Welcome and agenda

Tom Morgan 0:37

It's time for another ACP. It's been a busy few months. There's some news to catch you all up on, and we've got some before and after clips of Connect's new AI voice So that's coming up in a bit. So Amazon Connect Customer is AWS's cloud contact center platform. It's the software that runs the phone lines, the chats, and the emails for a company's customer service. As part of that, AWS ships small updates almost every week, so these recap episodes help pull out what really matters. But before we get into what's new if we are gonna talk through all of the latest goings on in Amazon Connect you'll be pleased to know that I'm not

Hello Alex and July recap

Tom Morgan 1:13

gonna try and do it. I've got Alex with me, who full disclosure, has been patiently explaining how contact center tech works to me for nearly three years now, I think. Alex, hello. July was busy, right? I thought tech companies were supposed to take the summer off.

Alex Baker 1:26

Yeah, it has. It's definitely been busy both from our work standpoint, but also it seems like there's been a continuous sort of flurry of of releases from AWS. So we've got lots of interesting stuff to talk about. Yeah, definitely. Maybe everyone's not going on holiday till August or something

Tom Morgan 1:42

I don't know. It'd be interesting because obviously we've got re:Invent coming up later in the year, so they can't stop. But but yeah, we've got a bunch of things to talk through. We're gonna talk about some new AI voices. We've got some more tools for team leaders, some more information there, and some data to help them. There's some new workforce planning stuff and then a new way of forcing calls through to specific agents and some auto redaction stuff to talk through. Okay.

AI voices overview

Tom Morgan 2:07

So let's talk about agentic voices a little bit. Some stuff got announced about not that long ago now. It was 20th of July, so it feels like quite a long time. It's probably about two weeks ago, right? And this is an interesting update in a way because there's no new features. It's just the existing things got better, right? We could do all this stuff before, but it was more the quality of the voices, that is the big upgrade, right?

Alex Baker 2:31

Yeah, it does seem that way, doesn't it? Which is good. And it helps to present your business better to your customers if you can just make improvements on those, make the voices sound a bit more realistic, that kind of thing. And language support as well. We're talking about that a lot at the moment, right? Companies over in the, the Nordic regions, for example, maybe traditionally have been a, a little bit further behind in the language support. This update deals with a, a whole load of additional languages as well.

Tom Morgan 3:00

Yes. So taking from the announcement on the 20th of July, 100 new voice options to pick from. And I've certainly seen that. So you've got the languages, so there's a bunch of different languages, I think 50-plus languages including some that weren't there before. But then within a language, there's multiple voices, and there isn't just one, or as was sometimes the case, one masculine, one feminine sounding voice. Now there's lots and lots of different ones to choose from. And then also there's some better, like the turn-taking, The support for interruptions is better, and less silences and less gaps as well. And then also you get a bit more control over the voices a bit. You can change the speed, the volume, the emotion and we'll go through some of that as well. So it's, yeah, it's good.

Alex Baker 3:45

The, yeah the sort of better turn-taking, that's quite an interesting one, isn't it? 'Cause I think that's a lot of what makes it seem like more of a, a realistic and natural conversation, so you're not having those slightly awkward, half a second pause before the, the AI agent responds, that kind of thing. I think probably 'cause we- we're in the, the space, we're a bit more used to that and just naturally adjust to it. Whereas I imagine if you're going in cold to it as a customer, that could be a bit more of a jarring experience.

Tom Morgan 4:12

Yeah, definitely. I don't know if this is still speech-to-text or whether this is now speech-to-speech. The announcement wasn't really clear about it. It didn't specifically say so. But looking at the documentation pages that are now available for it, they do call it speech-to-speech models. I don't know if that's the underlying sort of technical change that's happened here. It doesn't really matter to the perception or the experience and stuff. It's just interesting for us, right?

Alex Baker 4:36

Yeah, definitely. I-I'm not sure either. Yeah. And agree, I don't think it was-- I didn't see that it was made terribly clear perhaps

Tom Morgan 4:43

No. No.

