Enter your search text in the box above

Select a result to preview

Talk to Your Data: The Future of Data Interaction with AI

#fabcon2026 #session #learning #microsoft #fabcon2026-session

Description

In the evolving world of data, the ability to interact with your datasets and databases using natural language is a game-changer. This session will introduce you to the power of Azure OpenAI, Azure SQL, and Microsoft Fabric, showing you how to break down barriers that have traditionally required.

Key Takeaways

My Notes

Action Items

Slides

📥 Download Slides

Talk to Your Data: The
Future of Data
Interaction with AI
Me Myself and I
• DBA To MCE
• Junior SQL DBA for Next (UK)
• SQL DBA (Albelli)
• MCE (SBP)
• Azure Heroes Organizer NL
• MVP
• Redgate Community
• 2 kids, live in Laren NL
https://www.linkedin.com/in/danny-kruge/
Content
• Introduction & Data Challenge
• Solution Overview
• Demo
• Use Cases & Best Practices
• Wrap up
The Data Challenge Today
• The Traditional way to interact with Databases?
T-SQL, Python, Excel (of course) plus many
more….
• Barriers for business users & non-technical
professionals
Waiting for Data Team potentially
Reports, Queries need changing..
• The need for natural language interactions
Unblock the above
Business can talk to their Data
What If You Could Just Ask?
• Concept of Conversational Data Access
How AI is transforming data interaction
• How AI + SQL enables chat-based data
queries
• Connecting various datasets without
writing SQL
What is Azure AI Search & How It Works with SQL?
Reference Architecture
Stop ….. Its AI Search Demo Time
Stop ….. Its Demo Time
• What was covered?
• Demo against a dataset
• No Code setup
• The setup between the resources
• Azure AI search setup – Show settings, indexes
• Open AI Setup – change response style and setup data source
• Azure SQL DB setup
Example
My Database is Slow?
Use Cases
• Intelligent Customer Support System:
1.Setup: Azure SQL can store customer data, interaction history, and product information, while
Azure OpenAI can power an AI-driven chatbot.
2.Functionality: The chatbot uses stored customer data to provide personalized support,
answer queries, and process requests. Natural language processing ensures conversational
responses, and SQL queries fetch real-time data (e.g., order status, transaction details).
Use Cases
• AI-Powered Revenue Forecasting and Insights
• Setup:
Azure SQL stores historical sales, customer, and financial data, while Azure OpenAI analyzes the
data to generate predictive insights.
• Functionality:
AI identifies trends and patterns to forecast next-quarter revenue. Business users can ask natural
language questions and receive data-driven insights to support strategic planning and decisionmaking.
Use Cases
• AI-Powered Database Performance Insights
• Setup:
Azure SQL stores performance metrics such as query execution times, resource usage, and workload
patterns. Azure OpenAI analyzes this operational data to detect inefficiencies.
• Functionality:
AI identifies slow queries, performance bottlenecks, and unusual workload patterns. It provides
recommendations to optimize queries, improve indexing, and enhance overall database performance
Data Cleansing & Quality Enhancement
Setup Azure OpenAI & Azure SQL Prerequisites
• Azure OpenAI Service (Access to GPT models)
• Azure SQL Database (To store cleaned data)
• Azure Functions (To automate the process)
Data Cleansing & Quality Enhancement
Common Data Issues
• Typos & Misspellings (e.g., “Jhon” → “John”)
• Inconsistent Formatting (e.g., “CA” vs. “California”)
• Duplicate Entries
• Missing Data (Fill with contextual information)
Define a Function to Clean Data
def clean_data(input_text):
prompt = f"Clean and correct the following data: {input_text}. Ensure proper spelling,
formatting, standardization. Make no Mistakes please please please"
response = openai.ChatCompletion.create(
engine="gpt-4",
messages=[{"role": "system", "content": "You are a data cleaning assistant."},
{"role": "user", "content": prompt}],
max_tokens=100
) return response["choices"][0]["message"]["content"]
Data Cleansing & Quality Enhancement
raw_data = "jhon do, 123 main str, ca"
cleaned_data = clean_data(raw_data)
print(cleaned_data) # Output: "John Doe, 123 Main Street,
California"
cursor.execute("INSERT INTO Customers (Name, Address) VALUES
(?, ?)", cleaned_data.split(", ")[0], cleaned_data.split(", ")[1])
conn.commit()
Data Cleansing & Quality Enhancement
Would I Do this?
Spelling mistakes is a hard one, other
quality gates could be handled by the
application
This is not batch, Row by row.
Its pretty cool though!
More is to Come…
Walk Through
Stop ….. Its RAG Demo Time
Do you trust it?
Fabric?
My Opinion
Utilize the tools but do not rely on the tool
Meaning,
Analysis should be done by professionals, and AI can read this analysis
and provide it to the user.
For example, Financial data.
Tools Advancement / Agents
• VS code – github copilot
• SSMS Copilot + github copilot
• Copilot
• Data Agent
• MCP
• SQL Server 2025
MCP - MSSQL
https://github.com/Aaronontheweb/mssql-mcp
• Abstract
• Why does this exist? Because the other MCP solutions in market for this
are generally janky pieces of s*** that don't work - certainly not on
Windows.
Stop ….. Its MCP / SQL 2025 Demo Time
Wrap up
• Data Challenges
• What if you could just ask?
• Reference Architecture
• Demo and Examples
• Use cases
• Potential Data Cleansing
mechanism
• AI Skills in Fabric / Data Agents
Thank you!
Sound off.
The mic is all yours.
Influence the product roadmap.
Join the Fabric User Panel
Join the SQL User Panel
Share your feedback directly with our
Fabric product group and researchers.
Influence our SQL roadmap and ensure
it meets your real-life needs
https://aka.ms/JoinFabricUserPanel
https://aka.ms/JoinSQLUserPanel