Building an AI-powered insight repository

To make customer insights more accessible, I built a centralized research repository from the ground up—organizing 100+ projects into a standardized, searchable format. Each page included the research topic, timeline, methodology, key insights, links to supporting materials, and tags for filtering by topic, product, and method.

To scale access further, I launched a custom AI Copilot in Microsoft Teams—“Research Bot”—that let team members ask any research question and instantly retrieve relevant insights with links to source materials.

This put the voice of the customer at everyone’s fingertips, reduced reliance on the research team, and enabled faster, more informed decisions.

My role:
Lead researcher

Impact:

+100 research studies now self-serviceable
+Faster, insight-driven decisions org-wide

Tools:
Confluence, Copilot Studio, Microsoft Teams

What barriers were blocking insight-driven decisions?

Challenges

  • Insights were siloed and hard to find, limiting the use of existing research across teams

  • Team members often had to manually request past research, creating bottlenecks and inefficiencies

  • There was no centralized source of truth for historical research or insight tracking

  • Raw files and findings were scattered across decks, boards, and folders with no consistent documentation or tagging

  • Teams lacked a fast, self-serve way to pull relevant insights into decision-making in real time

Previous
Previous

Building a customer feedback system with automated insights & dashboards

Next
Next

Enhancing our service bot experience through testing and user flow mapping