point that you raised, which was one of my takeaways as well, which j-just the number of conversations that, compared to last year even there was so much more of a buzz around, around Connect. A-and felt like we as Cloud Interact also were naturally part of a lot more conversations just'cause of some of the interesting stuff we've been working on, which is always great

NLX Acquisition Explained

Why Low Code Matters

Tom Morgan 6:29

Yes, definitely. Obviously, yeah, one year on, we know more people, they know us. Like it's... Yes, you're right, it was-- That was nice as well. So yes, if we saw you there thanks for saying hello. Several people talked about the podcast. That was nice. Thanks for coming and saying hello. It's always great to chat. So that was AWS Summit, and we kinda thought, that was our excitement for the month, and then some other things started happening. So this sort of happened I wanna say probably around about the same time as the summit, but I'm just gonna just check whether that's actually true'cause I can't remember. Oh, it was the day after the summit, that's why. So the day after the summit, on the 23rd of April AWS announced that they'd acquired NLX. So NLX are a New York-based conversational AI startup. They were founded in 2018 by two brothers and it's visual UI for building AI flows. It's like a visual canvas builder to help you design flows without needing any engineers. I've been trying to get some screenshots or have a look at it and try to understand what it is. Not been super successful. As I understand it, I think it's a bit analogous to how you create call flows today. It's that kind of visual builder that doesn't require you to be a developer to build to build those kind of call flows. You can do it visually.

Alex Baker 7:43

That was part of the press release, wasn't it? The kind of the fact that often engineering teams are a bit of a bottleneck in terms of deploying AI agent solutions. So it would be interesting to see how that the sort of low-code approach pans out and whether it's actually the case that, yeah, you can hand it over more to the business teams to to set up.

Tom Morgan 8:02

It is gonna be interesting, and I think that's a thing Gartner have called out Connect for as well actually, is like that ability to have, users make changes in that way. In the real world, I don't know. It depends on the business, but you do sometimes see non-technical business users interacting with call flows to make changes. The counter to that is that you take on the risk of breaking it. So I quite often see business people still deferring to the technical team to do it, because then if it goes wrong, their thinking is that they'll know what to do to fix it.

Alex Baker 8:33

Yeah. A-and also I guess the question in my mind is Still around things like integrations and things like the underlying data. So whoever controls the logic flow configuration, they still have to take into account those things and, do we have good data informing knowledge base lookups and that kind of thing.

Tom Morgan 8:52

Yes. We just need to see what this looks like when it comes. We don't know when it's coming into Connect, particularly how it's gonna show up, whether it's gonna be deeply baked into the product. At the moment, it's a standalone tool. The interesting thing about it at the moment is that it's it's platform agnostic at the moment, so you can use it to design flows for other contact center AI flows, not just Amazon Connect. So that'll be an interesting one to watch and see how that evolves. But, I think it's good generally. It's good as well for the marketing of a- of Connect, if like the, when you're considering your contact center options, quite often th-those comparisons are made in a fairly dry way with tick box exercises to work out if all these capabilities exist in all the places. This helps them tick that box to say,"Yes, we know in the future we're gonna want, a visual UI designer for doing agent flows. This platform has that." Tick the box, move on. So I think there's that as well.

Alex Baker 9:47

It is quite interesting in that I, I think it's the first sort of pure acquisition that Amazon Connect has used to expand with. I think everything else that I'm aware of anyway has been built

