EVErica Vaughn

Case study 07 · AI adoption · Learning platform

A.I. University

An internal learning platform I conceived and built to teach employees, from the C-suite to every Microsoft 365 Copilot license holder, how to use AI well in their own day-to-day work.

01 · The challenge

Licenses in hand. No path to using them well.

The company was handing out Copilot licenses, but there was no structured way to build prompting skills, ground them in the company’s own work, or measure how ready the organization was to use AI well.

02 · What I did

What I owned and built.

  • Conceived and led the initiative to advance AI adoption, and designed the platform end to end in the company’s brand system.
  • Built the AI Prompting Course: four modules (fundamentals, being clear and specific, role and context setting, advanced techniques) with real examples from Sales, Analytics, Demand Planning, Operations, Marketing, IT, Product, and HR.
  • Built a 16-question Copilot readiness assessment, locked after completion to keep results clean, with completion indicators in the sidebar.
  • Implemented persistent progress tracking with OAuth-integrated load and save, designed so database errors degrade gracefully instead of breaking the experience.
  • Solved real-user identity capture in Databricks Apps using the forwarded-user header instead of placeholder data, and fixed production deployment failures with a diagnostic script.
  • Wrote executive-ready rollout communications and reminders to drive completion.

03 · Architecture

How it fits together.

Employee sign-in
SSO
Databricks App (Streamlit)
Identity via forwarded-user header
Courses + readiness assessment
Progress tracking
Unity Catalog table · graceful degradation
Leadership readiness baseline

04 · Key decisions

The judgment behind the build.

Why ground lessons in real department work?

Generic prompting tips don’t stick. Examples from each person’s own function make the skill immediately usable.

Why lock the assessment?

Retakes would muddy the baseline. Locking after completion keeps the organizational readiness data honest.

Why degrade gracefully?

A learning experience that crashes on a database hiccup loses people. Progress saving fails soft, and the lesson keeps going.

Why build it in-house?

Owning the platform means it can grow into champion programs and role-based paths on the company’s own data platform.

05 · Impact

What changed because of the work.

4course modules
16question readiness assessment
8departments in the examples
  • Every Copilot license holder gets a self-paced path to using AI well, tied to their own department’s work.
  • The readiness assessment gives leadership a baseline view of organizational AI readiness.
  • Creates a platform that can expand into an AI Champion program and role-based learning paths.

06 · What’s next

Where it goes from here.

Add videos and quizzes based on user feedback, then launch courses on AI agents, agent requests, and agent building.

This case study is a sanitized reconstruction. Organization names, proprietary code, data, and internal schemas are intentionally left out.