Overview of AI Insights

5 mins to complete

Course Creators:

Lizzy Burnam

Paolo Appley

Last Updated:

March 5, 2026

User Interviews was built to remove friction from research logistics. With AI Assistant, we’re extending that same philosophy into research execution and analysis—helping you move faster while keeping rigor, transparency, and human judgment at the center.

This course introduces AI Insights & Analysis and shows you how to use it confidently and responsibly across your research workflow. In this first lesson, we’ll cover:

The problems that AI Assistant was designed to solve
What AI Assistant does and doesn’t do
Which User Interviews features are included in AI Assistant

Why we designed AI Assistant

From our research with customers, we consistently heard three things:

Analysis fatigue is real: Researchers are generally happy with their outcomes, but the process is exhausting and slow.
Manual comparison doesn’t scale: Spreadsheets, sticky notes, and copy-pasting break down as projects grow.
Tool sprawl breaks context: Bouncing between recording tools, AI tools, spreadsheets, and repositories slows teams down.

AI Insights reduces that friction by keeping analysis close to the session data—where context is strongest.

The AI-assisted research pyramid

By combining world-class AI with research-specific engineering, User Interviews allows you to harness the power of AI without needing to be an AI expert. Below is the AI-Assisted Research Pyramid, which can help you visualize the technical foundations of our AI Assistant.

‍

‍

The foundation: Best-in-class LLMs

We’ve chosen the most capable AI models available (Claude, Open AI, etc) for their quality, safety, and fit. Note that we do not let the LLMs train using customer data.

World-class reasoning: Advanced synthesis, nuanced understanding
Broad training: Deep knowledge across domains
Long context: Process full transcripts without cutting them short
Safety & alignment: Built-in guardrails, responsible AI practices
Continuous improvement: Models get better; you benefit automatically
Tool use: Can invoke specialized capabilities when needed

As the AI landscape evolves, we’ll adopt what works best for research.

Research-specific tools

Our AI Assistant has custom capabilities built for qualitative workflows, including:

Cross-transcript search: Find patterns across all interviews
Quote extraction: Pull exact quotes from individual sessions
Topic-based observations: Auto-extract insights per topic
Speaker recognition: Distinguish researchers from participants

As with all of our products, we’ll continue to add more capabilities over time based on the researchers’ evolving needs.

Research methodology

At the next level of the technical pyramid is methodology: custom instructions that guide the AI to think like a qualitative researcher.

Quote-grounding: Claims tied to real words
Citation standards: Timestamped references
Synthesis patterns: Theme ID, comparison, gaps
Discussion guide awareness: Your structure honored
Participant voice: Respects how people actually speak

Research insights

At the top of the pyramid—powered by best-in-class AI, research-specific tools, and research methodology—are the actual insights and outcomes of your research:

Themes and patterns across participants
Every claim backed by actual participant quotes
Synthesis you can trust and verify

What AI Insights does (and doesn’t do)

AI Insights helps you:

Organize conversations into structured observations
Compare participants side by side
Surface evidence faster
Share insights more clearly
Verify accuracy and context with links back to real participant quotes

AI Insights does not:

Replace research expertise
Decide what insights matter
Produce conclusions without review

AI Insights is designed as an open, flexible analysis layer, not a replacement for the tools you already rely on. Over time, we’ll continue building ways for AI Insights to play nicely with other tools, whether through deeper integrations or additional export and sharing capabilities. The goal is to add value where analysis is most time-consuming—without disrupting your broader research ecosystem. You’re in control at every step!