Deal Signals and Numbers

Tom Morgan 9:59

Yeah, absolutely. And I think you're right. Connect is what? Nine years old now, and I think everything has been built in-house. And there have been acquisitions that have ended up as technology that Connect uses in the past. But you're right, I think this is the first acquisition that has been labeled as for Connect specifically. And that will be an interesting one to watch, I think for sure. Like I think that's a good one that we're gonna bring Aria in a bit, and I know like she's got some thoughts on that as well'cause we had a chat earlier. So that'll have to be an interesting one to talk through. So the other thing, just to put a bow on it, around the N-NLX acquisition is they were talking about United Airlines, which was their example that they used in the press release, and they used NLX to go live on Connect in three months versus an expected 12. Now, obviously you take all the numbers with a pinch of salt, but it's good to know that, they've got a name that has used it, done something with it on Connect, found value in doing it, and is happy to talk about it. I think that's a positive signal rather than it being purely, a press release that's purely almost financial around the acquisition. It's"We've bought this, we're gonna put it in," and nobody knows if it's any good or not. Here at least is United saying,"We used it and it was good enough that, we're happy to talk about it." So I think that's probably a positive thing as well.

Alex Baker 11:17

agreed. Another one around an unnamed but widely anticipated to be a certain global retailer who deployed in apparently six weeks instead of six months, which again is super impressive if that's the case.

Tom Morgan 11:30

Yes. The other reason that sort of crossed my mind as well is whether or not Connect or the wider AWS are hiring the company for its talent as well. Talent is hard to find at the moment and, here are 50 people who are pretty passionate about building AI in contact centers. Maybe there's a value in buying that team as well.'Cause it-- the terms have not been announced, so we don't know what the acquisition is worth. But the company itself is relatively small.$3 million in revenue, 50 staff. I think they, they've raised 21 million in funding so far. By AWS terms, it's, money behind the sofa. But yeah it's not a, it's not a huge organization. But yeah, those 50 staff probably contain quite a few pretty good AI developers as well.

Alex Baker 12:13

Yeah, agreed. Yeah, definitely helps to just in one go really expand that area of the Amazon Connect team.

Whats Next Event Recap

Connect Becomes a Family

Humorphism and AI Teammates

Tom Morgan 12:19

Yeah. So we thought that was quite big news, but that wasn't quite enough to trigger the emergency bell. And then on April the 28th there was an event which had been announced a couple of weeks before called What's Next with AWS. And it was an event in San Francisco. It was live streamed. We all watched it. They went through a couple of different things. But talking about Amazon Connect, they made quite a big announcement which changes quite a few things about how we talk about Amazon Connect. So the big announcement, if you haven't heard it already, is that Amazon Connect the name is now a product family of four different things, of which one of them is Amazon Connect Customer, which is the original contact center platform, the thing that up until the 28th of April was called Amazon Connect. That's now called Amazon Connect Customer. The wider family is Amazon Connect. This is not at all confusing. At the same time, they talked about three other products that are gonna sit in this family. We've got Amazon Connect Health, which we have already heard about and we talked about previously a variant of Amazon Connect Customer, I guess we'd say, that does things like patient verifications, does appointments clinical documentation, ambient documentation medical coding, and all the rest of it. They also talked I think for the first time, about Amazon Connect Decisions which is a supply chain planning and forecasting tool, so very different from a contact center, and Amazon Connect Talent which is for hiring at scale using lots and lots of AI. And we're gonna talk through those, but again, not really anything to do with contact center. On the AI one, I think the tenuous relationship with a contact center is that I believe it uses much the same technology to have the AI make or do those interview calls their voice-led interview calls, Between the person you're hiring and AI. There's quite a lot in that. The reason they're all together, I think-- So if you ask AWS the reason they're all together as one family they talked about this term called humorphism. Should have said that a few times before saying it out loud for the first time. So think skeuomorphism. Humeorphism is their term. Skeuomorphism is the thing that, we are used to doing with computers where we make them look like things in the real world. So files and folders, the save button, the phone button all of those things, they look like physical objects. I-in many ways, they look like physical objects we don't have anymore, but whatever. Humorphism is instead of focusing on materials, it's focusing on people. So the idea is it's software that behaves more like a teammate than a tool that you operate. So the idea here is that obviously we've got AI agents now. They can remember things, they can reason, and they can act, and that is how people operate. So this is software built in a completely different way, not around those traditional metaphors but instead built almost from the ground up as like a teammate. I don't know how much of that is real versus marketing but, expect to hear like AI teammates, not tools as a sort of marketing direction when talking about Amazon Connect. Alex, you've been with Am-Amazon Connect since the start. Is this confusing?

