Getting Intelligent Assistance from Dan Miller of Opus
“While it’s important to acknowledge we’re still in the “early days” in the development and acceptance
of Enterprise Intelligent Assistants, the proliferation of chatbots, voicebots and virtual assistants
has already reached billions of end users. With a growing audience, the number of use cases will grow
as well, and there is no turning back.”
That’s the opening line of a recent research note on conversational AI from Opus Research. I agree there’s no turning back. I’ve been critical of these so called intelligent chatbots, but I can now say I had a delightful encounter. There’s a common perception that humans don’t want to talk to machines. That’s not true. Humans don’t want to talk to dumb, unhelpful machines. I’d take a machine any day if it’s fast and useful.
The tech has recently crossed over into the early adopter stage and proving to be useful. It’s going to get much better with specialized mini apps, vertical use cases, and ongoing technical improvements. Plus, humans are beginning to accept them.
You can access TalkingHeadz podcasts on most podcast directories and apps.
Dave Michels 0:12
welcome to Talking Heads. Today’s episode will be with Dan Miller of Opus with lots of conversation and not so much intelligence. But that’ll be the topic. But before we get to that, Evan, big news this week, or is it last week, or I don’t know what week it was, but man, it’s every week every week. Facebook had a real tough fall on Wall Street. What do you think of that?
Evan Kirstel 0:32
Well, I was thinking how much money you lost being a big Facebook shareholder? I think what was it? Yeah, I don’t know how much of their market cap evaporated, but it was pretty significant, pretty scary for their investors.
Dave Michels 0:45
You’re pretty active on Facebook, you know, I’m not active on Facebook. But it just goes to show no matter how big you are, pivots are hard.
Evan Kirstel 0:53
Well, pivots and you know, technology comes in wave and, and we’ve seen Facebook kind of reach peak user adoption and even starting to decline now. And all the cool kids are on Snapchat and tick tock. And so you know, when you hit 2 billion users, you kind of run out of people to get on your platform.
Dave Michels 1:14
I don’t see how you can say all the cool kids are on there when he knows that’s not on those apps.
Evan Kirstel 1:20
By the cool kids going to VR, though, that’s the question because I don’t see a lot of kids running around with Oculus glasses. I don’t think VR is going to be the panacea that you see soaps it is, you know,
Dave Michels 1:33
we actually have a session on Enterprise Connect, I get to be a participant on we’re gonna be talking about the metaverse and they actually allocated the full timespan, what is it 15 minutes or something for a session? I I don’t know how we’re gonna go more than 510 minutes. It’s just kind of a bunch of bunch of smoke and mirrors, in my opinion. So
Evan Kirstel 1:50
Well, bad, and even fun use cases. I’ve been known to use my VR headset. After about 20 minutes, I get the shakes and the sweat. So I’m not sure of its practical use case beyond the geeks and the nerds like us. What people
Dave Michels 2:05
don’t realize though, is how much of our day to day is already in a Metaverse, although we just didn’t call it that. But right now, you and I are interacting on a you know, we’re in different places, different rooms, different facilities, and we’re having this conversation as if we’re together. It’s kind of a Metaverse conversation. And we do this all the time. And so there’s, you know, every video conferences, kind of like that, and every email is a form of, you know, interacting in a different space. It’s just kind of evolution. And I think that that people are confused that the metaverse is going to be something radically different.
Evan Kirstel 2:38
Well, speaking of something different, we have a good chat about conversational AI, which is a bit different.
Let’s get to it. Talking Heads is a semi monthly podcast with interviews of the top movers and shakers and enterprise communications and collaboration. Your hosts are Dave Michaels and Evan Kersal, both of which offer extraordinary services including research, analysis, and social media marketing. You can find them on Twitter, LinkedIn, or at talking points.com. That’s points with a Z and Devin kersal.com. That’s KR STL.
Dave Michels 3:13
Today we have with us Dan Miller. He’s the lead analyst and founder of Opus research. Opus is based in Minnesota, but as for as long as I’ve known Dan, he was either based in San Francisco or New York, it must be a tax thing. Welcome, Dan.
