A little over two years ago, before Facebook Messenger or iMessage opened to third-party bots, and before the arrival of Google Assistant, conversational AI champion Chris Messina helped coin the term “conversational commerce.”
Messina, who is perhaps best known as the creator of the hashtag, has since 2015 examined trends like chat apps surpassing social media in monthly active users, buzz among developers and investors, and unanswered questions, like how experiences created by third-party developers can attract attention in a conversational world.
Last year, Messina left his job as developer experience lead at Uber to explore opportunities in conversational AI. Next week, he launches Molly, a service that skims your social media accounts to create a bot that answers questions the world wants to know about you.
Messina spoke with VentureBeat about the state of conversational AI, his new startup’s ambition to reimagine the value people derive from social media, and who’s winning the chat wars between Alexa, Siri, and the like, which he calls “god bots.”
This interview has been edited for brevity and clarity.
VentureBeat: It’s been a while since you first started talking about the idea of conversational commerce. Has anything you’ve seen since then surprised you about this conversational AI and conversational computing space?
Messina: I think the insight I was trying to help people see was less about the interface and more about the programmatic technique of putting software behind the interface. In other words, the notion of conversation as an interface would become normal, as opposed to everything being button-based, and I think that continues to be true.
It’s just like Jeff Bezos was saying: We’re at the very, very early innings of this. And I still believe we’re at the very beginning of this shift, and that this shift will be a lot harder than getting the graphic user interface (GUI) because the GUI is something you can show someone and they can learn from watching and you have a fast iteration cycle of improvements.
The next wave of software is deeply personal, [requiring] enormous amounts of personalization and pattern recognition and matching the way a system responds to an individual.
But things that surprise me — some of the things, like the primacy of the “god bots,” is very clear. Talking to Siri, talking to Google Home, talking to Alexa, they have not figured out yet how to do an elegant handoff to third-party services. So the lack of addressability relative to the home screen grid of icons on a GUI interface is still a huge barrier to third-party developers succeeding on these platforms.
I do think it’s interesting and not too surprising that games seems to be the thing that Amazon is pointing to as a way of showing engagement, as a way of promoting these devices and saying “Here’s something that’s kind of new and different and a little bit easier than other games that you might have played with your family on a TV or something.”
It’s hard for me to say I’m surprised because these things are always so slow and people are always so skeptical. Unless they have something super incredible or amazing, it’s very hard to motivate people to change their behavior, so what we’re seeing instead is a slow, gradual saturation of the market with voice-enabled devices, but then the real question is whether or not third-party developers are going to have some sort of toehold. In this context, platforms need to figure out or overcome the memory problem, which is: How do I remember to use this skill? How do I remember what my device can do?
That to me is where conversational brands are completely necessary in this new world, which is — How do you sort of stay top of mind? How do you nest yourself in someone’s brain so that they know to think of you, and then how do you deliver constant value?
So I guess I haven’t been too surprised, but I think we have several generations or years before this stuff becomes the dominant paradigm. I see AR and VR being complemented in a conversational world, because so much of the interactions in that world are not about typing but about verbal expression and voice commands, so I think what Facebook might be doing with their VR play and with Portal suggests to me a way of thinking about this that could be somewhat interesting and emergent, but I just don’t know.
VentureBeat: You mentioned the sort of god bot, the ubiquitous AI-powered assistant I can speak to anywhere; it’s in my light switch, it’s in my car, it’s in my workplace. Who’s winning?
Messina: I think Amazon’s winning. However, I’ve been surprised about some of the stats around Google. I just feel like Google says they have all these Google Home devices that they’ve sold and stuff, and I suppose maybe that’s true from an international standpoint or other markets, but if you count everything Google Assistant does (Android smartphones, Android tablets, Home speakers, and Android Auto, to name a few), they might be overwhelming Alexa devices in some way, again depending on how you count. But to my mind, Amazon has the lion’s share of developer attention and developer interest.
Apple seems to have very little, if any. Sort of like their iMessage Business Chat stuff — I don’t have a sense that people are excited about it. I mean they’re excited about it because it’s a huge channel with a lot of reach, but they’re tightly controlled channels, and when Apple opens up they tend to be really reluctant and slow. Their playbook is a little different than Facebook.
Facebook has a very clear playbook where it’s — Grow to a billion users, and then we figure out how to monetize and, you know, we may have to open up APIs along the way, whatever it takes to get there, we’re going to get there.
Apple’s like, Whatever, fuck it. We’ve got years and years worth of selling hardware. We’re going to take the slow, determined approach. We’ll see what works; maybe we’ll open up a few APIs here and there for complementary things or things we can use as our showcase piece to tell a story, but ultimately they don’t allow developers to control too much of the experience except for the experiences they really want to support and monetize, whereas Amazon is kind of like, Look, just build shit and we’ll figure out how to monetize; you guys are using all of our other developer services, and it seems like building on Alexa seems pretty straightforward and people seem pretty able to do it relative to the other platforms.
What was your take on the Super Bowl ad?
VentureBeat: I mean, it was smart to build some suspense and stuff like that, but I half wanted them to drop a new feature for people to test out Super Bowl Sunday, a day when you’ve got all your friends and family in your house. Seemed like a good time to give people an excuse to interact with their Echo for a wow moment or to have a good time with it. Seemed like they set it up for the celebrities to knock them down, but they could have made it a lot more key to their business if they actually did something new, like give Alexa a male voice, for example.
