Voicelabs, a company that has been experimenting in the voice computing market for some time with initiatives in advertising and analytics, is now pivoting its business again – this time, to voice-enabled commerce. The company is today launching its latest product out of stealth: Alpine.AI, a solution that builds voice shopping apps for retailers by importing their catalog, then layering AI technology on top to better answer consumers’ voice queries, and ultimately convert those queries to purchases.
While this is the third product Voicelabs has developed over the past couple of years, this level of experimentation is to be expected in the nascent voice ecosystem. (And “labs” is even in the startup’s name, we should note.)
Initially, Voicelabs believed dabbled with third-party advertising. It allowed advertisers to reach a large smart speaker customer base by aggregating voice app developers into a network whose reach could be sold to advertisers en masse. But Amazon, sensitive to the way ads could disrupt the consumer experience, changed its policies and forced Voicelabs to shut down that product.
The company then shifted to voice analytics, a more competitive market with a number of existing players. From here, the idea for Alpine AI grew.
While Voicelabs counted over 3,500 analytics customers, it found that some percentage of those had interest in e-commerce. But voice shopping apps weren’t yet available on either Google Assistant’s or Amazon Alexa’s platforms, outside of the first-party shopping experience both provide.
Of course, in Amazon’s case in particular, that may be because it’s hoping that voice devices in the home will eventually become a key way consumers shop for products from Amazon.com, and not necessarily from retailers directly. Google, meanwhile, is more open to working with partners in the e-commerce space through services like Google Express, for example, where it has teamed up with big retailers, including Walmart and Target last year.
But retailers don’t necessarily want to hand off the entire voice Q&A process, including answers about their products, recommendation and discovery, and, ultimately conversions, to Amazon or Google entirely.
That’s where Alpine comes in.
In a matter of days, Alpine can digest the retailer’s website and CMS, then automatically create a voice app using that data. The app is then enhanced by machine learning technology, which learns from incoming consumer queries how to best respond to questions.
Some of these improvements are done before the app is even live. For instance, Voicelabs sent out an employee to document what sort of questions shoppers were asking about one of its client’s products in the retail store itself.
“Consumers – we started to see about six months ago – they’re asking about products,” explains Voicelabs co-founder and CEO Adam Marchick. “They’re asking ‘should I buy this or that?’ Or, ‘tell me about this mascara.’ Or, ‘tell me what’s new,’” he says . “We saw this happening both in third-party apps and on the platform itself.”
The AI technology comes into play as customers introduce new queries and to better refine the voice app’s responses.
For example, a customer may ask a beauty product retailer for a “long-lasting mascara,” of which several dozen products are available. The AI may know, based on earlier interactions, that the best way to narrow down that selection is by price. Later, if another customer asks for a long-lasting mascara that lasts for more than 12 hours, the system has to adapt to this new query. Machine learning aids in building out a new custom path for the app’s response here, too, again based on this existing data.
While today’s retailers are taking note of the voice assistance space, it’s unclear to what extent the platform makers themselves will allow other retailers to enable their own voice shopping experiences.
But Alpine could work around whatever restrictions are in place by offering a simpler discovery experience that later follows up with a customer email, for example. Or it could tie into the voice shopping experience the platforms offer – like allowing customers to complete the purchase on Amazon.com, if that’s an option for the given retailer.
However, Marchick believes the platforms be welcoming to retailers’ developing their own voice shopping apps.
“We’re in great communication with both of them. Both platforms know about Alpine, and that’s as far as I can go,” he says, somewhat cagily, referring to Google and Amazon.
“Here, we’re much more confident in Alpine being embraced by the platforms [compared with the earlier ad product], because it’s 100 percent in line with their goals and policies. The platform with the best consumer experiences is the one that wins,” Marchick says. “And Alpine will be powering the best path-to-purchase experiences.”
The company isn’t permitted to talk about its clients yet, but does have customers from the e-commerce 100.
Alpine is sold as a Software-as-a-Service solution with custom pricing, based on the retailer’s goals and e-commerce volume.
The company also closed its seed round a few months ago, in advance of Alpine’s launch. It’s now backed by investors including the Chernin Group, Javelin Venture Partners, Betaworks, as well as angels Tim Tully (Splunk CTO), Jon Brelig (InfoScout CTO, acquired), Scott Cannon (Mailbox CTO, acquired by Dropbox), and John Kobs, CEO of ApartmentList.com.
The size of the round is not being disclosed, beyond Marchick’s claim that Voicelabs is now “well-funded.”
Voicelabs’ analytics product will be live until March 29, 2018, to give existing customers time to migrate. Alpine, meanwhile, is open for business today,