Dan Miller 3:29
Thanks. Glad to be here. I’m glad to be anywhere I think the line goes.
Dave Michels 3:34
Dan and Opus are primarily focused on intelligent assistants, or conversational bots, aka conversational AI. This is the tech that’s used to power things like chatbots, or voice bots, or virtual agents and increasingly used for customer service. So let’s get to it. So Dan, I’m so excited. You’re here.
Dan Miller 3:57
Yeah. Thanks for having me. I’m looking forward to this conversation.
Evan Kirstel 4:01
Me too. I’ve been often been called Dave’s intelligent assistant, but I’m not sure if that’s the right context. I’m the only non analyst on the call so I’m gonna I’m looking for a good fight. I kind of like it when analysts fight and argue. So let’s start off with some basic stuff. And this whole conversational AI topic is wrapped up and other tech you know, you hear about speech processing and translation tech and ml and ml use and what’s going on, Dan? Well, what is conversational AI? Sure.
Dan Miller 4:34
Well, let me preface by saying I hate that term. I coined something called conversational commerce and intelligent assistants and it got sucked into a rat hole about bots. But I’m pretty specific. When somebody says conversational AI. I used to write flavors of AI that support better conversations, you know, between and among people, from people to machines, especially. So there’s a small subset of machine learning. That’s the core sort of thing that you have a lot of computer power, looking at patterns and detecting things in there. And then refining their ability to detect patterns better. One flavor of that gets into natural language processing. So that’s where a lot of attention is paid. There’s cognition, which I don’t think is a real technical term. But But this idea that a person can talk to a machine, some of this started out with just simple speech recognition is one thing. And then it gets married to what we’ve been covering forever in the contact center, his contact center automation happened, which was automated speech recognition, and text to speech rendering. So some marriage of that with a couple of other derivatives like applying pattern recognition or deep neural networking resources, to recognizing intent, and some of the language models that are either open source or developed by deep pocketed companies like Google and Facebook, to just do a better job of understanding what we say call intent out of it, apply some logic to matching intent to possible answers. That’s I guess what’s being encapsulated in conversational AI.
Dave Michels 6:22
It didn’t take long for you to mention the magic word or words which were contact center, and didn’t take you long to talk about companies like Google either. So So and now I know you don’t like conversational AI. So I’m gonna keep on using that term because technoid conversational AI must be a real thing because Gartner has come out with an MQ. And of course, Gartner is very selective about that they only do about 300 MQ is a year and then IDC came out with a marketscape on conversational AI. So this is obviously getting some sort of traction. When I look at that I was kind of set when I look at those reports, I was kind of surprised that they were their own kind of vendors, conversational AI vendors, and not as many seek as vendors or providers as I was expecting. Is conversational AI, a whole new sector, or do you anticipate that it will merge with see cats,
Dan Miller 7:10
so conversational AI, as embodied by virtual assistants, or chatbots, whatever you want to call them, will be a feature or a set of features offered by your seat has provider. And in the long run? Well, in the short run, that’s what we have. Now you can shop for them. And then in the long run, there will be some first among many, but basically, these companies are trying to respond to the organic demand from their customers to bring some natural language understanding to an automated interface. They’ll bring in, you know, Google dialogue flow, or Lex or Louis or whomever. But in the long run, they’re going to look for something a little more cohesive.
Evan Kirstel 7:52
Interesting. What surprised me about the Gartner and IDC reports was not only how many vendors there were I mean, gosh, not just dozens, but hundreds to almost 2000 vendors, but also the big tech vendors, Amazon, Google, Microsoft, we’re really not in the top positions. So why is that? I think that the
Dan Miller 8:13
deep pockets, we always call them the the behemoth of the cloud, will be
Dave Michels 8:19
Evans, not a Oh, you bet Microsoft, okay.
