Every product team knows they should talk to their users more. When you ask product managers what would improve their decision-making, user research consistently tops the list. Yet most teams interview fewer than five users per quarter.
The gap between intention and action isn’t about caring. It’s about logistics. Scheduling interviews across time zones, finding participants who actually show up, moderating conversations without leading the witness, transcribing hours of audio, and then somehow synthesizing it all into something the team can act on. It’s a full-time job disguised as a simple question: “What do our users think?”
The compromise everyone makes
So teams compromise. They send surveys — response rates below 5%, answers that are one sentence long, and no way to ask “why?” They watch session recordings — which show what happened but never the reasoning behind it. They talk to the three users who happen to be available and extrapolate wildly.
The result is products shaped by educated guesses instead of genuine understanding. Features built on assumptions. Roadmaps influenced by the loudest voices in Slack rather than the clearest signals from users.
What if the interview could run itself?
This is the question that led to Maren. Not “what if we automated interviews?” — because automation implies checkbox surveys in a chatbot wrapper. But genuinely: what if an AI could conduct the kind of thoughtful, adaptive, probing conversation that a trained qualitative researcher would?
The kind of conversation where the interviewer notices something interesting in an answer and digs deeper. Where follow-up questions emerge naturally from what the participant just said, not from a predetermined script. Where people feel heard and share things they wouldn’t write in a text field.
That’s what Maren does. You tell her what you want to learn, and she talks to your users for you. Real conversations, not surveys. And then she synthesizes everything — across dozens or hundreds of interviews — into the themes and insights your team needs.
Early signals
What surprised us most in building Maren was what participants told us. 83% said they shared more openly with an AI interviewer than they would with a human. Not because the AI is better at interviewing — but because there’s no social pressure, no judgment, no awkward silences to fill. People are remarkably candid when they feel safe.
The other number that jumped out: participants write 142% more words in Maren interviews compared to traditional text-field surveys. They don’t just answer — they explain, qualify, give examples, change their minds. That’s the difference between data and understanding.
What we’re building toward
Maren is still early. We’re building in the open, learning from every interview she conducts, and improving her ability to have the kinds of conversations that surface real insights.
Our goal isn’t to replace human researchers. It’s to make research accessible to every product team — so that decisions are informed by understanding, not guesswork. If you’ve ever wished you could talk to more users, more often, without the logistical overhead, that’s exactly what Maren is for.