Please don't hesitate to reach out. Also, feel free to check out my other links.
TL;DR: AI is genuinely good at the tedious part of research — clustering large volumes of qualitative feedback into themes — as long as you keep the human in the loop for interpretation. Use it to find patterns fast, then validate the themes against the raw quotes yourself.
Synthesis is turning messy raw input — interview transcripts, open-ended survey answers, support tickets — into a small set of insights you can act on. It's slow, and it's where bias creeps in. AI helps with the volume; you supply the judgment.
Paste your notes and ask the AI to group them into themes, with a count of how often each appears. This gives you a fast first map of what users are actually saying.
Ask it to sort themes by how often they appear and how painful each problem sounds. This separates loud-but-rare complaints from quiet-but-common friction.
Explicitly ask for quotes that contradict the majority. Outliers often point to a segment you're underserving — the opposite of what an averaging summary would show.
Before you trust a theme, click back to the actual quotes behind it. AI can over-generalize or invent a tidy narrative; the raw evidence keeps you honest.
Will AI misrepresent my data? It can, if you let it summarize without checking. Always trace themes back to source quotes.
Does this replace a researcher? No — it speeds up clustering, but deciding what the insights mean is human work.
Carlos Lastres is an Apple Design Award–winning product designer and software engineer based in Tokyo who works hands-on with AI tools to design conversion-focused products.