Dan Miller 8:23
Well, you know, Microsoft, IBM, all those guys, they are happy to support the initiatives of the contact center, the CCAFs folks with their own products, that’s you’ll see Google’s CC AI contact center, artificial intelligence, be a resource that every major cloud, every major C cast says, Hey, we’re working with Google, you know, and sometimes Genesis claims to be the first among equals five nines as it just goes down the list. So, and Google pretty much insists that they’re not going to be a CCAFs that they’ll be in several CCAFs offerings. It’s different for Amazon. But a couple of things going on there is I’m speaking specifically of the Amazon with some packaging of Lex and poly and lambda. That put some heavy lifting on the enterprise itself shoulders, but in the long run date Well, in the proven long run, what they’ve done is they’ll package attractive sort of packaging of these conversational AI resources, some grow out of speech analytics, some are just married to the fact that hey, they made the resources that power can save here power Alexa and they have the natural language understanding part down and you know, they’ll sell themselves but right now other than Alexa, they’re not packaging a bot necessarily things are market driven.
Dave Michels 9:57
Yeah. Is government ever mentioned the garden or not? He reports. It’s interesting how many vendors are but also that they’re not the reports aren’t the same vendors. And that’s because of what you’re touching on the different definitions and views. But you put out your own report, you already mentioned that 40 Page Decision Makers Guide to conversational AI, I was surprised was 40 pages down, I was thought he should have put it out as an interactive bot.
Dan Miller 10:22
Right, or at least put something in front of it that lets people query it. But I mean, to be fair, there, it’s 16 pages of sort of describing what’s going on. And then there’s a distillation of the market where we issued an RFI, and vendors tend to respond with their market where and then we would distill that into sort of an evaluation based on our criteria and what they’re providing us. But yes, we got very lengthy answers. And we have to give a lot of credit to our research director, Derek top for getting like 40 pages down to two pages for each vendor.
Evan Kirstel 10:56
How about one page total? Like, is there a cliff notes? Version? I could, I could read like, what’s the bottom line of your report, if you could share with the audience? What people gravitate
Dan Miller 11:06
towards is the graphic that isn’t King, we don’t do a magic quadrant, and we keep renaming it. But there’s leaders and I think we’re gonna say there’s the leaf. Well, every one of them is a leader. Because Evan, to your point, there are 2000 companies.
Dave Michels 11:22
It’s like preschool, everyone’s a leader. It’s all right.
Dan Miller 11:25
Now, the ones included in our in our report are leaders, when you look at the 2000 companies, you
Dave Michels 11:32
know, you had leaders and challenges, right, let’s be fair, you would broke them into two groups
Dan Miller 11:36
in this report. And we re label the challengers to noteworthy because challenging always sounds like a pat on the head. It has its own meaning. But yes, we have we have leaders and challengers and the pedigree goes back to the hooks in the contact center and customer facing resources. And the idea when we started was to look at the firm’s that compiled the language models and use cases to suit essentially customer care, we have a relatively high bar for inclusion. And we were looking at implementations and the volume of calls handled and stuff like that, which incidentally, when Gartner issued, you know, their first one this year, they had a challenge, you know, they said, Okay, there’s three different things. We look at natural language, understanding business, automation, and attention to UX. We at the time was saying, okay, sort of the natural language understanding is core, but it’s, it’s for the purpose of creating a better user experience. And as we mature it we were looking at, and then what are the hooks into the business logic. And we can, in brief say that a bunch of the solution fighters that you see out there will say, give me your chat history, we’ll put it through our natural language understanding will tell you what use cases handle like 90% of the traffic, you’ll have talking to your customers. And we can have this up and running in, you know, a week and then it’ll start learning and it’ll be actually doing stuff within three weeks. And it’s basically a glorified Question and Answer resource. I’m not saying it’s a you know, there are some that are just answer bots, just say, Hey, show me your frequently asked questions. I can automate that. So we’re farther along than that. And then this notion of how you do business automation, you know, what would move that to the core from opuses? Perspective? is, you know, where it’s going, is that, okay? There was a limited number of capabilities when this all started, and it was question answering and routing. Now, it’s evolving into, hey, we can do natural language search, or the company’s knowledge base, and we can do intelligent routing to the right resource, and all that. So that’s the challenge was a lack of understanding that you can’t look at the three things separate. You have to be kind of holistic. So Evan,
Dave Michels 14:08
there’s your Cliff’s Notes version. Of course, I think you could have read the report quicker than that response. But So Dan, let me ask you at a high level, how do these product offerings differ across all these vendors? Is it just about maturity and features? There’s more about verticals, or is it pricing models? What are the some of the big differences we’re seeing across these offerings?
