It seems like everyone wants to build a virtual assistant that you communicate with via text message. In the case of New York City startup Agent Q, it’s an assistant that helps you shop online.
Co-founders said Rémi Cossart and Pramod Thammaiah pitch their service as a combination of Magic and Consumer Reports. Users should be able to text Agent Q with the details of what they’re shopping for, then the startup will provide a recommendation and, if asked, even make the purchase for them.
For example, when Cossart and Thammaiah stopped by the TechCrunch office before the launch, they started a chat with Agent Q, saying that they wanted to purchase a bike pump. Maybe a minute later, it came back with a description of a recommended pump and asked if they wanted to buy it.
Behind the scenes, Cossart said the recommendation was coming from “a human augmented by AI.” So there’s a team of people (specifically Agent Q employees) responding to your text messages, but they’re taking advantage of the data the company has aggregated across the web (including professional reviews, customer reviews and more) and some pre-written answers for popular queries.
In the demo, I was also struck by how little back-and-forth there was. What about getting all the specifications of what you’re looking for and comparing different options? Not surprisingly, the degree of interaction will vary depending on the product. In some cases, Agent Q needs more details, but bike pumps (as Thammaiah put it) are “all roughly the same” — so why not skip ahead and give you a solid recommendation from the start?
Of course, if you don’t like Agent Q’s recommendation, you can always refine your search by replying, “I want something cheaper” or whatever. And over time, the service will learn more about your preferences.
So why use a more narrowly focused text assistant, rather than something that can accomplish a broader set of tasks, like Magic or Facebook’s M? Cossart argued that by limiting their product to shopping, they can deliver a better user experience: “By specializing and staying focused, we can be really good at helping users shop rather than being mediocre at doing a lot of things like finding a bar, food ordering, helicopter booking, etc.
“Our view is that this it’s all about user trust and focus,” he said. “The way to get user trust is to give more than just the ‘I’m feeling lucky’ result. The best user experience is giving users not only a curated choice but also a clear explanation as to why it’s the best choice.”
Thammaiah has some experience with online shopping, having worked as a product manager on Google Shopping. He and Cossart (formerly a senior analyst at Cornerstone Research) said they’ve been friends since fourth grade.
Agent Q is available for free. Thammaiah said the company is considering several different options to make money, but the key is to avoid anything that would hamper the user experience or interfere with the objectivity of the recommendations.
The startup is currently accepting sign-ups for the waitlist. The first 100 TechCrunch readers who register here will get to jump the line.
Oh, and in case you were wondering, Cossart said that Q doesn’t stand for anything, at least not officially. Maybe they’re open to suggestions.