AI-Assisted UX Research: How SketchMate Cut Interview Analysis Time by 70%

AI · 5 min read

SketchMate faced slow turnaround on qualitative research: interview synthesis took up to a week because researchers manually transcribed, tagged, and clustered notes. They piloted an AI-assisted workflow that used speech-to-text, automated thematic extraction, and a summarization layer that produced candidate insight documents for researcher review.

Critical to adoption were explicit guardrails: researchers retained final editorial control, AI outputs were labeled with confidence scores, and sensitive content underwent an extra manual pass. The team also implemented a validation loop where AI-generated themes were cross-checked against a sample of manually coded transcripts to measure precision and recall.

After three months, the pipeline reduced synthesis time by 70% and allowed product teams to act on insights within 48 hours of interviews. The AI-assisted summaries surfaced overlooked micro-themes that led to two rapid UX fixes; however, researchers emphasized the need for continued human oversight to catch nuance and to prevent amplification of sampling bias.