Dan Miller 14:28
How come you mentioned that big three? I do think pricing is always important. So let’s lay that aside, because pricing is very flexible at this point, and pretty competitive. Maturity is a great word meaning that if we look at the companies that can come in and say we have the use cases we can show you where there are results that suit what their prospects are looking for. Then that’s it. Big plus, which gets to your third one, which is verticals, it tends to happen within verticals. So the rolling out of intelligent assist enterprise intelligent assistance has not been horizontal there. There were companies, I suppose the largest companies in particular verticals. And it usually was around financial services, we’re getting more sort of pure e commerce, the sort of things, there was healthcare. There are specialists that have those mature use cases or a list or you know, language models, whatever it takes to say, okay, I can get you up faster and more effectively, because I have experience in your vertical. And then as they succeed there, then they go to look at the next one, and the next one in the next
Evan Kirstel 15:48
one. So speaking of bots, I mean, I’ve often been accused of being a bot on social media, many times, yes, that
Dan Miller 15:57
nobody thinks she’s asleep. Yeah, it’s my bot
Evan Kirstel 16:00
for seven. But in the context of customer service, I never quite know if a chatbot mean, is implying a voice thing or chat saying and other terms that make it clear to consumers, or should people have a preference do people
Dan Miller 16:18
we have a survey out later this year, we’re going to find out more about that the terminology that people arrived at was, well, chat bots, those were techspace. And they were kind of the first to be deployed, both on messaging platforms got a lot of attention, you know, Facebook, launched sort of a platform for Apple business chat, all of those have a way to add the company’s chatbot to your, you know, list of two people on your network and that sort of thing. So those are chat, those are text based. And I think they get called Chatbot. Voice bot will learn how many of those there are actually out there. And you know, what was happening is a bunch of large banks as part of their mobile offerings, we’re adding a way to talk to and you know, the one that people cite young Capital One had no Bank of America has Erica, others have elected not to give it a name. And those can be voice or chat. So you invoke them by pressing a microphone or something. And then it may be a very short lived phenomenon. You might have heard this here first. But one of the things we’re seeing is that those conversational bots are sort of giving way to natural language search, meaning that in the spirit of having systems learn from the users not trying to teach users how to use things, we’ve kind of learned that people are getting more comfortable using natural language to get responses, either from something like the Google Search slot or the you know, something in the mobile app. And it’s only in the instances where people are calling and encountering a voice bot, that voice bots are getting used and that that seems to be in the restaurant and hospitality vertical. So you’ll see things across verticals. And I know this is too long an answer. But they’re self identifying voice bots are ones that you talk to chat bots are ones that you’re typing toward, and all of them ideally are giving the same answers wet regardless of how they’ve heard it. So you know, the challenges for companies that are employing intelligent virtual assistants is that their source of information, whether it came in through the mobile app, or through the website, or into the contact center, we’d like those all to be the same.
Dave Michels 19:03
I like that somewhere in the middle of that response. You said, I know this is too long. But I want to get back to the conversational AI versus to see cows. I still find that surprising because so much of the CCAP conversation right now. Industry conversation is about conversational AI, if there are separate sectors, which will be handling more customer calls, and doing more customer conversations over the next few years, or is it going to be the same number because every time you talk to a bot you wind up asking for
Dan Miller 19:34
I like to think that seek has versus conversational AI will handle the exact same amount of calls because the myth I’m thinking of is that conversational AI will be involved in every interaction between enterprises in their customers or prospects or partners all that and I honestly think we’re seeing that so,
Evan Kirstel 19:56
but seek as is usually price per user per month. You know, around the agent, is that in conflict with conversational AI were session based or number of calls, what do you think?
