If You're Writing Survey Questions, You've Already Missed the Insight

01 /The Concept

Perspective built an AI engine that conducts user interviews on behalf of product teams.

User interviews are among the most valuable inputs a product team can get. Surveys tell you what users do – but not why. Interviews can reveal not just where users are today, but where they're headed, which is often the more useful signal for deciding what to build next.

The platform ships with a catalog of ready-made interview topics: why users decided to buy, how they got started, what's missing, why engagement has dropped off, which competitors they evaluated or are currently using, how they feel about a newly launched feature, and more. Teams can also define custom interview topics tailored to their product's specific questions.

Once a topic is selected, the team provides a list of contacts from the target user segment. The AI engine reaches out automatically, opens a conversation, and asks questions designed to draw out honest, detailed answers.

All conversations are stored in a shared database that team members can query in plain language – the AI answers based on the accumulated interview transcripts. After a series of interviews on a given topic, the engine also auto-generates a summary presentation with key insights and supporting quotes from users.

The quality of the conversations is backed by data: 91% of initiated conversations reach a meaningful conclusion, 85% of participants agree to be interviewed again, and 89% of platform clients say they've gained deeper product insights than they could have gotten through any other method.

The entry-level plan charges $0.99 per interview. Professional plans start at $99/month and include custom branding, a choice of AI model, and a lower per-interview cost.

Perspective officially launched two days ago, simultaneously announcing $4M in funding.

02 /Why It Matters

Open-ended conversations with users – current, potential, and churned – are the best way to understand what actually needs to be built. Standard surveys only surface answers to questions you already know how to ask. If you already know the question, you probably already know the answer.

Traditional user research – focus groups, moderated interviews – lacks scale and regularity. Preparing and running them is slow and expensive. Rally ([covered here](/review/horoshaja-jekonomija-horosho-prodajotsja)) has raised $9.35M for a CRM-style platform designed to make that process faster and better-organized, but the core constraint remains: human-led research doesn't scale.

Another pattern worth noting: many teams treat user interviews as a customer discovery activity – something you do before launch to validate the product concept. After launch, they switch to A/B tests and analytics, treating those as a proxy for understanding users.

But users change. The competitive environment changes. The insights that mattered at launch may be completely wrong 18 months later. Continuous conversation with users isn't just a nice-to-have – it's arguably the only reliable growth driver over the long term.

Cafeteria ([related review](/review/podsadi-klienta-na-podpisku)) understood this and built a subscription model around it: a chatbot that regularly surveys a brand's users and returns trend data over time. The platform targets Gen Z audiences, who are comfortable chatting with bots. Its survey engine is simpler than Perspective's interview AI, but the subscription mechanic still earned the company $3M in a first round.

Outset ([covered previously](/review/insajty-dvigatel-biznesa)) takes the opposite approach to presence: its AI interviewer has a face, conducting interviews over video calls with a customizable avatar, voice, and language to match nearly any user. Beyond that, the product is functionally close to what Perspective has built.

And then there's Boardy ([covered here](/review/produkt-kotoryj-sam-prinosit-investorov)), which demonstrated how capable AI interviewers have become. Boardy's AI talks to users, learns about their interests and working style, then connects them with people from previous conversations who'd be a valuable professional match. The company raised $3M in an initial round in October.

What happened next was remarkable: a venture partner used the product as a regular user, had a conversation on a Friday, was so impressed that he spent the weekend convincing his partners to try it – and by Monday, the fund had decided to invest. The Boardy founder then asked the AI to identify other users on the platform who might want to invest. The result: another $8M raised in early January.

03 /Opportunities

Product teams need to be talking to users continuously – not just at launch, and not just when growth slows. Especially when things are going well, because that's usually when the next threat is forming.

AI-led interviews solve two problems at once: they scale what was previously a high-effort process, and they remove interviewer bias – a human moderator, however experienced, will unconsciously steer conversations toward confirming their existing beliefs. An AI interviewer follows the thread wherever users take it, which is how you learn genuinely surprising things.

Perspective, Outset, and Cafeteria collectively prove there's real market demand for continuous user research as a paid subscription. That's the first direction: AI platforms for ongoing, structured user interviews.

There's also a more interesting secondary effect: when a platform helps companies understand their users better, it helps those companies build better products, retain more customers, and stay in business longer – which means they keep paying for the research subscription. Product quality drives platform LTV in an unusually direct way.

The broader opportunity is deploying AI interviewers in other contexts. Boardy uses them for professional networking. Kinnect ([covered previously](/review/100-let-zarabatyvanija)) uses a simpler chatbot for a family memory app – prompting users with questions to help them capture stories they wouldn't have known how to start telling on their own.

The most underserved domains are those where the quality of questions matters as much as the quality of answers – coaching, therapy, due diligence, onboarding. In each, the interviewer's ability to read the room and follow unexpected threads is exactly what AI is now learning to do. What behavioral data would the interviewer need to get there?