This workshop focuses on designing and prototyping an AI agent to complete a task in your daily workflow. Rather than focusing on AI features or basic prompting, you’ll create a working tool that supports a defined talent development process.

Design AI Agents for Business and Learning Workflows

To start, you will identify an actual L&D workflow and document the current process, establishing a baseline for the actual time, steps, or error rate for that task. From there, you’ll design your custom agent in portable, tool-agnostic formats. This design-first approach makes you the architect of the solution rather than the operator of one vendor’s interface: your work carries over to whatever platform your organization adopts.

Prototype a Custom AI Agent for a Real L&D Task

In the second half of the course, you’ll conduct your AI agent training—translating your design into a working agent prototype, adding a memory pattern that tracks what’s already been gathered across conversations, and stress testing it before deployment.

At the end of the session, you will remeasure the same task to quantify impact and document what remains human: the judgment, accountability, and edge cases you continue to own as the workflow orchestrator.

Test the Agent for Accuracy and Privacy

Because learning teams often work with proprietary materials and participant data, the workshop addresses security, privacy, and knowledge-base design throughout the development process.

You’ll leave with a secure, working, IT-ready agent prototype, clear before-and-after results, a reusable 5-step blueprint for building future agents, and a business case ready to bring to your leadership.

Some participants may need approval to deploy an AI agent within their organization. This workshop focuses on designing and prototyping a robust assistant, with the option to build an agent for those who have the necessary tools and permissions. At the end of the program, you will leave with the tools and business case necessary to upgrade your assistant to an agent once approval is granted.

Access to a paid artificial intelligence large language model (LLM) on your computer is required for this course. Acceptable tools include ChatGPT, Gemini, Claude or Microsoft Copilot with access to GPTs in ChatGPT, Gems in Gemini, or Agents in Microsoft Copilot. Trial versions may not include the necessary capabilities. If you’re using an employer-provided laptop, please confirm you can access both the tool and its custom agent features.