Dan Miller 20:09
I think they’re gonna find some pricing equilibrium, if you will. And I think somewhere in the world there, there’s a vision of the put $1 amount on it, you know, I mean, they used to talk about like the 116. A, well, anything from $35 to $116 a seat, depending on how many features you’re adding. But think about the $900 seat or something like I don’t know what the price is going to come out as, but I don’t think it’s difficult to map, whatever your consumption model was, you know, seats or ports or whatever, use that as a unit of measure, look at what the traffic has been with some sort of predictability based on your history, and figure out what you as a buyer willing to pay on a per seat basis to have these resources invoked. That’s kind of what I’m seeing. It isn’t exactly what’s happening. But it’s going to get there. I mean, I always think that people get, I mean, enterprises get used to pricing, you know, whatever the most advantageous pricing is that’s you know, whatever Micra
Dave Michels 21:17
but you’re saying the technology is going to different, but the price will stay the same. But But obviously, the the bots are cheaper to operate and to scale than hiring a bunch of agents. And so there’s got to be an advantage financially to using the box.
Dan Miller 21:31
Oh, correct. So, yeah, okay. Exactly. The pushback there is, it’s kind of old think, no one thought to that, once you do start thinking about seeds and productivity, and treating this as sort of an automation challenge that yes, when the bots are introduced, it’s like wage arbitrage. And they can definitely do it cheaper. And incidentally, Evan, that militates toward a proceed or, you know, per user pricing with fewer live agent, you know, there’s some scale for Okay, perhaps it’s a bot. I mean, one of the trends we’ve seen is that you hire a digital employee, and it comes in less expensive than a human and has certain tasks can do, there’s a period of training for it, all these sorts of things. And there’s been some traction for that approach by a number of the vendors and pricing can go any direction and that is actually
Evan Kirstel 22:39
hired several digital employees. The problem is he’s been fired them after the first
Dave Michels 22:43
you know, digital employees, either on or off, I’m all of mine are off. But so you mentioned you don’t like conversational AI. So another term that gets gets gets used a lot down is a conversational cloud. And I kind of avoid conversational cloud, because I don’t know what Cloud is. But let me ask you, is conversational cloud, inherently, or conversational AI, inherently a cloud service?
Dan Miller 23:06
And the answer is kinda, and I’ll try to keep this short. So the shortest of short answers is, yes. But it’s cloud in the spirit that, you know, establishing back in the 80s, and Enterprise Services bus, and then having ways to invoke services, you know, through connectors, you as a solution architect, just start not giving a damn where stuff resides. So I think what we already see companies that have quote, moved stuff to the cloud, like use Amazon Web Services, or Azure, or whomever else for certain tasks, or have, quote, moved to the cloud. And I know why you don’t like this term, there are public clouds, there are private clouds, and that sort of thing. But as a abstraction, the stuff moved to the cloud. You do a lot of the things to make them work that you would do, whether it resided on premises, you move to the cloud, you know, use Docker and you containerize things and you move them. Everybody’s sort of using the language of the cloud. And there’s no downside to pretending that, hey, it’s in my company’s cloud. That’s how it sort of playing out.
Evan Kirstel 24:22
And a lot of CCAFs companies are partnering, you know, nice with Amelia and Avaya with cogged and G. Several acquired as you know, better than anyone five, nine and inference. But do you think it’s too early or too late to acquire a conversational AI company?
Dan Miller 24:39
Wow. I’m not going to be doing and the companies that you listed are all in sort of different businesses.
Dave Michels 24:48
I’m sorry, Dan. Nice Avaya and five nine are not in Oh, I
Dan Miller 24:52
was saying that cognate G is not doing what Amelia is not doing. So In order for the C cast folks to provide enterprise Intelligent Assistance, there’s it’s not an either or, because they’ve established marketplaces where their customers can invoke whatever, whatever they want. And even though you know, saying five, nine bought inference, if somebody said, Well, inference itself would say, Hey, if you want your NLU from Google, we’re fine with that. So it’s not too late. It just may never be necessary, because
Evan Kirstel 25:34
it’s like an app store. You know, you just write, plug and play different apps. But I am surprised how big conversational AI
Dave Michels 25:40
has become very quickly to just out of nowhere. Yeah, talking heads. We’ve
Evan Kirstel 25:45
had several vendors cogged, the G grid space. You know, you’ve seen all these reports from Gartner and IDC, of course, yourself. Yeah. Specializing in it. So are we at peak conversational AI? Is this hype? Or is it going to get bigger or smaller?