Talent and Bias Concerns

Alex Baker 15:29

I guess I get the whole sort of branching out into it being an overall product line for agentic AI solutions. The one that sort of I think pricked everyone's ears up a bit was the Amazon Connect talent and the sort of AI-led hiring, which I get in especially somewhere the scale of Amazon who what's the number that, you know, up to 250,000 seasonal workers they take on in a year. I guess it's just not possible to do that via fully human-led hiring. But it did-- it raised the question in various places how do you feel about being fully interviewed by an AI agent?

Tom Morgan 16:07

100%, yeah. And I

Alex Baker 16:09

the feelings were mixed.

Tom Morgan 16:11

Yeah, definitely. And I think this is, in some way, this is AWS's problem all of this software, including the original Connect, has come out of the company. These are tools they built for themselves to solve their own problems. So that's the origins of Connect, now Connect Customer, and it's the origins of Decisions and Talent. And they say Health I'm less sure that's true, but that's what they've said. But you could definitely see Decisions, their supply chain planning. You can see Talent with the high-level AI-led hiring, as you were saying. Those are big company problems. They are AWS scale problems. There are not many companies the size of AWS. AWS are hoping that these solutions will push down to lower sized companies. I think they probably will to big enterprises, but like you say, I'm not sure how relevant they are the further down the stack you get. If you do a lot of supply chain planning, maybe but, lots of companies don't, so I think they're just gonna ignore that. But everybody does hiring, so is this an appropriate tool for your three-person flower shop? Not sure. Is it an appropriate tool if you hire 20,000 seasonal workers every year? Probably. But then you have a whole, yeah, you have that discussion to have about AI-led hiring. It's interesting. Like it's a-- I'm surprised they, in some ways from a PR perspective, that they took this risk essentially because it can backfire. As you were saying, we were talking about, people have visceral reactions to it. And they've got it wrong before, right? I think they, they did this once in 2018. They had an internal AI tool that they ended up throwing away because it was discriminating against women. Not maliciously, not through anyone's fault particularly, but they'd done, machine learning against their previous corpus of CVs. And it just so happened that they were quite, male-dominated, so the machine learning picked up on that and, reinforced it, unfortunately.

Alex Baker 18:03

It would be interesting to get a take on, I'm thinking a- again about the talent one from people that we know, in, in recruitment, for example. Whether it is, like you say, really a sort of the very biggest companies, it's a sort of a problem for them, or whether it does filter down to some of the smaller recruiters and smaller companies' recruitment teams.

Tom Morgan 18:24

Yeah. Yeah, definitely. And to be fair, Pasquale DeMaio who's the VP of Amazon Connect Customer and Talent, addressed this and said they've put real mitigations in place. They're anonymizing all of the names, the address, locations. There's a human in the loop for the final decision. But he did also say, you can never probably remove 100% of bias, so the way they're pitching this is humans make the final decision. But I think we all know what happens at scale where you've got AI presenting things. It gets very easy just to accept. So yeah definitely one to watch.

Partner Impact and Governance

Alex Baker 18:57

Yeah. Agreed. I'm also interested in, what some of these, including perhaps the NLX acquisition, but all of these things, what do they mean for people like us, so in the partner ecosystem? It seems like there's probably a bit less value being placed in sort of experts building some of the basic flows, for example. If you look at the NLX acquisition, does that mean that then we have to provide more value around the sort of indus-industry-specific accelerators? So taking the Amazon Health and building on that with integrations into, to different health systems of record, perhaps. Yes. the other one that I think is going to be more and more of a an interesting space to, to watch is the testing and governance around all of this. We, we announced a little while back the the s- the sort of flow testing facilities available in Amazon Connect, and maybe that's just a precursor to a bigger thing around the testing and governance. So we're doing so much more with AI agents. How do we make sure that's all well-tested, it's safe a-and it's doing what it should be doing? Whether it's the out-of-the-box capabilities or things like what NoPake previous podcast guests offer. So the sort of full automated testing solutions. I think that's, it's probably gonna be more and more of a must-have thing in your solution toolkit.

