AURA — AI Research Intelligence Platform

Impact:

• Research cycle: 3-4 weeks → 5-7 days
• Insights reuse: +60% across teams
• Democratized research: PMs, Support, Ops now launch studies independently

The Challenge

The organization generated substantial user knowledge, but in a fragmented way. Each team had its own approach to interviewing, note-taking, and sharing insights — making research slow, siloed, and hard to reuse.

Key Constraints

• No dedicated PM for the design workstream — I took ownership of translating research needs into product requirements: writing user stories, facilitating alignment, and defining the UX roadmap alongside the research team.
• AI skepticism — teams worried AI would “black box” insights. Solution: every summary links to source quotes, every theme shows evidence.
• 2 engineers part-time — we shipped MVP in 6 weeks (transcription + summarization + basic repository) instead of waiting 6 months for the “complete” product.
• Cultural resistance — teams had their own workflows. Solution: AURA integrates with existing tools; it enhances, it doesn’t replace.

Approach & Decisions

Working closely with the User Research team, we started with one question: what do internal teams need to trust AI insights day-to-day? Conversations with PMs, Support, and Customer Success revealed three needs: transparency in how insights are generated, accessible knowledge (not trapped in documents), and speed without sacrificing context.

Key decision #1. The project was initially scoped as a “transcription tool.” Research with internal users showed transcription was table stakes — the real value was connecting insights across studies. The team pivoted to a repository-first experience. I championed this shift from a UX perspective, designing how a living repository could surface patterns across past research in an intuitive way.

Key decision #2. Instead of presenting AI as a “magic” layer, I designed small moments of contextual help — suggested next steps, meeting summaries, cross-study theme connections. The principle: AI should feel helpful, not mysterious. This design direction shaped how the entire team thought about AI integration.

The Solution

AURA starts with a dashboard showing active studies, recent insights, and documented decisions. Launching a new study is a guided flow — define objective, participant type, and channel, while AI suggests questions and formats.

Once sessions are recorded, AI handles transcription, summarization, and theme grouping — offering views adapted by user type: executive summaries for leadership, granular detail with quotes for researchers. I designed the information architecture so users never lose context: every insight traces back to its source, every decision links to the learnings that justify it.

Results

• Research cycle reduced from 3-4 weeks to 5-7 days
• Insights reuse increased ~60% with a centralized AI repository
• Non-researchers (PMs, Support, Ops) now launch studies independently
• Teams stopped repeating studies and started building on past learnings

What I Learned

• Trust over magic — showing AI sources and confidence levels drove adoption more than any feature.
• 10x value or die — cutting research from weeks to days was the only threshold that mattered.
• Cultural change beats interface design — 30% of the project was workshops and demos, not screens.