Throughout my career, I have been conducting interviews with customers and users in various forms: customer development interviews, user feedback sessions, and even exit interviews. Always seeking qualitative insights to better understand what drives user behavior.
While I genuinely enjoy these conversations, I’ve found that conducting them regularly can be surprisingly challenging. Scheduling calls, coordinating calendars, and dealing with fragmented workdays means I inevitably end up doing fewer interviews than I should.
I strongly believe there’s no true substitute for founders personally speaking to customers. Yet, the reality is that time constraints limit how often these interactions happen, especially for less critical (but still important) research tasks. This dilemma sparked my curiosity: Could AI potentially handle some parts of this process?
Another motivation behind this project is my passion for building things. With AI now making coding incredibly fast and enjoyable, I wanted to explore developing a meaningful AI-driven product. While I have experimented casually with AI agents, I haven’t built something like this before.
Thus, I created a simple MVP of this concept and called it SelkoDialog. The name comes from “Selkeä”, meaning “clear” in Finnish, so “SelkoDialog” translates roughly as “clear dialogue.” It felt fitting since the tool’s main purpose is to facilitate clear, insightful conversations through AI agents. Admittedly, the naming was partly influenced by finding an available “.com” domain, which is quite hard these days. But I’m happy with it.
SelkoDialog is designed to complement traditional user interviews by enabling what I call “AI mass interviewing.” It conducts open-ended conversations with users, asks meaningful follow-up questions, and explores deeper motivations much like a human would. These interviews happen asynchronously, anytime the user is available, removing the need for scheduling altogether. And since transcripts and summaries are generated automatically, it’s easy to quickly review findings without losing the nuance of each conversation.
We recently tested this MVP with Props, where I’m a co-founder and which is my main focus these days. Props is a rewards app, and we wanted to gain deeper insights from churned users. I wrote a case study on how we used SelkoDialog at Props to uncover hidden reasons behind user churn.
Overall, this early test suggests there’s real potential here. I plan to continue developing SelkoDialog as a side project and am excited to see where this journey takes me. Stay tuned.