Before and after demo

Tom Morgan 4:44

All right. So it's all well and good to describe these things and talk about them, but what we thought we'd do is have it so you could hear what the voices are like. So we've got some clips and for this I used Aria who is overdue a comeback, I think. If you don't know who Aria is, then go back a couple of episodes. Aria is a one of our podcast uh, co-hosts but is also a completely autonomous AI agent running inside Amazon Connect. In the same way that you'd run a contact center phone line we run Aria as a podcast host. And anyway, we've got some clips for you to listen to. All right, so here we go. So this, first of all, the first clip I'm gonna play is the old Aria. So the old voice it's in US, and this is what Amazon Connect sounded like prior to these announcements.

ARIA 5:30

Hi, I'm Aria, your AI host here on ACP, the Amazon Connect Podcast. And yes, the voice you're hearing right now is a synthetic one, which is rather fitting this week because the headline is a big leap in how Amazon Connect's AI voices actually sound. So stick around for the before and after, and let's see if you can tell how far we've come

Tom Morgan 5:51

Okay, so it's not terrible by any stretch of the imagination, but I think you can definitely tell it's an AI voice, right? There's some artifacts there and I think the rather, the way she says rather like it's... There's something about it, about the, the way it slightly grates on you a little bit, I think, and so those are the things you pick up on. All right, That's how it was, and now I'm gonna play how it is now. Again, US, tried to keep everything else the same. This is how it is now.

ARIA 6:17

Hi, I'm Aria, your AI host here on ACP, the Amazon Connect Podcast, and yes, the voice you're hearing right now is a synthetic one, which is rather fitting this week because the headline is a big leap in how Amazon Connect's AI voices actually sound. So stick around for the before and after, and let's see if you can tell how far we've come.

Tom Morgan 6:36

It's quite different, right? If you hear that rather fitting, that sounds very natural now compared to how it did before, and maybe you can still tell. I don't know what you think.

Alex Baker 6:44

It's interesting, isn't it? It's almost like how Have you seen that quiz where you can get shown a bunch of pictures and you have to choose which is AI and which is a real picture, and that sometimes there's just some r- really subtle tells, but it's hard to even pinpoint why you think it is sometimes. I, I think, yes, it w- it sounded to me quite a bit more natural. Could I describe exactly why that is? I'm not sure I could all that well

Tom Morgan 7:09

Yeah, definitely.

Accents and customization

Tom Morgan 7:10

One of the things we said as well was about different regional support for different regions, different languages, and different accents, and things like that. And so what I've also done is used the Australian, one of the Australian voices as well, so we can hear what that sounds like.

ARIA 7:24

Hi, I'm Aria, your AI host here on ACP, the Amazon Connect podcast, and yes, the voice you're hearing right now is a synthetic one, which is rather fitting this week because the headline is a big leap in how Amazon Connect's AI voices actually sound. So stick around for the before and after, and let's see if you can tell how far we've come.

Tom Morgan 7:44

I don't know enough Australians to know if that is realistic sounding or not. Sounds it, but, yeah. It's

Alex Baker 7:49

Sounds, it sounds it in a sort of might have watched "Neighbors" once when you were a kid kind of knowing different Australian dialects, doesn't it? But yeah, sounds pretty convincing to me

Tom Morgan 7:59

Absolutely. We've got a couple more voices. We're going to the UK for the next one 'cause I know, obviously we're from the UK, so maybe we can judge it slightly better. But Aria has always been US to this point, so we wanted to start there. And I'm gonna-- We've got both both male and female versions of these, so we'll do the female one.

ARIA 8:14

Hi, I'm Aria, your AI host here on ACP, the Amazon Connect Podcast. And yes, the voice you're hearing right now is a synthetic one, which is rather fitting this week because the headline is a big leap in how Amazon Connect's AI voices actually sound. So stick around for the before and after, and let's see if you can tell how far we've come.

Tom Morgan 8:35

See, I'm not sure, it's got a twang of New Zealand or Australia to it, but I really like that one for how much emotion is in the voice and how much the intonation and stuff like that. And it really, the emphasis on some of the words really does feel quite natural

Alex Baker 8:48

Yeah. I totally agree with you actually the slight twang of... I thought maybe the, the Australian was still on from the previous one initially,

Tom Morgan 8:55

It's weird, isn't it?