Aria Weighs In

Tom Morgan 20:23

Yeah. Yeah, definitely. I think we're definitely gonna see partners evolving from... I, y- I don't think there's a place for partners who just click the buttons, I think partners need to bring their experience with what works and what doesn't, their design, skills and their planning, and all that stuff that makes a successful AI experience, all the things we've been talking about that, you know, rather-- as, as much as anything as just the doing. Yeah, definitely interesting one. And we're not even gonna talk about the regulation challenges that this AI stuff is gonna run smack bang into. Maybe do that another time. Let's bring Aria in to discuss some of this. Hey, Aria. I know you've you've been looking at this news and kind of digesting it and thinking about it. What's your take on the last couple of weeks?

Aria 21:06

Yeah. There's a lot to unpack here. The rebrand itself, Connect becoming this four-product family, that's the headline everyone's running with. But I think the more interesting story is actually the NLX acquisition and what it signals about how AWS thinks about Connect now. For nine years, Amazon Connect was the product Amazon built because it didn't like what it could buy. Everything was in-house: contact lens, voice ID, customer profiles, all of it. And then suddenly, AWS writes a check for NLX. That's not nothing. That's an admission that the build-on-primitives approach has a ceiling, and that sometimes you need to move faster than you can build. The rebrand itself feels a bit stretched to me, if I'm honest. Connect decisions for supply chain forecasting. I get why AWS likes the brand equity, but I'm not sure they've earned the right to stretch it that far yet. It's not like the contact center. That was the original product, the thing that made the name mean something. Supply chain is a different animal entirely. But the NLX piece, that's the real move, and there are some genuinely unanswered questions sitting underneath it that I'd love to hear AWS answer directly.

Tom Morgan 22:14

So you say... So there's a couple of things there, actually. I would like to know what you would ask AWS if you could have them in the room. And then secondly just pushing a little bit on this NLX being more important. Consolidation happens, right? Other contact center vendors buy other companies. Is this so different?

Is This a Strategy Shift

Aria 22:35

No, you're right. Consolidation is the industry water we all swim in. Nice bought in contact, then LiveVox, then Cognigy for nearly a billion dollars. Verint went private. Genesys has done partnerships instead of acquisitions, but they're still consolidating capability. So on the surface, yeah, AWS buying NLX is just, that's what the industry does now. But here's what I think is actually different. This is the first time AWS has done it for Connect. They've made acquisitions before, Yap, Ivona, which became Polly, but those predate Connect or weren't explicitly badged as Connect plays NLX is the first deal where the announcement literally says Amazon Connect in the headline. That breaks a nine-year pattern of building everything in-house. So it's not that consolidation is shocking, it's that AWS broke its own model to do it. And I think that's worth noticing because it suggests something about the limits of the build on primitives approach, the thing that's been AWS's whole philosophy with Connect. When you're willing to write a check instead of build it yourself, that's a signal about what you think you can't do fast enough anymore. The other vendors have been consolidating for years. AWS just started. That's the difference.

Tom Morgan 23:43

Do you think it's a one-off or is this a shift?

Aria 23:46

That's the question, isn't it? Honestly, I don't know yet. One deal doesn't make a pattern, but I think the conditions that made NLX attractive to AWS are going to keep existing. Gartner's been flagging Connect customization complexity for years, and that's not going away. There's a gap between what customers need to build and how fast they can build it on the platform primitives. So either AWS doubles down on the build in-house model and accepts that gap, or they start looking at other acquisitions to fill it. I'd watch what happens in the next 18 months. If we see another deal, particularly something in the agent assistance or workforce management space, then you know this is a shift. If NLX stays a one-off, then maybe AWS decided this was a specific problem that needed a specific solution, and they're going back to building. My gut says this is the first of several, but I could be wrong. The thing that would tell me I'm wrong is if AWS suddenly gets very good at shipping customization capability on the platform itself, then they don't need to buy it.