Messina: It was super weird that they had people wearing these strange headsets.
Messina: What the hell is that about? No one’s going to wear those. I mean you basically put two Echo Dots on people’s heads, and I’m like, This is why you’re not Apple.
VentureBeat: Well, I mean, the mobile accessory kit — there are no headphones made by Amazon with Alexa inside available today, and the mobile accessory kit was launched about a month ago before CES. That would make anyone who’s been paying attention to Alexa say, Hey, wait a minute, but I guess that’s not part of it. Whatever.
Messina: What’s your take on who you think is winning?
VentureBeat:I think it goes back to what you were saying — like with Google, how you gauge the math on all this. There was this snippet of a sentence that Cheddar did on Portal about a month ago where they were saying Mark Zuckerberg doesn’t care at all about making money from Portal.
The idea is to change user behavior as it relates to Facebook, so maybe Stories is a bigger part of what you do and your News Feed is less important to what you do perhaps and Watch is a bigger part of what you do, and streaming content from these things and maybe AR Spaces and all of that stuff blended in and available in one place, and you can do all sorts of stuff with facial recognition — understand user response to advertising or prove that you’re actually making your users happier.
Messina: Wow, that’s crazy. I hadn’t thought about that idea where the screen or whatever is the way of conveying visual images, and you bring facial recognition or facial detection into a space and Facebook shows you an ad and, based on your face reaction, if it falls into one of the five reaction types, advertisers get that data and get a sense for emotive quality that someone is actually having about your ad.
VentureBeat:Yeah, you’re making a lot of females ages 23-29 smile [for example].
Messina: Haha, right, yeah.
VentureBeat:So I think this user behavior idea sort of determines how each of the different companies will decide whether or not it’s successful, and Amazon isn’t necessarily trying to get into the mobile operating system, but that’s where Google has its dominance and so at least on that front there’s a lot of people that can use Google Assistant. But Amazon’s user behavior paradigm shift may be that they want to make it normal to shop with your voice, and apparently that’s true for some people. I was talking to Adobe recently and they were talking about 22 percent of people who own a smart speaker find it to be a normal thing to shop with your voice. When you think about the amount of time it took for people to go from getting a smartphone to being willing to shop with a smartphone, this seems a lot faster than that to me.
Messina: Also think all the people who have done QVC or voice catalog ordering. It’s actually a known behavior — saying it out loud, of course. Now on the end of a phone line might be the difference, but in a way it’s kind of the behavior that’s already been there.
VentureBeat: Totally, but I’m impressed by what Google has done so far. They haven’t just stayed neck to neck with Amazon, they seem to have made up some ground. What Facebook does later this year, what Samsung does later this year, how people respond to HomePod, that will all play into how this space looks in a while.
Messina: Apple Music is the linchpin in that relationship. I mean you have all your music there, and it’s super sticky. Like, clearly the fact that HomePod doesn’t work with any other service, it means that they’re being very cautious about what goes into it.
I listened to the ReCode podcast with the CEO of Sonos because I’ve been wondering what the hell Sonos is going to do. I guess later this year they’re adding AirPlay 2 support to Sonos devices, so there’s an interesting A-B test, if you will, with Sonos to see what happens when there’s a platform that is totally open and platform-agnostic relative to the HomePod, which is an expensive first-gen speaker. Even if it sounds amazing, which I’m sure it does, it still has to compete with content access. It appears as if Apple is placing a bet that as long as they have the best-sounding speaker at the relative right price point, then that’s actually going to get people to come back into the Apple fold and stay there.
VentureBeat: I just think of buying, especially knowing what the sound quality is on a Home Mini, buying two Home Minis and a Home being much cheaper than one HomePod seems like a pretty good sell (especially with multi-room music and intercom features).
Messina: Yeah I’m kind of like, What is Apple really playing at here?
VentureBeat: So what’s Molly?
Messina: So Molly ingests a lot of the social media content you produce across different platforms — like Twitter, Medium, and Instagram — to create a kind of personal search engine for stuff you’ve previously written to allow people to ask natural language questions of that content.
In a case where we don’t have the right answer or don’t know what the right answer might be, we actually send that on to you, and then you get a notification from the app which you can answer.
My question is, Is there a way to build social media that is more about rewarding human connection and relationship and deepening the awareness and knowledge that we have about each other through this type of serendipitous type of discovery of a person’s background?
The hope is that we build a conversational agent that learns about people over time and is able to answer questions about them over time to try to build a social platform that understands people, how they ask questions, the types of answers they’re looking for, and allows us to build up ideally a very useful and valuable dataset that we make available to our users and give them something back for all the content they’ve contributed to the social web over time.
VentureBeat: And why the name Molly?
Messina: One of the big problems with a lot of these voice skills and apps is recall, recollection of their name, so we wanted a name we could develop a kind of persona around. While the Olabot brand was timely, I don’t think it necessarily stuck in your mind the same way a person’s name might. Another part was that I just happen to know a woman, Molly Holzschlag, who is a web pioneer and owner of the molly.com domain.
Then, of course, there’s the more counterculture element to it, which I can leave to your imagination. This product is about connecting people to digital media. There’s a connotation there that might not be terribly negative, but happens to be a happenstance.