Getting answers to tough, qualitative questions about products from users can be costly, both in terms of time and money.
At least, that was the experience of Aaron Cannon, a former strategist at Deloitte, where he was responsible for facilitating research projects for Deloitte clients. Cannon and his team would spend hundreds of hours on a client project, only to have to devote additional time — and resources — to scheduling and moderating interviews with the said client’s customers.
“Enterprise decision makers expect faster and faster results from insights teams,” Cannon told TechCrunch in an email interview. “Researchers are feeling that pressure every day, particularly after being hit hard by 2022 layoffs. The biggest risk to the industry today is that the increasing speed of decision making leads to a decreasing ability for insights functions to keep up. That’s why researchers need the tools to accelerate and amplify their work.”
So Cannon teamed up with Michael Hess, who he met while working at Untapped, a talent recruiting startup, to found Outset. A Y Combinator-backed company, Outset autonomously conducts — and synthesizes — interviews.
“The broader slowdown and associated layoffs have hit research and insights teams disproportionately hard. But the demands from business leaders to make more informed and strategic decisions has not slowed down, leading to expectations of doing more with less,” Cannon said. “This is a tailwind for Outset as people look to technology to amplify their work.”
Outset taps GPT-4, OpenAI’s flagship text-generating AI model, to lead interviews with participants in research studies. How, exactly? Outset users create a survey and share the link with prospective survey takers. Then, Outset — powered by GPT-4 — follows up with respondents to clarify, probe on answers and create a “conversational rapport” for deeper responses.
For every question, GPT-4 generates themes, tallies up counts and highlights quotes to “uncover the story,” as Cannon puts it.
“Today, much of the work to collect and analyze qualitative data is done manually. In that way, we’re competing with the late nights I used to spend reading transcripts and scheduling interviews as a consultant,” Cannon said. “We believe that Outset will grow the market for research, making user insights faster and more accessible to more teams across the business.”
It’s early days for Outset. But, despite GPT-4’s imperfections and limitations, Outset’s already seen some success with a household brand: WeightWatchers. WeightWatchers was able to conduct and synthesize over 100 interviews in 24 hours, according to Cannon, the results of which are now being used to propose a new framework at WeightWatchers for user segmentation.
“We’re currently working with 15 enterprise insights teams at companies like Opendoor and other large consumer-oriented companies to help them make smarter, faster user-centered decisions than ever before,” Cannon said.
Outset, which recently raised $3.8 million in a funding round led by Adverb Ventures with participation from Weekend Fund and Jack Altman, Sam Altman’s brother, plans to expand the size of its team from four full-time employees to six by the end of the year.
“We just raised our seed and our team is small, so we’re keeping our burn rate low,” Cannon said. “Despite the broader economic slowdown, there’s increasing demand for AI-powered tools in everyday knowledge work, giving us another tailwind. Between our seed round of funding, low burn and accelerating tailwinds, we’re well-positioned to weather any storm.”