Alex Baker 8:56

was only brief

Tom Morgan 8:57

Yeah. Yeah. And then we can we can do a male version as well.

ARIA 9:02

Hi, I'm Aria, your AI host here on ACP, the Amazon Connect podcast, and yes, the voice you're hearing right now is a synthetic one, which is rather fitting this week because the headline is a big leap in how Amazon Connect's AI voices actually sound. So stick around for the before and after, and let's see if you can tell how far we've come.

Tom Morgan 9:22

I think the only criticism of that one is if you were having a conversation with that person, I'm not sure you could place where they came from in the UK. It's not-- It's almost not...

Alex Baker 9:30

It's pretty neutral, isn't it?

Tom Morgan 9:31

It's got no region to it at all, does it? Which is probably intentional. But yeah, it's slightly odd in that way, isn't it? That it's... Yeah. Yeah

Alex Baker 9:38

That's a-- It's an interesting point you make though, 'cause I, I remember talking to a-at least a couple of people who've asked specifically, can we get these voices in a, Northeastern accent?" Or whatever the accent is, 'cause they want it to be really specific and tailored to their customer base. I haven't seen that's available still.

Tom Morgan 9:57

No, I don't think so. No, I think w- there's more variety, say for Australian, Irish, and those kind of variants of English, but but no, the regional is not something I've seen yet. But I know you're right, yes, that has definitely been an ask. One of the other things you can do is this thing of changing the pitch slightly, changing the speed, changing the, like intonation and stuff like that. You can also insert laughter as well, so I tried that. I'm not... I don't think I'm the biggest fan of it, just to set it up. But yeah, see what you think.

ARIA 10:26

Hi, I'm Aria, your AI host here on ACP, the Amazon Connect podcast. And yes, the voice you're hearing right now is a synthetic one, which is rather fitting this week because the headline is a big leap in how Amazon Connect's

Tom Morgan 10:39

Yeah, so I don't love that one. Maybe in the right context it makes a bit more sense. But yeah, you can insert these bits of laughter as well. I haven't tried putting like laughter blocks one after the other to see what happens, but

Alex Baker 10:51

So you actually actively put a laughter block in there to do that, do you? Or d- does it handle it

Tom Morgan 10:58

Yeah, exactly. Y-it's literally square brackets laughter

Alex Baker 11:01

Yeah. Got it. Okay. 'Cause I, I was gonna ask you mentioned about the sort of, you know, intonations and stuff like that. 'Cause we had... when you look at Polly, for example, which is a sort of speech synthesi- or text-to-speech model in in AWS you can use SSML markup language. I- is it similar for these? So you use SSML markup

Tom Morgan 11:20

It's similar. Yeah. No, it is similar. They what's it called? A... They just-- They don't-- I don't think they call it anything specific. But yeah it is like that. There's not as many things to it. I haven't found a kind of reference. I've just found lots of examples. So the technical documentation lists a bunch of examples around different things you can do, but I'm not sure if it's a conclusive reference guide. They look like they might be SSML-ish. So speed I think is similar. They, they look like that kind of markup. Laughter's the only one that doesn't, which is in line with square brackets. All of the others are like XML tags for like emotion and volume and speed. So yeah.

Alex Baker 11:59

Yeah, that's cool. I was slightly wary to start with thinking that the laughter was just, it was in the power of the, the AI agent to decide when to put the laughter in, and I was thinking that might be a

Tom Morgan 12:08

terrifying. Can you imagine? Yeah, getting it wrong or putting it in the most inappropriate places and yes

Why voice quality matters

Tom Morgan 12:14

not good.