Dan Miller 26:00
I think it’s still on the the early growth stage. You know, one of the reasons it got very big very fast is that a lot of the methods, a lot of the programming methods used to do the natural language understanding to do all this pattern matching and stuff are very old, and build on dnn. And now these language models that are even that rely on the speed of processing that just wasn’t available until like two years ago. So so in a funny way, the technology is out the capabilities of the technology providers is way ahead of what businesses are about to ready to absorb. So the capabilities there, there’s some recognition and there’s some successes out there, there’s probably more solution providers that are going to survive. But in terms of the outnet spending, to bring the capabilities enabled by conversational AI by the you know, deep neural networking and speedy processors, that is just starting. And there’s still some organizational barriers, the barrier is now organizational, where it gets absorbed, you know, the fact that was talking about whether it’s for C CAS or UCaaS, or whatever, it’s going to be ubiquitous resource. I mean, we’re already seeing that there’s bots sitting on conference bridges, and recording everything and doing summations and figuring out calendar. So we’re just starting to accept it. The hype curves, an interesting thought. But it’s not hype any longer. There’s, there’s a way to deliver it on it. And the barriers to acceptance are not the technology at this point.
Dave Michels 27:43
Well, hype is when we have these unrealistic expectations, I think, I think we’re beginning to learn how limited the technology is until this last time, but not because because we’ve adjusted our expectations. You say it’s we’re in early stages. But as it gets bigger, isn’t that just going to wait Google, Amazon, Microsoft and IBM and just kind of in that’s just going to be the end of the story, the end game? Maybe?
Dan Miller 28:06
I mean, we’re seeing how this whole we’re seeing how those giants, the behemoth are positioning themselves. So like I said, I don’t think Google has, Google will be like a source of of enablement. IBM is pointing sort of in that same direction. Companies don’t get very far from their founding, Amazon, Amazon will play the role of supporting just like they did in CCAP. You know, we’ll support as many people until we realize how we can do this ourselves. And we’ll do it ourselves. But when it comes to Intelligent Assistance, we’re still in the lead a million flowers bloom, we’ll hit the let 100 Flowers bloom. And then there will be three or four enterprise intelligent assistant providers that aren’t the Giants. Not that they’re small either because I mean, Verint but next it and they’re, you know, folding that into their overall enterprise infrastructure. They don’t have a C cast the way their rival nice has with CX one. And then you know, having as you mentioned, CX one took the tack of hey, yes, we have Neva, but we also will work with with Amelia for, you know, to solve some of these more complex customer facing problems and enterprise wide. So there’s no way to say well, they the answer you Yes, the big guys will prevail. They’ll always be there but they won’t crowd out, you know, three or four or five of the demonstrably working EIA providers that have a proven track track record. Customers,
Evan Kirstel 29:47
every analyst has a different opinion. You’ve clearly shared yours. It’s been great on this episode. I think Dave even has two or three opinions on any topic at any given time. But I noticed in your list and then the This report, you know, the leaders are different than the leaders gardener select and different from the leaders in IBC. Why is there so much variation among the analysts? Well,
Dan Miller 30:11
I can’t speak to the other analysts, I was trying to articulate, you know, our selection point were that they were basically into customer support. So something as broad as conversational AI, sort of cuts across a lot of things. So So we’ve evolved because we were trying to do a buyer’s guide to people who wanted to introduce what you might politely just call bots. So that that was a self defining group of people, you know, cognitive G has come up a couple times here, I sort of thought of cognitive G as centered around business process automation and adding sort of a conversational front end to that. So I didn’t see them as in, you know, direct competition with the nuances and next it and, you know, 24/7 of the world that that we’re bringing conversational entities, these enterprise intelligent assistants, you know, customer facing or ating live agents exclusively.
Dave Michels 31:13
24/7 is actually a really good example, because Opus you rated 24/7, a leader, and neither Gartner nor IDC did. Now, I know you can’t speak for Gartner, IDC. But can you speculate why there’d be such a gap and why you liked 24/7 so much?
