AI Insights Hub

Problem

Morningstar was investing significantly across multiple AI platforms — M365 Copilot, GitHub Copilot, ChatGPT Enterprise, Claude, Gemini, AWS Bedrock, and Copilot Studio — with no unified way to measure adoption, engagement, or return on that investment. Leadership needed a single view of what was being used, by whom, and whether it was moving the needle. The data lived in seven different places, in different formats, with no consistent definitions of what 'active' or 'engaged' even meant.

Approach

• Established consistent metric definitions across all seven tools — standardizing what counts as an active user vs. an engaged user regardless of platform.
• Built a Power BI data model that ingests telemetry from Microsoft Viva Insights (M365 Copilot), GitHub, and internal system APIs, normalizing across different granularities and reporting cadences.
• Developed a Python + Claude Code automation pipeline to generate formatted weekly AI briefings for senior leadership from multi-source data, reducing manual reporting effort.
• Integrated Workday org data via Power Automate to keep cohort definitions (business unit, geography, seniority band) current without manual exports.
• Designed the reporting experience specifically for an executive audience — summary KPIs first, drill-down available, commentary embedded directly in the report.
• Extended the architecture with an AWS Lambda → S3 → Athena pipeline to bring in data from tools outside the Microsoft ecosystem, future-proofing the hub.

Tech stack

Impact

• Delivered quarterly to senior IT leadership and the CEO, covering 12,000+ employees across a global organization.
• Unified reporting across seven AI tools in a single dashboard — previously tracked, if at all, in separate spreadsheets or not tracked.
• Enabled data-driven conversations about AI ROI at the executive level for the first time, replacing anecdote with trend lines.
• Automated weekly executive briefing generation, significantly reducing manual reporting time.

Screenshots