Video Interview with Al Cook, VP and General Manager of Artificial Intelligence at Twilio
Editor's Note: This and other original video content (interviews, demos, etc.) from ProgrammableWeb can also be found on ProgrammableWeb's YouTube Channel.
Audio-Only Podcast of Interview with Al Cook
Editor's note: ProgrammableWeb has started a podcast called ProgrammableWeb's Developers Rock Podcast. To subscribe to the podcast with an iPhone, go to ProgrammableWeb's iTunes channel. To subscribe via Google Play Music, go to our Google Play Music channel. Or point your podcatcher to our SoundCloud RSS feed or tune into our station on SoundCloud.
Full-Text Transcript: Al Cook, VP and General Manager of Artificial Intelligence at Twilio
David Berlind: Hi, I'm David Berlind. Today is Wednesday, February 12th and this is another edition of ProgrammableWeb's Developers Rock Podcast. We love developers and today we're talking to one of the darlings of the API economy. They make APIs that a lot of developers love. It's Twilio. You've heard of them, I'm sure. If you haven't, we're going to catch up on what they're all about. With me today is Al Cook. Al is the VP and General Manager of Artificial Intelligence at Twilio. Al, thanks for joining us today.
Al Cook: Thanks, David. Great to be with you.
David: Oh, it's great to have you. We haven't had you on the podcast before, so this will be a first time. Looking forward to it. For those people who've been living under a rock, don't know who Twilio is, tell us what Twilio does.
Al: Yeah, sure. So, Twilio is a cloud communications platform. You can use Twilio, whether you're a developer looking to integrate communications into your application, looking to add text messaging or phone calling into your flows, or whether you're an enterprise looking to deploy a flexible contact center or solve any of your customer engagement problems. Folks come to Twilio to be able to really tailor communications with their customers so you can build the exact experience and really get a great customer experience out of that.
David: I was reading some of the notes coming into this interview. You guys do like an extraordinary number of API transactions per day. What's that number?
Al: That's right. Overall, annually we power 800 billion human interactions every year and that's across 170,000 different customers who are using the platform for one type of customer engagement problem or another.
David: Yeah, that's amazing. I think I read that it amounts to something like 64 billion transactions per day. That's unbelievable! You must have a crazy scalable infrastructure in order to support all of that.
Al: Yeah. We've been building this for over 10 years and we're honored to be able to be part of pairing those communications that our customers companies are building on top of us. The volume is just testament to how important it is for folks to be able to really engage their customers in the right way.
David: What's a really good example of what one company is doing with your APIs?
Al: Yeah, the interesting thing about Twilio is there's all sorts of different use cases, whether people are using it for text messaging, voice calling, video calling, two-factor authentication or whether they're deploying a full-scale contact center.
Al: So for example, Lyft had deployed Flex, which is our application platform, which is a contact center that allows you to deploy it out of the box, but you can customize and tailor anything that you want about it. So, you can really integrate the front end-user experience, the agent experience, into your CRM, into all of the different backend systems that you need to be able to use as a contact center agent, and also customize all of the routing and all of the backend logic as well. And so, folks from Lyft and Shopify and all sorts of different, both digital-native companies and large enterprises have been using the Twilio sector to really build the experience they want.
David: You just said something that — want to stop you there because we're using that same phrase in an upcoming report about all the different business models that are out there for the API economy. You said, digital native. What do you mean by digital native?
Al: Yeah. Some companies grew up in a kind of a different era and they've been going through a sort of digital transformation and used Twilio to power that digital transformation and to move, to be able to accept the omni-channel communications and to be able to talk and move, to be able to engage in the way customers want.
Al: Other folks, have kind of grown up in this economy and have always been built to be on the very forefront of technology and using everything in their toolset. We work with folks on both sides of that spectrum and everything in between. But digital natives, folks who've really grown up in this economy have a very, very rapid way of adopting technology and are able to really, really use that to their advantage to disrupt industries, and move fast and make a big difference.