Dan Miller 31:31
Sure. These I guess make them when we were starting with our shortlist in the in the first year, they they were one of the few that brought us actual use cases where they were being employed in customer facing situations. They had a mature set of tools for building intelligent assistance and then incorporating it into the customer care workflows. So they’ve always been included there. And I can’t speak to why they weren’t included as a leader in in the other guy.
Dave Michels 32:02
That’s easy, because they the other elements. Of course, if we were talking IDC, or Gartner I’d say, like just because I’m a But alright, well, thanks for clearing,
Evan Kirstel 32:12
I guess that’s as close as we’re gonna get to a fight. But okay, that’s. So it seems like intelligent virtual agents are, are in are all the rage IVRS are obviously out. Ah, are there any best practices emerging? Like why do you say, represent speak with a representative instead of pressing zero or one credit card? Instead of entering it with a pad like, yeah, are we on all these little nuances?
Dan Miller 32:39
Both those questions are over on the UX side. So best practices, understand whatever is something whatever somebody is doing, regardless of what they’re doing, and you should never say it’s better to say representative rather than zero out, you handle either one. So, you know, this is one of those ironic things with any really, really good UX is if you’re doing it, right, you’re not drawing attention to yourself, handle, let people use their own way of conveying their intent is probably the core along with do no evil. But then Evan, UX aside, one of the things we’re learning, well actually stick with UX, the supporting natural user interface, you know, regardless of whether it’s voice or text, or whatever, is sort of the primary rule. So that’s what I was trying to express by saying, hey, here, whatever they’re saying, over on the, Hey, Are there best practices for implementation? In these documents, we have a few sort of roadmaps for getting started. I don’t know if that helps. But you know, we usually say you know, the classic start small, pick a high impact use case, get that right. By high impact, we mean, make sure it’s visible both to your customers. So your frequent, they’re frequently using it and getting a positive result, and also conspicuous internally to your company. So either you’re, you know, proving ROI. You know, I hate the idea of like you’re capturing these things, but you’re fulfilling the objective of the caller or contactor successfully, and in doing so, it has a positive financial outcome. And then to repeat that, that whole formula, you know, do it with other use cases. But the good news is, initially, it was just a project for I know, introducing an effective intelligent assistant was for deep pocketed companies that could hire a staff that included computational linguists and UI designers and that sort of thing. And the good news now is we’ve moved from the early adopter deep pocketed project sponsored by the digital transformation group or You know, existing in a pilot over into the critical path with customers with a lot of firms. So, you know, what you’re looking for are the companies that bring solutions that are ready know what the types of questions that are going to be answered. So the machine learning part is just making sure it works in your company in your discipline has your brand names that sort of stuff and then learns from your conversations, and that there’s robust enough tools that you’re able to dip into back office systems and knowledge bases and stuff without disrupting a conversation.
Evan Kirstel 35:43
So Tommy, we’ve talked a lot about practical applications and seek as in contact center customer support, you know, very useful but kind of boring. Give us some other use cases, maybe something a little sexier, in the enterprise for conversational AI, you know, what am I gonna get my Jarvis Iron Man assistant, or my Metaverse assistant with conversational AI,
Dan Miller 36:06
our first Metaverse mentioned in W three. So here in web three world, let me generalize to say that the objective is to make people better at doing the tasks that they’re doing. So there’s two sets of answers to what you’re talking about. And one would involve smart endpoints, where the intelligent assistant and we thought of this actually, in our first report, is there’s a difference between an assistant, an advisor, and then an agent. And we’re somewhere in sort of the between an assistant and advisor level for applying, you know, what we’re calling conversational AI on this call. So an assistant is just sort of a ask me anything, and I’ll do my best to find an answer. An advisor is one that sort of preloaded with suggestions, for best, next best action, politely put, that can apply either to an agent or, you know, through a bot to the actual customer or end user, we’ve seen some dazzling things, especially on the voice side of these resources that will hear us speak in longer sentences, you know, basically saying something along the lines of I need to fly to Detroit, next Tuesday, and come back the following Friday. But basically, you use your own words, and somewhere, it’s saying, Oh, it’s Dan, he flies on United Airlines. And this is his frequent flyer number of years, his preferred means of payment and that sort of thing. And we’re moving closer, that would be moving from sort of advisor to personal agent. That’s what Opus is hoping for, were sort of somewhere on a map going in that direction, where that intelligence resides, where the where you store your payment information, and preferences, and all that sort of stuff gets gets kind of interesting. But it’s kind of a right now unobtainable point on the horizon, that I think it would behoove us all to move towards,
Dave Michels 38:19
pay down in my recent newsletter, I wrote about a great experience I had with a voice bot, I guess it was Farmers Insurance. And setting up a good experience like that is is really impressed me. And it’s presumably pretty hard to do. If a business wants to do that, you know, who do they turn to? Do they turn to their contact center provider? Or do they turn to another kind of company or CRM? Or what do they do?
