Listen Labs Bets That AI-Powered Customer Interviews Will Replace Traditional Market Research

In less than a year, Listen Labs has gone from a viral hiring stunt on a San Francisco billboard to a $500 million valuation on the promise that AI-run customer interviews can outpace—and eventually replace—traditional market research. For product leaders, UX researchers, and growth teams used to slow, expensive studies and noisy survey data, the company is positioning itself as an alternative: automated, open-ended conversations at scale, turned into executive-ready insights within hours.

Backed by $69 million in new Series B funding led by Ribbit Capital with participation from Evantic and existing investors Sequoia Capital, Conviction, and Pear VC, Listen Labs has now raised $100 million in total. In nine months since launch, it says it has grown annualized revenue 15x to the eight-figure range and run more than one million AI-powered interviews.

Underneath the attention-grabbing billboard story is a more fundamental bet: that AI can fix some of the most persistent pain points in a $140 billion industry built on surveys, interviews, and panels—and in the process, change how product decisions get made.

From cryptic billboard to $69M in funding

Listen Labs first drew broad attention with a $5,000 billboard in San Francisco covered in what looked like five random strings of numbers. For most passersby, it was gibberish. For the engineers the company wanted to hire, the numbers were a puzzle: AI tokens that, when decoded, led to a coding challenge.

The challenge asked candidates to build an algorithm that would act as a digital bouncer for Berghain, the Berlin nightclub known for turning away most people at the door. Thousands attempted the puzzle. Four hundred thirty people solved it; some were hired. The winner flew to Berlin on an all-expenses-paid trip.

That stunt—viewed roughly five million times on social media, according to CEO and co-founder Alfred Wahlforss—was less about branding than necessity. The company needed to hire more than 100 engineers while competing with compensation packages as large as $100 million from tech giants. Early employees joined before the startup even had a working toilet.

The billboard and the hires that followed became the foundation of the engineering-heavy team that built Listen’s platform. The company says about 30% of its engineering team are medalists from the International Olympiad in Informatics (IOI), a competition that has produced founders of other prominent AI startups. Wahlforss’s co-founder previously worked on Tesla Autopilot and was a national champion in competitive programming in Germany.

That concentration of technical talent is now focused on a very specific problem: turning qualitative depth into something that can move at the speed of modern product and marketing cycles.

Why traditional market research struggles to keep up

Listen Labs’ core premise starts from a critique of how most organizations do customer research today. Product and marketing teams are typically forced into a trade-off:

• Quantitative surveys offer statistically clean-looking numbers but often shallow, biased, or inauthentic responses.
• Qualitative interviews and focus groups deliver the nuance teams want but are slow, expensive, and difficult to scale beyond small samples.

Wahlforss argues that surveys in particular create a misleading sense of certainty. When most people are clicking through the same multiple-choice options, the resulting percentages look precise but can hide distortion.

“Essentially surveys give you false precision because people end up answering the same question,” he said. He points out two problems: outliers are hard to surface, and people often aren’t fully honest when given fixed options and clear incentives to complete a form quickly.

Traditional one-on-one interviews solve some of that: researchers can ask follow-ups, test for understanding, and explore unexpected directions. But they don’t scale. Recruiting participants, scheduling calls across time zones, conducting sessions, transcribing, coding, and synthesizing findings can take weeks or months. By the time insights arrive, the product decision may have already been made.

Listen is designed to compress that cycle into hours and make in-depth conversations as accessible as sending a survey link.

Inside Listen Labs’ AI-first research workflow

Listen’s platform is structured around a four-step workflow aimed at replacing separate tools and vendors with a single, AI-driven layer:

1. Study design with AI assistance. Teams define objectives—such as testing a new concept or understanding churn—and the system helps draft the study structure and questions.
2. Automated recruitment. Listen recruits participants from a claimed global network of 30 million people, targeting the desired profile.
3. AI-moderated interviews. Instead of surveys, participants engage in open-ended, often video-based conversations with an AI “researcher” that can ask follow-up questions.
4. Synthesis into deliverables. The platform packages key themes, highlight reels, and ready-to-use slide decks for stakeholders.

The emphasis on open-ended video responses is central. Where surveys nudge participants toward guessing the “right” answer among four or five options, open responses give people more freedom—and, according to Wahlforss, tend to elicit more candid, detailed feedback.

From a product or UX team’s perspective, this is meant to blur the line between traditional quantitative and qualitative methods. The interviews retain the conversational depth of qualitative work, while the AI summarization attempts to deliver something closer to the clarity and speed of a dashboard or survey report.

Listen’s pitch to teams is that this format can bring “the customer into every decision,” from go-to-market to UX changes. The business growth figures the company shares—15x annualized revenue growth to eight figures within nine months—suggest that at least some organizations see enough value to integrate it into existing research workflows and budget lines.