Alex Baker 12:14

All in all, though, pretty impressive, I think. I definitely felt like they were a little step towards being a bit more natural

Tom Morgan 12:21

Yeah, it's interesting, isn't it? It seems to matter way more than I thought it would, and in a way think it should. How it feel, like how it sounds, how natural it sounds. But, it does matter to people. The survey data says that like consistently, one of the complaints about, phone bots is that they feel robotic and they sound robotic. They and so taking some of that away. I think in real use though, the value will be in the, the way they handle interruptions better and don't, interrupt you, but also can be interrupted and that, that kinda turn-taking and pacing stuff. I think if that feels better, that was the thing that will make a big difference. Along with, like you say, the multilingual reach. And then I think the ability to customize a little bit. I, I don't know how many companies think about brand consistency when it comes to the voice. Maybe a little bit, I don't know. But also there, there'll be some, if it's really good it would be amazing, wouldn't it, to get to a point where it's so good that companies use it as part of their kind of sales tactic, as part of their adverts. It's like, talk to our bot because it's the most realistic sounding bot or whatever. That would be really interesting.

Alex Baker 13:28

Yeah. It does. It feels like there's still a, a way to go. Take the, the comparison of what we know looking at all this new tech and then actually as a customer calling into some customer service phone lines that have bots or chat, whatever you want to call them on them. It still feels like there's, yeah, there's a way to go in terms of adoption. I was also gonna say, though it's come on so far. It's amazing now that it was almost like in the past, if you were demoing a a chatbot, a voice bot, you almost had to set the scene as, this is maybe a little bit rubbish, but it will do of what you want it to do. Whereas now it's so much more impressive and yeah you almost... You don't have to be apologetic, do you, when you're showing it and demoing it to people?

Tom Morgan 14:13

That's true. That's true. And I wonder how people will perceive, as obviously they're gonna get better and better. There's a few things I think that e-excite me here. One of, one of them is AWS's continued investment in this stuff because, there's a lot of work that went on here for actually no new feature. It's just the existing features got better. So I think that's, that says a lot. I think AWS are fighting hard to keep up with the voice quality, if you like, of a lot of these agents and how they feel. And I imagine they're gonna keep getting better as well. We're into that uncanny valley a little bit, I think of it sounding really human. And then we're into those questions around, and we've talked about this before, regulation and disclosure and saying that you're talking to an AI agent and stuff, along with a whole bunch of concerns around frauds and deepfakes and all the rest of it. But, it's exciting, right? It's yeah. The, the voice here is definitely better than it was before.

Alex Baker 15:09

Definitely. Do you think, and I think you put this point on the notes, but the fact that the, the voices have suddenly got so much better, is that raising people's expectations then of the back end part of the system? Which this is something that we frequently say, you've gotta have all the back end and your knowledge sources and integrations absolutely nailed as well for this to be a success.

Tom Morgan 15:31

Yeah, definitely. I think it's definitely true that people fixate on the quality of the voice, but I don't think it's b- anywhere near the most important thing, actually. I, I think can you... even before the voices got better in, in this upgrade, they were still good enough to understand what was being said and to make yourself understood. Leaving aside the voice lang- the languages support stuff, 'cause that's genuinely very good and helpful, but certainly for English speaking, it was good enough before to get what you needed done. The the all the other bits were the bits that really needed the attention. I think that's definitely still true. You still need good prompting. You still need good guardrails. You still need good data. You still need to hook it up to tools in sensible ways for it to actually be really useful. The voice model's kind of the cherry on the top, but it's always good to have a bigger bit of cherry, right?

Alex Baker 16:18

Yeah,

Tom Morgan 16:18

yeah. So there they are, the new voices. Do you like them? Do you not? What do you think about them? Let us know. Interested to get your comments. We've spent so much time with them now it's starting to get hard to evaluate them properly because we hear them so often and we hear them in lots of different situations and stuff. Again, I think this stuff is changing and improving over time. There won't always be announcements for when things improve 'cause AWS like to improve things gradually iteratively. So actually give them a try, and if you find some things are not quite right, and we've definitely found a couple of things that are not quite right give that feedback back to AWS. They do wanna hear about it. They are taking that feedback, and it is improving, and those things are getting fixed as well. So that's also, for me, really good.

Supervisor coaching updates

Tom Morgan 16:59

let's talk about some other things, 'cause otherwise we're gonna spend the whole time talking about voice models. There were some other things, right? Couple of things for supervisors and kind of coaching, and these both happened in July. So the first one is there is some new metrics on the dashboards for agent queues. Is that right?

