On a cold, dark evening at the FedEx Institute of Technology, Memphis Data Professionals convened students, analysts, and Power BI learners for a practical working session titled “Copilot in Power BI: Rule Your Analytics Workflow.” The focus was not on what AI promises, but on how Copilot actually behaves inside real analytics workflows. What Copilot can reliably do today, where it needs human judgment, and how teams should prepare their data to get value from it.
The room reflected a familiar cross-section of the local data community: University of Memphis students seated beside working analysts and local professionals conversing over hot pizza (a Memphis tech tradition).
Session Highlights & Project Takeaways
Copilot as a Foundation, Not a Finish Line
Guest speaker DeNisha Malone grounded the room with a simple truth: Copilot is most valuable as a starting point. Through live, end‑to‑end demos, she showed how quickly Copilot can scaffold visuals, generate report pages, and reduce setup friction. While making it clear that interpretation, refinement, and decision-making still belong to analysts.
It was a message that resonated with a room full of practitioners who know speed only matters when accuracy follows.
Prompt Quality and the “Middle Ground”
One of the most practical lessons centered on prompt design. Malone emphasized that Copilot performs best when prompts avoid extremes, neither vague nor overloaded with instructions. Clear intent paired with structured data produced the most usable results, reinforcing that AI effectiveness is closely tied to human clarity.
Data Hygiene: The Real Prerequisite
Before anyone could lean on AI, Malone brought the room back to fundamentals:
Clean, labeled fields
Standardized values
Star schema modeling
Disciplined data prep
Copilot mirrored whatever structure it was given, good or bad. The takeaway was unmistakable: AI doesn’t replace foundational analytics work. It amplifies it.
From Questions to Full Report Pages
A standout moment came when Malone asked Copilot to generate entire report pages from existing data. Within seconds, visuals, filters, and written insights appeared. This was followed by live edits that showed how analysts can shape AI‑generated work into something usable.
The room leaned in as Copilot generated DAX expressions on the fly, turning abstract capability into something tangible.
Trust, Hallucinations, and Saying “I Don’t Know”
When an attendee asked about hallucinations, the conversation shifted toward trust. Malone demonstrated how Copilot often responds with “I don’t know” when context is missing. A behavior that drew nods across the room. She walked through Verified Answers and Prep Data for AI, sparking a thoughtful discussion about when AI summaries help and when human review becomes non‑negotiable.
Teaching by Showing, Not Telling
Rather than slide-heavy explanations, Malone answered most questions by building solutions live. When questions extended beyond the session’s scope, she acknowledged the limits and invited follow-up conversations afterward. This “show and tell” approach felt grounded, generous, and deeply community‑oriented. It allowed participants to see complete workflows, from raw data to finished insight.
A Community That Compounds
Before the demos even began, the room buzzed with familiar Memphis energy: people catching up on family news, comparing holiday stories, and sharing what they hoped to learn next. It was a reminder that these gatherings aren’t just technical meetups. They’re continuity.
Malone’s own story, from University of Memphis student to SQL Saturday learner to Microsoft program manager, underscored how local communities build careers over time. She closed by pointing attendees toward the Power BI Women group, the Fabric Community, and Microsoft learning resources, offering clear next steps for anyone ready to keep growing.
Resources:
Closing Reflection
What distinguished this session was not just the tooling on display, but the way the room treated learning as a shared responsibility. People arrived prepared, asked detailed questions, and stayed present through live problem-solving rather than polished conclusions.
In a space that felt organized, welcoming, and intentionally set up for work, the takeaway was clear. Memphis’ data community is less interested in chasing AI trends than in understanding how new tools earn their place: one workflow, one question, and one shared lesson at a time.