Reasoning Models Workshop

Welcome to the Reasoning Models Workshop!

Reasoning models represent the next step in generative AI. While traditional LLMs like GPT excel at generating fluent responses, reasoning models are designed to:

• Think through problems step by step
• Follow multi-step logic
• Deliver structured, explainable answers

Why does this matter?

In industries such as finance, insurance, healthcare, and law, it's not enough for AI to sound smart—it must:

• Reason through complexity
• Support real decisions
• Trace and explain its logic

That's where reasoning models come in.

This workshop is designed to give you both understanding and hands-on experience. Whether you're from a low-code, code-first, or no-code background, you'll gain the tools and intuition to build powerful AI solutions that don't just talk—but think.

You'll explore reasoning models through interactive Jupyter notebooks and real-world use cases, learning how to leverage their capabilities for complex analytical tasks and decision-making scenarios.

Workshop Structure

The workshop is organized into sections:

Getting Started

1. Setup & Environment - Learn how to set up your Azure OpenAI environment and configure the necessary credentials.
2. Basic Text Reasoning - Explore fundamental concepts of reasoning models using simple text-based examples.
3. Advanced Features - Dive into advanced capabilities including:
   • Developer Messages - Set reasoning goals and context
   • Structured Outputs - Generate formatted responses
   • Function Calling - Integrate with external tools
   • Vision Support - Process and analyze images
4. Model Comparison - Compare outputs between GPT and Reasoning models across different tasks.

Use Cases

Explore practical applications of reasoning models across various domains:

• Credit Risk Assessment and Management
• Customer Relationship Management
• Data Analysis and Insights
• Fraud Detection and Prevention
• Insurance Claims Processing
• Insurance Plan Analysis
• Loan Agreement Analysis
• Market Sentiment Analysis
• Portfolio Optimization
• Risk Assessment for Underwriting
• Underwriting Analysis

Key Features Covered

The notebooks demonstrate key capabilities of reasoning models:

Feature	Description
Developer Messages	Set reasoning goals, persona, and context
Structured Outputs	Generate strictly formatted responses
Context Windows	Handle large input and output contexts
Reasoning Effort	Control the depth of analysis
Vision Support	Process and analyze images
Function Calling	Integrate with external tools and APIs

Prerequisites

1. Azure Account with OpenAI access
2. Python environment with required packages:
   • openai
   • azure-identity
   • python-dotenv
3. Basic understanding of:
   • Python programming
   • Jupyter notebooks
   • Machine learning concepts

The notebooks are self-contained with detailed explanations and examples to help you understand each concept thoroughly.

1. Walk away with a sandbox (and notebooks) that you can use, to experiment with application ideas, or to explore advanced reasoning concepts.
2. Star or watch the repo for updates. We'll continue to add new labs and scenarios to keep pace with the fast-growing set of models and capabilities for reasoning.

Questions or Comments?

1. File an issue. We welcome feedback, bug reports, and suggestions to improve the workshop.
2. Visit the GitHub repository to star, watch, or contribute to the project.