Alex Baker 17:19

Yeah. So there was, for every agent in Amazon Connect customer and it was in- introduced a while back, they had their own sort of agent queue for specific routing to the, to them as individuals. But traditionally there, there hasn't been the visibility of those that there was on the standard queues. It's another one of those kind of maybe a bit under the radar announcements that's actually pretty useful if you are relying on those though. So yeah, supervisor could be a bit blind to what's going on across agent queues. But you can now, report on those in the same way as your standard queues, which is good.

Tom Morgan 17:57

Nice. The other thing was I don't really understand this, so I'm just gonna read it out and you can help, hopefully help me through it. Automatic example evaluations for agent coaching. So when you've got a manager that's looking to coach an agent, Connect will help pull up some recent calls where the agent did it well, but also where they didn't do it well. So is this based on the information that comes out of evaluations?

Alex Baker 18:23

Yeah and to be honest, I'd-- this one had-- I'd missed this before seeing it in the notes, but it sounds really cool. I guess the, a lot of the power behind being able to automate evaluations in general is the fact that, that there are so many, may hundreds of thousands of calls or whatever it is going on in your contact center every month. It's hard for managers or for QA teams to go through and do the scoring on those, but also when we're talking more about coaching like this to find examples to back up how they're marking things on forms. And how I read this is that it will help a manager or a, a QA analyst pull together some examples when coaching an agent to say Tom you're you have de-escalation as a skill mentioned in y-your, your roadmap. This is a call that you did well at that, whereas this is a call where it wasn't so good." And being able to pull those comparisons together so that in your coaching session, it gives that, that real sort of tangible evidence and something to move towards. I think it sounds like it would be really useful.

Tom Morgan 19:24

That's cool, isn't it? Using even more of that sort of contact lens information, that real-time information and yeah, I like that. And it's also reinforcing this story that we say quite a lot, is that all of this stuff is absolutely not about completely replacing people. It's not that at all. It's about making the people better at doing the things where you need people to do them and help develop them and taking away the things that are sucking up their time but are not super high value. So I think that's nice.

Alex Baker 19:53

Helping to frame it a bit better, isn't it? I think a lot of the time, analytics tools in general in contact centers can be seen as a bit sort of Big Brother, can't they? Whereas this is, it's a bit more helpfully saying, this was an example where you did something well, so let's try and strive towards that."

Tom Morgan 20:09

Yeah. May-maybe that's the thing as well, like if you're running a contact center and you're bringing in this AI stuff because it does clearly help make it more efficient, run it better, run it at higher traffic, maybe with the same number of people because you're reallocating some self-service stuff. Maybe things like these, tools like these are the things that help win over your agents a little bit who might just feel like either they're being replaced, but also they're being yeah, surveilled a bit and monitored all the time. This can help be the thing that actually we're trying to make you better, we're trying to improve, and trying to pull out the things you're doing really well as well and recognize them. When we were doing the QM metrics for Apollo, one of the things that went in as a result of customer feedback was positive behaviors. And when we were focused quite a lot on negative behaviors, right? Because we're thinking QM-ing and coaching and evaluations, it's all about finding the things that go wrong and fixing them. Treating a contact center like a big software project where you've got to find all the bugs. But actually it was really important for them as well to call out the things that were going really well, so they could feed that back to agents to say, "Hey, well done." But also, to say to one group of agents, "Hey, there's some really good stuff going on over here with this group of people." Learn from it and, everyone can improve.

Alex Baker 21:20

Yeah. Yeah, definitely. Yeah, I think that it sounds like, sounds really powerful,

Workforce planning expansion

Alex Baker 21:24

that one

Tom Morgan 21:24

And then there's some stuff for workforce planning for tasks and emails. So Amazon Connect Customer has workforce planning built into it, workforce management stuff, that, that whole thing about predicting how many staff you need and when you need them. Up until now, that was around phone calls and chat, but it now also covers emails and tasks. So tasks are like back office work. They're not instigated necessarily as part of a, an inbound communication, right? They're things that you can set up to have your agents do other work that is not necessarily responding to customers. But that, all of that can feed into forecasting and planning. Is that right?

