AI Series | Session 3: Advanced| Virtual | Host Chapter: Suncoast

Why a Tiered Series?

AI sessions can feel too intimidating for beginners or too basic for experienced users. This 3-part series is designed with clear levels, allowing you to join the session that suits your comfort and skill level.

This advanced session is for L&D professionals who want a clear, practical understanding of how AI-powered practice and coaching experiences are designed and improved in real-world environments.

Jon Meisburg, a L&D leader who uses AI, automation, and experience design to make training faster, smarter, and more human, will start this session by demonstrating a fully working AI-powered customer simulator, showing the end-to-end learner experience from live practice to coaching feedback.
He will then walk through how the solution was first built as a GPT-based simulator, and explain why and how it later evolved into a more custom Realtime API implementation.

The session focuses on the inquiry-driven process behind the build.
Participants will see examples of the questions asked of an AI assistant, how those questions lead to draft code or configuration, how the results are tested, and how the system is refined over time.

The emphasis is on understanding design decisions, tradeoffs, and workflow across both GPT-based and Realtime API-based approaches.
Participants will leave with a clear mental model for how to prototype, iterate, and evaluate AI-powered practice systems, and how to choose the right approach for their needs.

✨ What's In It For You?

• Understand a fully working AI-powered practice simulator and how its components connect from learner interaction to coaching output.
• Break down a production-ready GPT-based simulator, including how role definition, guardrails, phase logic, and evaluation criteria support realistic practice and consistent coaching.
• Understand how the same design was extended into a Realtime API solution, including key building blocks such as the front end experience, session service, scenario and guardrails management, transcript capture, and coaching evaluation.
• Decide when to use a GPT-based approach versus a Realtime API solution based on learner workflow, integration needs, scale, and cost.
• Follow a high-level inquiry sequence with an AI assistant to see how ideas become draft implementations that are tested and refined over time.

• A clear picture of how a production-ready AI-powered practice simulator is structured from learner experience through coaching feedback.
• A concrete example of how a GPT-based simulator can be designed using clear instructions, guardrails, and evaluation logic.
• A high-level understanding of how that same design can be extended into a Realtime API solution, including the major front end and back end components.
• A reusable set of inquiry prompts and example chat logs that show how to move from ideas to draft implementations, test them, and improve them iteratively.
• A reusable set of inquiry prompts and example chat logs that show how to move from ideas to draft implementations, test them, and improve them iteratively.
• Practical guidance for choosing between a GPT-based approach and a Realtime API solution based on workflow, integration needs, scale, and cost considerations.