David: Yeah, they have a distinct advantage. And so a lot of those companies that are going through the digital transformation, they kind of aspire to move into a digital native state—
David: ... much like the people who are trying to disrupt them. Okay. You guys just released a survey. It's got some fairly significant findings in it. You've also made some predictions. So, what is the survey that you've released?
Al: Yeah, the survey is about the state of communications. As Twilio, we have this really interesting ability to survey the landscape because so many folks are coming to us from so many different angles and saying, "Hey, help me solve my problem. Help me engage my customers in a better way. Help me improve this flow." And so we get to see these bright spots of what is going on in the industry at large.
Al: And so, through the survey, we've covered trends like businesses really, really increasingly need to engage in long term conversations with their customers, not just the sort of transactional interactions, but long term conversations that last and really promote the value around the lifetime customer value. We also see a lot of trust issues that we've been working with the industry to resolve. So, things like robocalling has made a huge impact on, particularly here in the U.S. Last year, there were 58.5 billion robocalls in the U.S., which is just a staggering number. You think about how many that is per person per day. It is a truly staggering number. Robocalls kill trust in the phone and we've been working every day to build new systems that can help people kind of regain the trust of their phone.
David: I agree, by the way. I mean, I read in the report that 2019 was the year that people stopped answering the telephone and I can't argue with that. Here in my household, we literally let the landline just ring all the time. We never bother to answer it anymore. Sometimes, you look at the display to try to see who it is, but it's very hard to tell and half the time when you pick up the phone because you think it might be a local number, they've tricked you. They somehow constantly reprogram their systems and make sure they're calling local numbers from local numbers, so it looks like they're somebody from around the corner or the school or something like that, and it turns out to be one of these robocalls. So, I completely distrust the public telephone system right now.
Al: Right, right. We've been working to improve that, and working on systems whereby enterprises can authenticate their identity and working with partners to be able to show on your phone when you answer that call, that you know it really is from a call that you want to take.
David: Yeah. I read about this thing called SHAKEN/STIR. What is that?
Al: Yeah, SHAKEN/STIR, it's actually a federally mandated thing that companies have to adopt to be able to really authenticate and identify an endpoint and that's a big part of the problem. But also getting the identity down to the user's actual devices is part of that as well. We're working on solving the entire problem end to end. And then I think the other part is as folks look to engage in different channels and different methods, people need to be there on the channel that the customer, that the user, wants to be engaged on.
And we talked a bit about that's important for business, but we also see that being important in politics as well. We power a lot of political communications on our platform and we see that politics are moving beyond the polls and things like being able to text message your constituents, being able to engage, again, in a long term and a meaningful dialogue is really, really important and we're helping to help political parties and political candidates reach people in the way that they want to be reached as well.
David: Yeah, I saw that one of the big trends was the fact that now the engagement cycle in politics is longer. In fact, we're constantly engaged. Once an election's over, we're re-engaged almost immediately right up to the next election, whether it's two or four years away. And so, I can imagine, especially given what you just said about the distrust over robocalls, the different channels of communication that are available to us, text messaging, mobile applications, our landlines, on the web, et cetera, I'm assuming that as users or most users would like to establish some sort of preferred channel of communication and then keep it there so that they're not getting overwhelmed on these other channels by the same company or organization. Is that a part of your—
Al: Right. Yeah, I think that's definitely true. People have their preferred channels and we see this all the time, that customer experience is increasingly a differentiator and the companies who are doing great at this are folks who are really reimagining the journey from the ground up from a customer perspective and thinking about things like well, what channel do they want to be contacted on? What is the right way and what is the right method of communicating exactly the right information at exactly the right time? And these are the kinds of things—
David: I went through this just yesterday. I was in a retailer store, I won't mention who they were, and I went to check out and it asked me at the checkout counter, the credit card machine asked me if I wanted to receive offers of specials over my mobile phone via text. But I already have that retailer's application on my phone and I was surprised that I was being asked this when I already have their mobile app and they could actually communicate with me that way as well. So, there's probably a bit of a challenge on the back end to create this sort of 360 degree view of a customer in a way that you know what channels they have available to them, whether or not you can use those channels or not and then which one's the priority.
