Weighing the Local Promise of Bots and Conversational Commerce
Are bots the future of the internet? Maybe, maybe not; like the buzz around Google Glass in 2013, we’re in the midst of a moment when it’s hard to tell the difference between hype and technological breakthrough. In one fell swoop, Facebook’s announcement at this year’s F8 that it would open its Messenger platform to bot developers transformed a vibrant but under-the-radar startup community into a certified phenomenon. At least it seems that way at this early stage, though the reality of the bot user experience has so far fallen short of its promise.
In a lengthy VentureBeat piece, Ken Yeung quotes Mark Zuckerberg as well as the visionaries behind such startups as Gupshup, Sonar, Assist, and Begin, to the effect that messaging powered by AI is a bold new platform in which bots are the equivalent of websites for desktop or apps for mobile. Natural-seeming AI-powered conversations, the theory goes, can reach consumers in an environment where they already hang out all day, and will eventually provide a more engaging alternative, across messaging apps and internet connected devices like Amazon’s Echo, to the walled-garden ecosystems of apps and websites. Bots will unobtrusively shortcut the steps to finding information, scheduling appointments, keeping up with the news, and most importantly, buying stuff.
Even the most committed proponents of our bot-powered future will readily acknowledge that the technology is still in its infancy, a fact borne out by early reviews of Messenger bots from companies like the Wall Street Journal, CNN, 1-800-Flowers, Shop Spring, and Poncho. Darren Orf, in a scathing review on Gizmodo, describes user experiences significantly more frustrating and less informative than a typical website visit.
It may be partly due to the newness of the technology, but thus far it’s hard to see how the typical search, browse, and buy model of online commerce has been disrupted by the notion of “conversational commerce,” where a chatbot guides you through the buying process by helping you find what you’re looking for and offering curated suggestions. In Orf’s example, a purchase via Messenger from 1-800-Flowers felt diminished in comparison to a website transaction.
I tried out Shop Spring and had a similar experience. The bot guided me down a predictable path, somewhat like a Choose Your Own Adventure novel or a text-based RPG from the 1980s, from “Men’s items” to “Accessories” to “Sunglasses” to a price range ($75-$250) and finally to a list of “five items we think you’ll dig.” If I didn’t like any of those five, I could keep searching, but in the end it felt like a narrow, circumscribed experience, with few choices and not much freedom to move around. For instance, if I decided I didn’t want to look at sunglasses anymore, I had to type “Go shopping” to start over again from the beginning. Compared with the web’s foundational concept, the non-hierarchical hyperlink, this particular chatbot’s imagination felt needlessly self-limiting.
On the other hand, early signs point to a more promising second phase in bot technology. Invite-only startup Ozlo got a favorable review the other day on BuzzFeed, suggesting that its focus on local food and drink recommendations fits the virtual assistant model particularly well. According to the review, not only can Ozlo find local cafes and restaurants, look up menus, and give directions; it can learn your preferences and provide better, more customized recommendations over time. Even better, Ozlo asks you to rate its answers with a Pandora-like “thumbs up” or “thumbs down,” letting users train the tool to improve over time.
Even more cutting edge is the new virtual assistant Viv, built by Norwegian developer Dag Kittlaus, the brain behind Apple’s Siri. Last week at TechCrunch Disrupt in New York, Kittlaus demonstrated Viv’s ability to parse complex questions that would defeat Siri or Google Now, like “Will it be warmer than 70 degrees near the Golden Gate Bridge after 5 p.m. the day after tomorrow?” Using stored credit card info and partner integrations with companies like Hotels.com and Uber, Viv can also book a room or hail a ride with simple, conversational interactions.
Ozlo and Viv aren’t yet ready for prime time, but they point to a brighter future for bots than the current technology might suggest. In particular, it’s notable that both tools focus much of their attention on mobile “here and now” queries. Compared with Shop Spring’s not entirely successful attempt to simplify online shopping, Ozlo and Viv want to get you where you’re going in the real world with a minimum of fuss. Virtual assistants that help you navigate the world around you might be the best means of bringing consumers into the fold.