AI Agents at Work: Advanced
Design, deploy and govern agent workflows for your team or organization
AI Agents at Work: Advanced is for people who design, deploy or govern AI agent workflows for a team or organization. It builds on AI Agents at Work: Foundations and Practitioner, and goes further: from a single well-specified agent to agent workflows that are tested, secured, monitored, governed and adopted at scale.
Throughout the course you'll apply the ETA Agent Autonomy Ladder (Assistant, Copilot, Task agent, Workflow agent, Autonomous agent) action by action, and use ADAPT (Assess, Direct, Assure, Protect, Tune) to structure your decisions. The rule you'll practice in every module: climb only as high as the risk, the reversibility and your oversight allow.
What you'll learn
- Choose processes for agents using five lenses: volume, variability, risk, reversibility and ROI
- Design workflows with approval gates, escalation, hand-offs and, when justified, multiple agents
- Write a production-quality agent specification with per-action rungs, limits and change control
- Compare agent platforms (as of late 2026) on governance criteria, and decide when to configure, assemble or build
- Build evaluation sets, define launch criteria, and monitor agents in production with logs and alerts
- Manage agent identity and access, and threat-model workflows using the OWASP Top 10 for LLM Applications and the OWASP Top 10 for Agentic Applications
- Run agent incident response and organize governance around the NIST AI Risk Management Framework
- Lead adoption, redesign roles responsibly, measure value honestly and brief leadership
Who it's for
Team leads, operations and process owners, IT and digital transformation staff, risk, compliance and security professionals, and managers in any sector who are responsible for agent workflows. You should have completed AI Agents at Work: Practitioner or have equivalent experience with agent specs, connectors, least privilege and human checkpoints.
How it works
Four modules, about 180 minutes, 3 credit hours. Each lesson includes workplace examples from different industries, weak-versus-strong comparisons, job aids, and a hands-on exercise in your approved AI tool. Each module ends with a knowledge check. You'll finish with a capstone, an agent workflow specification and governance plan for a real process, and a final assessment.