Al: Right. Well, David, that is exactly the problem when you're deploying a sort of siloed SaaS application that just does what it was intended to do and nothing else because you can't integrate it with any of your other systems. And so you might end up with piece parts that seem like that makes sense on their own, but you can't tie the whole thing together to make a cohesive customer experience that is really to your point behaving how you want to do and kind of making sense together. And it's only when you have developer APIs and when you can really tweak and change and fine tune the behavior, that you can actually get the experience that you want. You have to be able to customize things.
David: Now, you're the GM and VP of AI, so tell me how artificial intelligence fits into the trends that you've spotted and some of your conclusions.
Al: Yeah, our trend was around conversational AI and really what we see is conversational AI, we often talk about conversational AI. Folks who hear conversational AI might think about things like a voice assistant, like an Alexa or Google Home or they might think about messaging bots on a website where you can type in and try and get help from a bot.
But really what we see is this is just a tip of the iceberg of what conversational AI can do. I think you see the early deployments of these, kind of these two main categories of things, because that happened to be a good place for folks to get started. It's been a good place for folks to experiment.
But the real power of conversational AI, when you can truly talk to a computer or type to a computer and it absolutely understands what you're saying, what you're doing, I don't think we've even begun to unlock the potential of that. And it doesn't, by the way, have to be just talking to a voice assistant or talking to a computer. It can be that computer is sort of in the loop on a human-to-human interaction, and then being able to extract more value out of that. Whether that's, for example, in a contact center, a use case, being able to help the agent and guide the agent in real time because you're able to understand how the conversation is going or being able to extract insights out of what's going back and forth. And those are the areas that I'm really interested in.
David: Well, you are the VP and GM of AI at Twilio. So, given all of those challenges, it sounds like there's a lot of opportunity, but there's some challenges. What is Twilio going to do about it? What do you have coming that's going to help businesses wrestle all of these opportunities to the ground and take advantage of them?
Al: Yeah, We've been working for a few years now in this space and we have a product, Autopilot, which is a self-service messaging bots framework where you can build conversational IVRs, you can build bots on them. I think one of the things that we've seen there is when conversational AI is deployed well, it's not deployed in a siloed standalone system. Often when you're talking to one of these systems today, you often find that consumers don't trust or don't believe that they're going to be able to get out of this interaction what they're looking for. And very often, that's because, either it's because not kind of true AI and it's just picking up keywords and you may be able to, it's sort of a glorified FAQ tool, but you can't actually get it to do a thing for you. Or it's because it's just not deployed and it's not integrated into a way and so you end up with a system where you're talking to a bot and then the bot gets to the point where it can't take you any further and then you're kind of ejected out of that experience and then into a sort of traditional human experience. That kind of forced eject is not a good experience for anyone really.
David: I think we've all been through that where you're going through the IVR system, you're hitting buttons and suddenly you hit the end of the road and you're like, "How do I get a human because this thing just died on me here?"
Al: Right, right.
Al: Folks who have been building on Twilio have been able to build a better experience in that whereby by you can have agents monitoring a whole bunch of bots interactions and being pulled in and out as they're needed and being able to kind of supervise those interactions and using the bot to sort of superpower the human rather than using it as a firewall in the front. And I think that makes a big difference from customer experience.
So, as we are working on that space, we've invested a huge amount in natural language understanding and being able to really analyze and understand a conversation. And as I was saying, I think being able to deploy that technology in other use cases becomes really, really interesting, right? So, take your contact center, for example. A contact center is this treasure trove of information that is typically not used very well. If you think every single one of your customers whoever calls you or ever messages you on their website, what they are saying about your products, about your company, about what they're struggling with, about what they're hoping to do, that sits in that contact center and very often that's where it remains, right? It never comes out of that.
And you think about, rewind, I don't know, 10 years ago or something and you think about the power that Google Analytics had in helping companies understand the pipeline of their prospects and like how do people navigate a website, and you could derive a lot of meaning from that. Well, take that and apply it to what your customers are actually saying to you in your contact center.
