Data Meets AI: Highlights & Take‑aways from the June Minnesota Power Platform Community Call

June 21, 2025
5 min read
Contributors
Subscribe to newsletter
By subscribing you agree to with our Privacy Policy.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

A Quick Catch‑Up: Why This Call Mattered

The June meetup marked the second annual gathering of the Minnesota Power Platform Community, drawing makers, analysts, IT pros and Microsoft MVPs from the Twin Cities—and well beyond—to explore how data flows into AI‑enabled solutions.

Key stat: Attendance included participants from Minnesota, India, Egypt and more—proof that low‑code passion knows no borders.

AI—Beyond the Buzzwords

Guest speaker Adedapo Adeniran (Microsoft MVP & Data Scientist) peeled back the curtain on large‑language models (LLMs). His central point: “Everything an AI does is mathematics operating on numbers—no magic, no hidden ‘black box.’”

What this means for makers

Misconception Reality for Power Platform pros
Engineers don’t know what’s inside the model Model behavior is traceable to math & training data.
AI will steal our jobs AI creates new roles and accelerates existing ones—think prompt engineering, governance, and model monitoring.
AI “looks up” data at inference time Once trained, a model relies on learned parameters; it does not live-query its training corpus.

How Data Powers Intelligence

“A model is only a mathematical function; data is what makes it smart.” – Adedapo Adeniran

Structured, semi‑structured, unstructured—we met them all. Understanding which format your business data lives in (CSV, JSON, images, audio, etc.) is the first step toward choosing the right Power Platform or Azure AI service:

  • Dataverse & SQL – best for structured tables; great with Copilot Studio.
  • SharePoint & OneDrive – semi‑structured docs, perfect for Power Automate AI Builder forms processing.
  • Blob Storage – unstructured images/video for Computer Vision or Azure AI Search.

Demo Spotlight: Visualizing the Math

Adedapo capped the session by animating a 3‑D function to show how parameter tuning “teaches” a model. Watching the surface warp as error dropped made gradient descent feel tangible—a perfect mental model before you fire up AI Builder’s prediction control.

Five Practical Take‑Aways for Flowdevs & Friends

  1. Start with the Business Question – fancy models flop without clear outcomes.
  2. Know Your Data’s Shape – drives choice of connector, AI Builder model or Fabric workload.
  3. Prototype in Hours, Not Weeks – Power Apps + sample data + AI Builder = quick ROI.
  4. Human Oversight Is Mandatory – keep makers “in the loop” for approvals and exception handling.
  5. Saturday Sessions Work! – community voted weekends best for learning & networking.

Looking Ahead: Join the Movement

Our next meetup welcomes an in person event involving board games and low code!

Call to Action
• Follow Minnesota Power Platform Community on LinkedIn for event drops https://www.linkedin.com/groups/14688471/
• Submit a lightning‑talk idea (5 min) to showcase your Power App or dashboard. (ask on linked in)
• Share Ideas for a August workshop in Linkedin.

A Quick Catch‑Up: Why This Call Mattered

The June meetup marked the second annual gathering of the Minnesota Power Platform Community, drawing makers, analysts, IT pros and Microsoft MVPs from the Twin Cities—and well beyond—to explore how data flows into AI‑enabled solutions.

Key stat: Attendance included participants from Minnesota, India, Egypt and more—proof that low‑code passion knows no borders.

AI—Beyond the Buzzwords

Guest speaker Adedapo Adeniran (Microsoft MVP & Data Scientist) peeled back the curtain on large‑language models (LLMs). His central point: “Everything an AI does is mathematics operating on numbers—no magic, no hidden ‘black box.’”

What this means for makers

Misconception Reality for Power Platform pros
Engineers don’t know what’s inside the model Model behavior is traceable to math & training data.
AI will steal our jobs AI creates new roles and accelerates existing ones—think prompt engineering, governance, and model monitoring.
AI “looks up” data at inference time Once trained, a model relies on learned parameters; it does not live-query its training corpus.

How Data Powers Intelligence

“A model is only a mathematical function; data is what makes it smart.” – Adedapo Adeniran

Structured, semi‑structured, unstructured—we met them all. Understanding which format your business data lives in (CSV, JSON, images, audio, etc.) is the first step toward choosing the right Power Platform or Azure AI service:

  • Dataverse & SQL – best for structured tables; great with Copilot Studio.
  • SharePoint & OneDrive – semi‑structured docs, perfect for Power Automate AI Builder forms processing.
  • Blob Storage – unstructured images/video for Computer Vision or Azure AI Search.

Demo Spotlight: Visualizing the Math

Adedapo capped the session by animating a 3‑D function to show how parameter tuning “teaches” a model. Watching the surface warp as error dropped made gradient descent feel tangible—a perfect mental model before you fire up AI Builder’s prediction control.

Five Practical Take‑Aways for Flowdevs & Friends

  1. Start with the Business Question – fancy models flop without clear outcomes.
  2. Know Your Data’s Shape – drives choice of connector, AI Builder model or Fabric workload.
  3. Prototype in Hours, Not Weeks – Power Apps + sample data + AI Builder = quick ROI.
  4. Human Oversight Is Mandatory – keep makers “in the loop” for approvals and exception handling.
  5. Saturday Sessions Work! – community voted weekends best for learning & networking.

Looking Ahead: Join the Movement

Our next meetup welcomes an in person event involving board games and low code!

Call to Action
• Follow Minnesota Power Platform Community on LinkedIn for event drops https://www.linkedin.com/groups/14688471/
• Submit a lightning‑talk idea (5 min) to showcase your Power App or dashboard. (ask on linked in)
• Share Ideas for a August workshop in Linkedin.

Subscribe to newsletter
By subscribing you agree to with our Privacy Policy.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Share
More

Related Blog Posts