Sign up for our weekly newsletters, and you’ll get the latest insights from our Dylan's Desk and DeanBeat columns before they’re published on VentureBeat.
Click here if you’d like my weekly column sent directly to your inbox. It takes less than a minute to sign up, and you’ll get the stories before they’re published on VentureBeat.
For decades, “artificial intelligence” was a long-sought-after ideal in computing circles. Now it’s finally starting to become real — just not in the way anyone expected.
Sure, the field of AI has produced a few stunning successes, like IBM’s computers beating chess master Garry Kasparov in 1997 and Jeopardy champion Ken Jennings in 2011. But it hasn’t yet delivered computers that can pass the Turing test or protocol droids that can act as interpreters during delicate negotiations with galactic separatists. Instead, today’s AI technologies help connect people with things they want, usually for the benefit of marketers.
An early-stage startup, Solariat, illustrates what I’m talking about: It uses powerful AI-derived technologies to insert itself into online conversations, providing relevant information in response to tweets.
Solariat is following a path blazed by Apple’s natural-language assistant, Siri, probably the most successful product of the AI field ever. Like Siri, Solariat took inspiration from SRI’s Calo project, an ambitious attempt to knit together many fields of AI and natural language research for military applications. The name “Calo” is partly an acronym for “Computer Assistant that Learns and Organizes” and partly a shortened version of the Latin word “calonis,” which means “soldier’s assistant.” The idea was to create an intelligent assistant that could provide soldiers with up-to-date information in the field, by responding to simply voice commands and queries.
It turned out that building a working, functioning Calo was beyond the capabilities of a Darpa research grant, even a very generous one. However, SRI generated several successful spinoff companies from its work. One was Siri, which Apple acquired an then integrated into its iOS mobile operating system. Another was Social Kinetics, a small startup that provided technologies that learned from interactions with customers and which Redbrick Health acquired in 2010.
Jeffrey Davitz, the founding chief executive of Social Kinetics and a former program manager for the Calo project, is now back in the game as the founder of Solariat. His new company is building quasi-intelligent agents that mine Twitter for examples of people expressing some kind of intent that’s relevant to Solariat’s customers.
For instance, if you post a tweet to your friends about how you need to buy a new laptop because your daughter spilled orange juice on your current one, Solariat could chime in with an @ message to you pointing to recent reviews of laptops.
The key, Davitz said, is that the technology identifies an expressed need accurately. “Otherwise you turn into Clippy,” he said, referring to Microsoft’s infamously obnoxious Office assistant.
Once the technology identifies that need, it matches it with content provided by one of Solariat’s customers — laptop reviews, for instance — and delivers it in the form of a targeted tweet. The goal: engaging customers and getting them to click on the offers, or as Davitz terms it, building an “engagement platform” that lets companies respond to people’s actual intentions, not just keywords.
Just as Google found that its search-driven ads don’t annoy customers as long as they’re relevant, Solariat is finding that contextually-relevant marketing tweets are welcome, too — as long as they’re on topic. Davitz said the company is seeing 20-30 percent clickthroughs on its targeted tweets, which is a remarkably high percentage.
The technology can be used for customer service as well as marketing. Or, Davitz suggests, publishers might use it as a kind of content distribution network: Instead of waiting for people to find your site via searches, Solariat could deliver relevant articles to you via Twitter @ replies whenever you express a relevant desire.
It’s a bit creepy, this notion of a computer combing through the entire flood of human expression on Twitter, searching for little tidbits of desire that it can latch onto and use to deliver something marketable. On the other hand, it’s also a potentially powerful marketing tool and has the possibility of decreasing the amount of junk you’re subjected to online. Rather than those ubiquitous and untargeted promoted tweets that keep popping up in my timeline, I might see tweets that are actually interesting to me.
It’s also a good business: Davitz says his company has successfully tested the technology, has paying customers, and is looking to build a more complete product offering in one or two vertical markets. So far, Solariat has raised $3 million, led by KPG Ventures, and aims to raise $3-5 million in its next round.
Marketing, customer service, and content delivery can be clever and profitable uses for AI — even if they aren’t exactly what Alan Turing envisioned 60 years ago.
Photo credit: pinkpurse via photopin cc
VentureBeat and marketing technology analyst David Raab are working on a new Marketing Automation usage and ROI study
. If you currently use a marketing automation system, help us out by answering the survey.
If you do, we'll share the resulting data with you.