Short CourseBeginner

Governing AI Agents

Instructor: Amber Roberts

Databricks logo
  • Beginner
  • 9 Video Lessons
  • Instructor: Amber Roberts

What you'll learn

  • Explore the four pillars of agent governance: lifecycle management, risk management, security, and observability.

  • Apply governance to a data analysis agent: restrict the agent’s access to sensitive data and grant it access to only the data it needs.

  • Add observability to track the agent’s inputs, outputs, and decisions to ensure transparency and enable debugging.

About this course

Introducing Governing AI Agents, a short course built in collaboration with Databricks and taught by Amber Roberts.

As AI agents autonomously access larger and more sensitive data, governance becomes essential. Without proper controls, an agent can accidentally expose personal information, modify sensitive records, or operate beyond its intended scope. As a developer, you need to design agents that are not only capable but also safe, compliant, and observable in production.

In this course, you’ll learn how to integrate governance into every stage of your agent’s lifecycle, from defining access control to monitoring runtime behavior.

Explore what it means to govern an agent, how to apply governance policies to a real dataset in Databricks, and how to add observability to track and debug performance. Through this course, you’ll know how to build agents that handle data responsibly while maintaining visibility, and safety.

Skills you will build:

  • Apply the four pillars of agent governance (lifecycle management, risk management, security, and observability) to build safer, production-ready agents.
  • Use Unity Catalog, Databricks’ centralized governance layer, to organize data, manage permissions, and enforce least-privilege data access for your agents.
  • Manage data permissions for Databricks identities and assign your agent an identity with appropriate access.
  • Apply governance to an agent analyzing an HR dataset: create anonymized views, mask personal information, and build tools that provide only the data needed.
  • Build, evaluate, and prepare your agent for production using MLflow to log, version, and deploy it with proper governance.
  • Deploy your governed agent with a secure, traceable endpoint in Databricks.

By applying these governance practices to your own agents, you’ll build observable systems that handle data securely!

Who should join?

Any AI Builder who wants to ensure their agents handle data safely and transparently. Basic familiarity with Python is recommended to make the most of this course.

Course Outline

9 Lessons・0 Code Examples
  • Introduction

    Video・3 mins
  • What is AI Agent Governance?

    Video・6 mins
  • Establishing Governance in Catalogs

    Video・5 mins
  • Adding Identity to an Agent

    Video・4 mins
  • Links & Resources

    Reading・1 min
  • Lab 1: Building the Governance Foundation

    Video・22 mins
  • Building, Evaluating & Deploying Agents

    Video・8 mins
  • Lab 2: Building the HR Analytics Agent

    Video・8 mins
  • Lab 3: Evaluating and Deploying the HR Analytics Agent

    Video・15 mins
  • Conclusion

    Video・3 mins
  • Quiz

    Graded・Quiz

    ・10 mins

Instructor

Amber Roberts

Amber Roberts

Course access is free for a limited time during the DeepLearning.AI learning platform beta!

Want to learn more about Generative AI?

Keep learning with updates on curated AI news, courses, and events, as well as Andrew’s thoughts from DeepLearning.AI!