Fraud: the hidden cost center in a $140B industry

Beyond speed and format, Listen Labs is explicit about another problem it believes is undermining research: fraud and low-quality responses across the panel ecosystem.

When the company began building out its global participant network, Wahlforss said they ran into what he described as one of the most surprising discoveries in the space: the prevalence of bogus respondents, even in panels run by large, established companies.

“Essentially, there’s a financial transaction involved, which means there will be bad players,” he explained. According to him, some of the largest vendors—companies with billions in revenue—were sending participants who claimed to be enterprise buyers but were flagged as fraudulent as soon as they entered Listen’s system.

To address this, Listen built what it calls a “quality guard,” a set of checks that includes:

• Cross-referencing LinkedIn profiles with video responses to validate identity and role.
• Checking consistency across how a participant answers different questions.
• Flagging suspicious behavioral patterns that may indicate bots, professional survey takers, or misrepresentation.

Wahlforss claims that with this kind of vetting in place, people “talk three times more” and show greater honesty, especially on sensitive topics such as politics and mental health. While those specific multipliers come from the company’s own reporting, one customer example illustrates the potential upside.

Online education provider Emeritus, which uses Listen, reported that roughly 20% of its prior survey responses fell into fraudulent or low-quality categories. With Listen, Assistant Manager of Customer Insights Gabrielli Tiburi said they “did not have to replace any responses because of fraud or gibberish information.” For teams that spend weeks cleaning data or running replacement samples, eliminating that 20% loss can materially improve both costs and timelines.

The broader implication for product and marketing leaders is that the real cost of research may be higher than line items suggest. If a fifth of responses are untrustworthy, entire strategies can end up anchored to misleading signals. Listen is effectively arguing that AI can serve not just as a researcher, but as a gatekeeper for who gets into your sample in the first place.

How Microsoft and consumer brands are using AI interviews in practice

Listen Labs’ value proposition becomes more concrete in how customers describe their use cases. Across enterprise and consumer brands, a common theme is compressing timelines while preserving (or improving) depth.

At Microsoft, Senior Research Manager Romani Patel described a familiar challenge: traditional research cycles of four to six weeks meant that by the time insights were delivered, product or marketing decisions had already moved on. With Listen, Patel says the team now gets insights in days, and often within hours.

Microsoft used the platform to collect global customer stories for its 50th anniversary, asking users how Microsoft Copilot was helping them “bring their best self forward.” According to Patel, video stories that once would have taken six to eight weeks to gather were collected in a single day.

Smaller consumer brands report similar shifts:

• Simple Modern, an Oklahoma-based drinkware company, used Listen to test a new product concept. It took about an hour to write questions, another hour to launch the study, and around 2.5 hours to get feedback from 120 people nationwide. That sped the team from “Should we even have this product?” to “How should we launch it?”
• Chubbies, the shorts brand, used Listen to overcome scheduling barriers in youth research—school, sports, dinner, and homework. Director of Insights and Innovation Lauren Neville reported a 24x increase in participation, from 5 to 120 youth participants, by letting kids respond asynchronously instead of joining live focus groups.

In Chubbies’ case, AI-led interviews also surfaced a product issue that might have otherwise gone unnoticed. Through repeated conversations, the system detected recurring complaints about the liners in kids’ shorts being “scratchy.” The brand responded by redesigning the product, which Wahlforss says went on to become a “blockbuster hit.”

For practitioners, the examples hint at two distinct modes of use:

• As a rapid validation loop for concept decisions that would previously have relied on gut feel or small internal samples.
• As a continuous listening channel, where unexpected patterns—like the “scratchy liner”—emerge from accumulated qualitative data.

Why cheaper research may lead to more research, not less

Listen Labs isn’t just positioning itself as a cheaper, faster alternative to incumbent research suppliers. It also argues that lowering the cost and friction of talking to customers will actually expand the total amount of research that organizations conduct.

The company points to a roughly $140 billion annual market research industry estimate from Andreessen Horowitz as evidence of the current spend on surveys, panels, and agencies. Wahlforss says Listen is already replacing existing budget lines that are “super costly,” slow, and locked into the survey-or-interview dichotomy.

But he frames the more important dynamic using the Jevons paradox, an economic concept originally applied to resource use: when a resource becomes more efficient to consume, overall consumption can actually increase rather than decrease. Applied here, the “resource” is customer understanding.

“What I’ve noticed is that as something gets cheaper, you don’t need less of it. You want more of it,” he said. In his view, demand for customer insight is effectively infinite. When teams can get usable feedback in hours instead of weeks:

• Dedicated researchers can run an order of magnitude more studies across the product lifecycle.
• Non-researchers—like product managers, marketers, and operators—can run their own lightweight studies as part of their day-to-day decision-making.