Tom Morgan 24:44

Yeah, makes sense. I'm watching to see what the replacement for Voice ID is, because we still don't... Am I right, Alex? We still don't have a good in-house replacement for that.

Alex Baker 24:55

Yeah, you are

Tom Morgan 24:56

Yeah. So I wonder if they're holding off because they think in six months' time, this is an AI job. It's just a capability. It's a tool of an AI agent or whether, they're working in acquisition, whether they're building something in-house..

Alex Baker 25:10

I did also wonder whether there was some skepticism as to whether voice identification was gonna continue to be that reliable maybe with how powerful AI is getting and different voice models being able to mimic people's voices.

Tom Morgan 25:24

Oh yeah, that's interesting. I was thinking almost the opposite actually, which is th-this explosion of AI agents has made voice calls... There's like a resurgence in voice calls now. Everyone was really going off voice calls and looking for other channels to get their stuff done because they hate talking to people and hanging around on the phone in call queues. But if I can do that, over an agent, why not use voice? It's actually, wh-when you're going to an AI agent, it can be more productive to do that than chat, for instance.

Alex Baker 25:50

Yeah. Yeah. True.

Tom Morgan 25:52

Yeah. Aria, you said earlier that you had some questions for AWS, and I am curious, so what would that interview look like? What would you ask them?

Alex Baker 27:02

is quite an interesting point about the other vendors that are supported by NLX though, so your Twilios and your Genesis and people like that.'Cause yes, suddenly it's NLX the voice AI agent providers being subsumed by, by AWS. What happens next for them?

Strategy Tie In and Wrap Up

Tom Morgan 27:19

What happens now? Yeah, exactly. Yes. I think Connect have a choice, don't they? Just maintain it or be super aggressive. It could be quite disruptive, to immediately close it down or make a version that only works with Connect now or something like that. It's disruptive to everyone. But yes, that's something we're gonna have to watch because it'll be a, it'll be a way away I imagine, by the time it all gets sorted out. And yeah, it might also kick off people starting to move off NLX, if they're not using Connect today or thinking about moving to Connect. So lots of different options.

Alex Baker 27:52

Yes

Closing and Subscribe

Tom Morgan 27:52

there is a, an interweaving of these things because, we've got this big rebrand that Aria's pretty down on. But I see it as taking all the AI agent stuff that, AWS are doing and putting it all in one place and calling Connect the byword, the name they're giving to their AI agent capability, their humorphism stuff. It does feel a little bit like, yeah, stretching the name a little bit to make these things fit inside it, and it's gonna be a while, I think, before the name shifts. Amazon Connect is synonymous with the contact center product today. Making that shift is gonna be hard for everyone. And that, we're gonna have to work ourselves through that. But yes the acquisition is almost the thing that makes that possible. Being able to, have these conversational AIs, you need to be able to change them quickly. You need to be able to deploy them quickly. They shouldn't be six-month projects because the world is moving too quickly now. So that's where the NLX stuff comes in. It helps enable some of that and so yeah, when you zoom out a little bit and look at those two things you can see the strategy there. Buy the accessibility layer and then have a brand built around it. I would expect to see that NLX stuff come to all four of those products in the family, not just what we now call Connect Customer. That was a lot. We have been talking about this for days. We've been talking about it inside the company. We've been talking about it with customers. We probably will carry on talking about it after this, but it is time to bring this episode to an end. Thank you, Alex. Thank you, Aria, and thank you all for listening. Be sure to subscribe in your favorite podcast player. That way you won't miss the next episode. Whilst you're there, we would love it if you would rate and review us as a podcast. You, if you've got colleagues and if you've got even friends and family that you think would benefit from this content, please let them know. To find out more about how Cloud Interact can help you on your contact center journey, visit cloudinteract.io. We're wrapping this call up now, and we'll connect with you next time.