Dan Miller 38:48
Hence why I say you go to your cloud provider. So the most innovative yet real applications we see are often the product of working with a business process outsourcer, who’s sort of acts like your personal shopper, among best case scenarios across a number of technologies that you’re talking about. So can’t speak to farmers. But if it followed the pattern that that we’ve witnessed, looking at other leading edge companies, you know, it starts with a vendor selection in the, you know, among that list that that’s in our report of companies that were doing enterprise intelligent assistance, and then just discovering that if you go with a single vendor, and you’re in a large organization, chances are somebody went for a vendor in the contact center, somebody else, you know, went to upgrade the IVR. Somebody else built the Chatbot for, you know, the web world and somebody built the mobile app. So we’re at a point where you need advice or some assistance from a third party. that helps you figure out how you move from having multiple initiatives like this to sort of that holistic approach, because as I recall from what you described, it probably came out of the mobile group. You initiated a phone call, or did you use the chat bot?
Dave Michels 40:18
Yeah, roadside assistance? Yeah, that was $1 800 Number sent me a link.
Dan Miller 40:24
Right. So I mean, the best advice I can give now is, chances are, you know, it started with a demo from your contact center provider, but it leans on so many other things that it’s closer you know, something that involves text and fulfilling and then following something on a map gets you to not see calf but a sea pass kind of approach. Because once you go across voice to messaging, and that sort of thing. That’s what C cast is moving toward, I guess if it’s a customer experience thing and and initiates with a phone call, you might I’m now contradicting myself, you might go to your your CCAFs provider, but chances are, it’s a multi vendor solution right now and you’re gonna end up with some kind of need for a system integrator. That’s my
Evan Kirstel 41:20
feeling well, just give them your phone number and figure it out. You can have thanks, but that was a fascinating tour de force of conversational AI. Thanks for all the insights and guidance and just ignore Dave Michaels criticism of being too verbose. I learned a lot and it was really good to chat again.
Dan Miller 41:41
I’ve been looking forward to this so thanks. Awesome.
Dave Michels 41:44
That does it thank you so much, Dan.
Evan Kirstel 41:48
All right. So we had a nice chat with Dan from Opus Saturday work with you analysts you guys like to collaborate and you’re all very chummy. It’s sort of like a club right?
Dave Michels 41:58
Well, it’s definitely a selective club but Dan Dan is more like a Woody Allen character to be enjoyed immensely He’s a very smart guy and when he kind of specialized in this area it confused me it’s like didn’t seem like that significant of an area to specialize in but boy he called it right what a great way
Evan Kirstel 42:17
to travel in packs. I mean, all the analysts go from event to event to be good but you go into any in person events over the next few months, which
Dave Michels 42:26
well the next one is going to be Enterprise Connect that’s coming up pretty soon but oh, no, no, the next one is actually actually you’re going to be there CenturyLink is having an analyst event
Evan Kirstel 42:35
which is now known as lumen and it’s going to be in the beautiful city of Santa Barbara Southern California. I
Dave Michels 42:40
can’t believe I said central my gosh, what a dinosaur
Evan Kirstel 42:43
you still have I think some backlines from Century Yeah, you have to unplug that as I move over there. We’ll talk we’ll talk there.
Dave Michels 42:51
Alright, thanks for listening Oh, man I gotta get out of the phone don’t know me. No, it’s me.
Transcribed by https://otter.ai