Mentoring AI in Research - Part 1a: From Interviews to Insights

null null
Oct 23, 2025
3 min read

Part of the “Mentoring AI” series - guiding AI like a junior teammate across the UX process.

TLDR

AI can accelerate research but it still needs guidance and mentorship to be useful and actionable. In this post, I walk through how to mentor AI during the early research phase: from raw transcripts to clustered pain points and emotional insights.

With the right prompts and structure, AI becomes your research intern... speeding up synthesis while keeping real user nuance intact.

👉 This is Part 1a of a multi-part series.

Up next: Part 1b – From Personas to Journey Maps.

Why Research with AI?

In enterprise UX, research bottlenecks are real. We’re often juggling hours of transcripts, dozens of interviews, and multiple synthesis documents before we ever reach actionable insights.

AI can’t replace real users, but it can help you move faster through the mechanics of research. When prompting AI... it can:

Transcribe interviews instantly.
Auto-group pain points into clusters.
Free up time to focus on empathy and strategy.

Slide titled “Why Research with AI?” Text explains that AI doesn’t replace research but speeds up mechanical tasks, allowing designers to focus on empathy and nuance.

PSA 1:

AI doesn't replace research but it can speed up the mechanics so you can focus on empathy and nuance.

Slide titled “The Leadership Role.” Copy reads: “AI can generate, but it’s design leadership that makes outputs actionable.” Emphasizes the need for human direction in AI workflows.

PSA 2:

AI can generate, but it’s design leadership that makes outputs actionable. You’re not delegating judgment... you’re accelerating insight.

Step 1: Transcribing Interviews

AI transcription is now baseline in our toolkit. Tools can capture live conversations and generate searchable transcripts in real time turning interviews into analyzable text within minutes.

Where AI helps: speed, accuracy, structure
Where humans matter: noticing tone, pauses, and subtle emotional shifts

Slide titled “Step 1: Transcribing Interviews.” Text notes that AI tools can capture live conversations and instantly generate searchable transcripts, speeding up early research steps.

Mentoring Moment 1:

AI records what was said... you record how it was felt. Add your own annotations to capture tone, hesitation, and emotional nuance that transcripts alone miss.

This is where empathy meets automation... your emotional notes become the context AI can’t see.

Step 2: Clustering Pain Points

Once transcripts are ready, it’s time to coach AI to cluster feedback into meaningful insights.If you simply ask it to “summarize pain points,” it’ll return vague buckets like “Confusion” or “Frustration.”

To go deeper, guide it with structure.

Slide titled “Step 2: Clustering Pain Points.” Text explains that AI can group pain points but needs coaching to go beyond vague, high-level categories to produce useful insights.

AI can group pain points but it needs coaching to move beyond vague or high-level categories.

Guide AI With Structure

Don’t just prompt AI to “summarize pain points.” Tell it how to organize what it finds by theme, emotion, and friction.

Example prompt:

You are an expert UX researcher and are particularly skilled at finding highlights and pain points that users experience just from reading through interview transcripts. Analyze the attached interview transcripts and provide highlights, supporting quotes and insights that align to our research goal: [include your goal here] For each theme, include: A clear label (e.g. Trust Gaps, Filtering Friction), a short description of the underlying issue, and a few supporting pain points from the journey (quoted or paraphrased) Focus especially on emotional friction (uncertainty, frustration, etc.) and functional friction (navigation, clarity, decision-making, etc). Assume this input is grounded in real user data and findings based on screenshots and realistic behavior.

This level of detail helps AI understand why something matters, not just what was said.

Prompt for Depth and Evidence

Once AI begins clustering, continue mentoring it toward evidence-based insights. Ask for examples, not summaries. Push it to connect emotions with actions, and behavior with context.

You’re not just teaching AI to summarize... you’re teaching it to synthesize.

Mentoring Moment 2:

Push AI for specificity and quotes... golden nuggets rarely appear on the first try. Good clustering is iterative. Each refinement builds clearer patterns between emotional friction (e.g., anxiety, confusion, etc.) and functional friction (e.g., navigation, clarity, etc.).

The Leadership Lesson

AI accelerates the mechanics... but you provide the empathy and meaning. In the research phase, think of AI as a junior researcher:

AI: transcribes, clusters, drafts.
You: refine, validate, and contextualize with real user insight.

By mentoring AI through this process, you save time on the repetitive work and spend more time advocating for users inside the enterprise.

Conclusion: Raising the Floor of Research

Research is the foundation of enterprise UX. By mentoring AI to handle the heavy lifting... transcription, clustering, and early synthesis... you raise the floor of your workflow without sacrificing quality.

Closing slide with the text “Up Next: Part 1B – From Personas to Journey Maps.” Encourages readers to save the post and continue to the next phase in the Mentoring AI series.

Disclaimer: The thoughts shared in this blog are solely my own and do not represent the perspectives of my professional relationships or clientele.