For organizations, that could mean a shift from large, episodic research projects to a more continuous, embedded approach. The risk, of course, is information overload or the temptation to chase speed at the expense of methodological rigor. Listen’s bet is that workflow design and automation can keep teams focused on high-value questions while the platform handles recruiting, moderation, and synthesis.

Engineering culture and growth behind the product

Listen Labs’ product approach is shaped by its origins. Before building the current platform, the founders created a consumer app that hit 20,000 downloads in a single day. The scramble to understand those early users is what led them to prototype the AI interview system that became Listen.

The company has leaned into hiring engineers not just for core product work, but across marketing, growth, and operations—a reflection of its belief that technical fluency is increasingly important in every function in the AI era. In 2024, Listen grew from 5 to 40 employees and plans to scale to around 150.

The early, scrappy days—complete with the lack of a working toilet—contrast with the current funding and valuation, but they also illuminate the culture that produced experiments like the Berghain billboard. In a Bay Area environment where AI talent is heavily contested, Listen’s leadership has been willing to try nontraditional tactics to attract and retain the kind of engineers who can push AI products into new territory.

For customers, that emphasis on technical excellence is partly a signal: the same team that cracked hiring challenges through a puzzle-laden billboard is now designing systems for fraud detection, interview moderation, and data handling in sensitive research contexts.

Looking ahead: synthetic customers, automated actions, and guardrails

With its new funding, Listen Labs is planning products that move beyond interviews and summaries. Wahlforss outlined two major directions: simulating customer voices and triggering downstream actions automatically.

First, the company is working on the “ability to simulate your customers,” using the corpus of interviews already conducted. The idea is to extrapolate from real conversations to create synthetic users or simulated user voices. In practice, this could allow teams to test messaging or product changes against virtual customer personas derived from prior data, without running a full live study each time.

Second, Listen wants to connect research findings directly to operational systems. Wahlforss imagines agents that don’t just suggest actions, but can be “spawned” to execute them: changing code, reacting when a customer is about to churn by automatically offering a discount, or making other targeted interventions.

He is also explicit about the ethical tension in such a roadmap. “Automated decision making overall can be bad,” he acknowledged, emphasizing that the company plans “considerable guardrails” to keep companies in the loop. For product and research leaders, this underscores a likely future trade-off: how far to push automation in customer-facing decisions while maintaining oversight, fairness, and compliance.

On the data handling side, Listen says it takes a conservative stance. Wahlforss states that the company does not train models on customer data, and automatically scrubs sensitive personally identifiable information (PII). In contexts like investor-related research, its AI is designed to detect and remove potentially material, nonpublic information from outputs.

All of this depends on the continued improvement of underlying AI models and on enterprises’ willingness to trust AI-mediated research and action. A 2024 MIT study, which found that 95% of AI pilots fail to make it into production, is a reminder that most initiatives in this space don’t yet reach durable, scaled use. Wahlforss cites that statistic as a reason for his focus on quality over impressive demos.

What AI-mediated research could mean for product development

If Listen Labs’ model gains widespread adoption, it could change not just the research function, but how entire product teams operate.

One customer example illustrates this potential shift. An Australian startup using Listen structures its workflow around time zones: developers code during the day, then launch a Listen study overnight aimed at a U.S. audience. By the next morning, they have feedback on what they shipped, which they can feed into coding tools like Claude Code to inform the next iteration.

This creates a near-continuous loop: build, collect customer reactions asynchronously, refine with AI coding assistance, repeat. In that sense, Listen extends Y Combinator’s classic advice—“write code, talk to users”—into a more automated cycle. As code generation becomes increasingly automated, “talk to users” is also on a path to automation via AI-run interviews.

Whether this leads to better products will depend on how teams use the tools. Faster cycles can mean more experiments and closer alignment with customer needs, but they can also amplify the impact of any biases or blind spots baked into how questions are asked and how results are interpreted.

Some customers report that the benefits go beyond speed. Microsoft’s Patel says Listen has “removed the drudgery of research and brought the fun and joy back” into her work, suggesting that automating rote tasks—like recruitment logistics and transcription—can free researchers to focus on framing questions and interpreting results. At Sling Money, a stablecoin payments startup, marketing manager Ali Romero said the ability to create a survey in 10 minutes and get same-day results is “a total game changer.”

Wahlforss frames the company’s philosophy with a line from Nat Friedman, former GitHub CEO and Listen investor: “Slow is fake.” In an industry built on caution and methodological rigor, that’s a provocative stance. Listen is wagering that in the AI era, the organizations that learn to listen faster—without sacrificing quality—will have a structural advantage in how they build and iterate products.

The open question for product, UX, and growth teams is how—and how quickly—to incorporate these new capabilities into their own practices. As AI-mediated interviews become more common, the teams that get the most value may be the ones that blend old and new: borrowing the discipline of traditional research, while using automation to talk to more customers, more often, at the moments when decisions are actually being made.