Alex Baker 22:04

Yeah. I think this is-- it's important for a couple of reasons. So Connect Customer is an omnichannel solution, and you'd really want-- if we're saying it's omnichannel, you wanna be able to do everything across all your channels. So it's nice now that you can include-- 'cause we, we see it with customers where email will be a much bigger channel than any of the others. But if you couldn't forecast and forecast demand and schedule resource around that, then that's a bit of a hole there, isn't it? So it's good that's been filled. Cause the maybe the back office part of the organization were underrepresented. It's good now that you can put those back office tasks or back office units of work through as tasks, and you can, yeah, start to, to represent that, that part of the organization and forecast and plan for it within the same platform.

Tom Morgan 22:55

Yeah. I think it's become more obvious, hasn't it? As the tools have got better and better, it becomes really obvious then that oh, but I need this to cover a hund- 100% of what the agent does, otherwise I can't get the full value out of it. And so I think that's really good as well. So this all just makes planning much easier for people, so it's all good. It's all

Alex Baker 23:13

Yeah. Yeah, important one this, I think

Tom Morgan 23:15

yeah,

Interrupt agent flow routing

Tom Morgan 23:16

definitely. And then a couple of sort of specific things. Again, all-- this is all still things that came out in July, which is crazy. There's an interrupt agent flow block. So this is a new building block for call flows that means a call can be pushed to a specific agent, even if that agent is already at their limit or they're marked as unavailable. And you called this one out, didn't you? As a kind of... I th- I can't remember what you said. You said it was quietly quite impactful and

Alex Baker 23:39

Yeah. It's been-- So I suppose anyone that's ever wanted to use Amazon Connect customer as more of a back office telephony system, o-one of the, the slight barriers to that is the fact that it's set up for contact centers really. So when you log in to CCP or to Agent Workspace, you have to go available to be routed contacts whi-which is perhaps a bit of a, a culture change from people that are used to, a-another UC platform, whether it's Teams or whatever. This at least means that you can kind of override that. So if somebody's not in an available state, you can use this block to route a call to them anyway. So it's, yeah, possibly quite useful for things like that, where it's if you want your agents or your users to have sort of direct dials, you could use it for that or, real time critical urgent stuff or maybe VIPs. The, your top VIP is calling about something that, that's broken and is really important, you just wanna get that, that call to somebody immediately. You can override them being in another status.

Tom Morgan 24:43

Yeah, this feels like a powerful weapon to be used sparingly because you could-- You're circumventing all of the stuff that Connect puts in place to make things run efficiently in terms of like the, the routing and the balancing and the making sure agents don't have more than their capacity and that they don't get contacted when they're marked as doing something else, and this kind of can punch through all of that. So yes. Yeah.

Alex Baker 25:07

con-consider how you're going to do it, and think about all the different edge cases. You're right, you don't wanna just... It's not one to just throw in and enable per se. It's those, it's to catch certain edge cases, I think

Tom Morgan 25:20

Yeah. I think if you do find yourself using it a lot, maybe it's a reconsidering of some of the other rules that you've got in place 'cause maybe

Screen recording redaction

Tom Morgan 25:28

they're not working for you. And then we've got something as well around rule-based redaction for agent screen recordings. Just in case you don't know, it is possible to record a, a contact center agent screen and tie that to the call as well. A lot of time this is used for kind of quality and training but often those screens can show lots of sensitive data and PII stuff. This feature actually blurs windows or browser pages based on the rules you set, which sounds like magic, so I imagine there's AI involved. Apparently it's not very hard to set up. You just essentially write rules to say, hide any pages matching this web address or hide this application window and then the masking is baked into the recording. I don't know how much flexibility you get in that. W- is it just web pages in a browser? Is it just application window names? 'Cause you can imagine with a CRM, just hiding the entire CRM actually now isn't very helpful for training, but might be the right thing to do for day-to-day operational recordings. It depends what you're trying to do with your recordings as well. If you're doing the recordings to make sure that agents aren't, like, using other applications they shouldn't be using, this might be quite useful. If you're using it for training, I think maybe it might be a bit limited. So I don't know. Yeah. What do you think? Yeah