Al: That's even more detailed information. If you can analyze out and say, "Well, here are the reasons why people were frustrated, why they churned, why they... This is the set of events that need to happen for people to upsell." That is a hugely powerful business analytics source of information that really until conversational AI has got to the point it is at now, it's been very, very hard to do anything with that, right? So, typically a contact center might review 1% or 2% of the calls, but you can do so much more with that if you can analyze it 100% automatically.
David: Right. And I completely agree. I mean, who of us has not made a call to a contact center and suddenly you realize things aren't going so well. You start complaining to the agent that you're talking to. You know that the call might be recorded because they say so when you first make the call, but you're thinking in the back of your mind, "This is never going to get to the people who have to change this. And I'm about to leave this company as a customer because of the poor customer service," or something like that. You really have no faith that the conversation you're having with that agent is actually going to get to somebody who can do something about it.
David: So, and I can understand from the company's point of view, they're frustrated. They need to be able to surface the most important feedback that's coming through that channel and then be able to act on it in a way. And AI clearly, especially when it's scalable, can sift through that haystack looking for the needles and get them to the right people at the right time.
I want to come back to the predictions. You guys made four predictions as a result of the survey. Can you quickly go through those?
Al: Yeah, so the predictions are that conversations with businesses will be increasingly important. That conversational AI, we're just at the tip of the iceberg, that robocalls will kill trust in the phone and folks need to work on improving that, that politics moves beyond the polls and candidates need to engage in a thoughtful way and that customer experience is increasingly important as a differentiator.
David: Yeah. The report said that robocalls will be conquered in 2020. Are you confident about that? Are we there?
Al: I think we are making some huge progress as an industry with things like STIR/SHAKEN as we were talking about. And that number, 58.5 billion robocalls, I think that is a number that we can make a material difference on in 12 to 18 months and we are working incredibly hard and I think we'll see them very significantly diminished. Will they go down to zero? That's a different question, but will they be significantly diminished? I think we can make a big, big impact on that.
David: Well, I pray that you're right. And for developers, because this is, after all, the Developers Rock Podcast, what is your advice to developers here given all these trends? I mean there must be some developers out there thinking, "Hmm, you know what, this whole thing about customer experience is a big deal. I should get smart about that. And when I'm out there building applications for my clients, let's say, or for the company that I work with, I should kind of take a little more ownership of the customer experience side of things to make that organization more successful at what they do." Is that kind of one of your current efforts to get developers educated on these issues?
Al: Yeah. We spend a lot of time doing design sessions with development teams within our companies. One of the things that I think is interesting across all of these trends is the need to really, really redesign customer engagement players almost from the ground up and really think about how do you build the right experience.
Take conversational AI, for example. There's a lot of thought that goes into how do we want our voice to come across as a brand? How do we want this kind of flow? Or how do we, what kind of feeling do we want out of this? It's much more than just a technology and so the interactions that I enjoy the most are where we have developers and designers sitting in a room together and really thinking about how do we design this from the ground up without being kind of shackled by the kind of traditional legacy systems where you can kind of hammer the thing into submission and make it do what you want.
David: A lot of points of view have to be brought into that conversation. People who know something about artificial intelligence and what it's capable of, people who understand the customer experience and, of course, developers, because at the end of the day, it's their job to string it all together into something that's really frictionless, right?
Al: Right. Yep. That's exactly right.
David: Okay. Well, thank you very much for joining me today.
Al: Yeah, thank you. It was great talking with you. I really enjoyed it.
David: It was great to have you. We've been speaking with Al Cook. He's the Vice President and General Manager of Artificial Intelligence at Twilio, one of the darlings of the API economy.
When you see this podcast, you may also find the full-text transcript on ProgrammableWeb. Just search Twilio in our search box. It'll probably lead you there and please come back to both our YouTube channel at youtube.com/programmableweb for more interviews like this one, and please, of course, come to ProgrammableWeb.com where you'll find all of the text as well as the audio where you can just listen to the podcast on your iPhone or your Android device. Thanks for joining us. We'll see you next time.