Agentic Specialist. Build agentic workflows.

By the end of this course, you will be able to:

Establish a clear baseline

What your workflow costs, how fast it runs and how well it performs, so every improvement can be measured.

Improve until it is reliable

Fewer errors, less rework, and the evidence trail that justifies scaling up.

Show stakeholders what is possible

A real, working agent demo you built yourself, no tech team required.

1. Module 01

   Define Success

   Lessons

   2

   Exercises

   2

   Hours total

   ~9

   Before you build, know how you'll win. Take the automation you built in AI 101—or a workflow you designed yourself—and turn gut feelings into hard numbers. Create an answer template that shows what excellence looks like, build scoring guides that measure performance objectively, and set up automated tests that catch problems before they break production. By the end, you'll have the proof system that separates real results from wishful thinking.

   What you'll learn

   • Create an answer template that defines what excellence looks like

   • Build scoring guides that measure performance objectively

   • Set minimum passing standards that gate what ships

   • Set up automated tests that catch problems before production

2. Module 02

   Improve and Iterate

   Lessons

   2

   Exercises

   2

   Hours total

   ~8.5

   Transform your single-agent automation into a specialized team where each agent does one thing brilliantly. Then use meta-prompting to let AI improve AI—while you validate what actually works. Learn to read how agents reason, not just what they output, so you fix root causes instead of patching symptoms. Build once, improve forever.

   What you'll learn

   • Decompose monolithic agents into specialized teams

   • Use meta-prompting to let AI improve AI

   • Read agent reasoning to fix root causes, not just symptoms

3. Module 03

   Build Smarter Workflows

   Lessons

   3

   Exercises

   3

   Hours total

   ~13

   One agent can't do everything—but a team can. Evaluate workflow blueprints and pick the pattern that matches your risk tolerance, speed needs, and data sensitivity. Then orchestrate communication between your agents, other systems, and humans to create a workflow machine that divides work intelligently, keeps costs down, and scales without breaking.

   What you'll learn

   • Pick the workflow pattern that fits your risk, speed, and data sensitivity

   • Orchestrate communication between agents, systems, and humans

   • Divide work intelligently to keep costs down at scale

4. Module 04

   Know What You Built

   Lessons

   4

   Exercises

   4

   Hours total

   ~16

   Most people can build a workflow. You'll master them. Create working Python-based agents from prompts, build an interface to measure real performance against realistic conditions, and analyze the generated code to understand exactly how your system operates. Debug individual agents before problems cascade, then debug agent teams to trace exactly how prompts decompose, how work gets delegated, and why agents surprise you. Go under the hood. Understand the machine. Own the system.

   What you'll learn

   • Generate working Python-based agents from prompts and analyze how they operate

   • Build an interface to measure performance against realistic conditions

   • Debug individual agents and agent teams to trace delegation and surprises

5. Module 05

   Launch with Confidence

   Lesson

   1

   Exercise

   1

   Hours total

   ~5.5

   Results without a rollout plan are just expensive demos. Package everything—your working prototype, performance data, cost projections, risk mitigation, governance checks—into a launch-ready plan stakeholders can actually act on. This is how pilots become production systems.

   What you'll learn

   • Package prototype, performance data, cost projections, risk mitigation, and governance into one plan

   • Ship a rollout plan stakeholders can act on

   • Turn pilots into production systems