Thursday 10:50 AM - 12:00 PM · Room 1

From Prompt to Production: Building Responsible Gen-AI Apps with Azure AI Studio

Daniele Mario Areddu

AI & Big Data Enthusiast| Freelance Full-Stack Developer| Student @ ITIS Vallauri Velletri | Digital Culture & Change Model Specialist @ UNIDEA | Indie Author @ KDP ! Community Lead @ Velletri-Dev

Azure AI Studio, now generally available with access to 1,600+ foundation and open-source models, lets Microsoft 365 and Power Platform professionals go from a clever prompt to an enterprise-ready Copilot in hours, not months. In this demo-heavy session you’ll learn how to:

Design multimodal prompts and chains in Prompt Flow.

Ground large-language models with your SharePoint or Dataverse data using Retrieval-Augmented Generation (RAG).

Govern outputs with built-in Content Safety, evaluations and cost dashboards.

Deploy as an Azure function, a Teams app, or a Power Automate connector—then surface the same logic inside Microsoft 365 Copilot.

We’ll walk through a live scenario that turns an internal policy PDF library into a secure “Ask HR” Copilot, highlighting pitfalls (token limits, latency) and best-practice mitigations. You’ll leave with a GitHub starter repo, a cost estimator sheet, and a checklist for responsible rollout. Microsoft

Key Takeaways

How Azure AI Studio’s Prompt Flow accelerates prototyping and A/B testing

Implementing RAG with Azure AI Search in three steps

Mapping governance features (Content Safety, evaluations) to your compliance needs

Packaging and sharing your Copilot inside Teams, Power Apps, or Outlook