Alex Baker 26:40

also not sure how granularly you can get with the rules as well. But I think it's another, another important one, isn't it? The other thing I was gonna mention is because screen recording is included in your Amazon Connect customer, previously known as unlimited AI pricing you probably want to be able to use it and make the most of something that you're paying for. But we have seen it with our customers where turning on screen recording just wasn't an option because it's not doing any of that PII or wasn't doing any of that PII redaction. So I think it's bringing that almost paid for part of the product making it a little bit more accessible to those that want to try and use it

Tom Morgan 27:20

Yeah, that makes sense. Yeah. Yeah, keeping security teams or legal teams happy with some rules to make it workable. That makes sense.

Connect product family rebrand

Tom Morgan 27:27

Yeah, it's worth just reminding everyone actually about what you just said, because I'm definitely guilty of interchangeably using Amazon Connect and Amazon Connect Customer throughout conversations at the moment. If you're new into the space, if you haven't been listening to these episodes, back in April of this year AWS split out Amazon Connect from being one product, which was the contact center product, into a family of four, all covered under an agentic AI banner. And there's Amazon Connect Customer, which is that kind of classic contact center product. That's what w- everything we've been talking about today relates to and what used to be called Amazon Connect. But there's also Amazon Connect Decisions, which is supply chain planning; Amazon Connect Health, which is a kind of healthcare patient engagement scenarios; and Amazon Connect Talent, which is AI-assisted hiring. That one is still in preview. But actually the whole family is being pitched by AWS and sold as an agentic AI story. So the AI doesn't just talk about stuff, it actually takes action as well, whether that's the hiring stuff, whether it's the supply chain stuff. These new voices, I imagine, go into all of it. We get exposure to them in Amazon Connect Customer the most but I think they're going into all all of the products across the board. And so that's why it's interesting. Go back a couple of episodes, listen to our interview with Hannah Bloking. And then in a couple of episodes' time we've got someone else from AWS as well who I know wants to talk through this rebrand as well and how we should think about it. Hopefully that helps as well. So yes, it's worth going back home, back over some of that stuff, which I know we've talked about before, but if you're not, listening to every word of every episode, it's it's easily missed I think.

Alex Baker 29:03

Been good to get a bit of the inside track from a few people at AWS around it a-and understand it a bit more, hasn't it?

Tom Morgan 29:09

Yeah, definitely. Yeah. 'Cause it, it was a kind of a bit of a surprise when it came, and then it was so it's good to have that yeah, like you say, the, the background to it because actually it's been worked on for quite a long time, and it's been... that's been the direction for quite a long time.

Alex Baker 29:22

Yeah

Wrap-up and next episodes

Tom Morgan 29:23

All right. It's been a busy couple of weeks, right? Busy as you can tell by the episode. There's a lot of stuff for voice, I think, in July with that the AI voices now sounding, so much better. And then a whole bunch of changes as well to plant them. The, supervisor changes, planning changes, and then routing and security and stuff as well. Yeah. I think there's probably still lots for everyone to do. The voice I think it's getting really hard now to look at all of this stuff and say, "I'll wait for the tech to be ready. It's still not ready for my contact center." I think you've got to really question that now 'cause I think it really is, it's probably not worth waiting now if you wanna maintain a competitive advantage. So I think that's interesting. And then like I said at the start, there's gonna be more stuff coming because we've got reinvent, back end of the year, late November and there's bound to be stuff that comes out of that. And then Gartner refresh their magic quadrant around September, I think. So again, they'll be looking at that and working out what features they can ship to meet those things as well. So yeah, so that's what they're doing. I've already talked a little bit about what we're doing. Got some more interviews coming up with customers, with some partners and with AWS as well, so lots and lots going on. I feel like we could chat about this stuff all day, but it is time to bring this episode to an end. Thank you very much, Alex, 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 here, we'd love it if you would rate and review us. If you have colleagues 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.