Agentic AI track

What I learn about building AI agents, and what I teach back. Based on Andrew Ng's Agentic AI course on DeepLearning.AI.

I add a lesson each time I learn something new, and I mark every change in the lesson's update log. Lessons marked EN | DE are available in English and German.

  1. Lesson 1 of 6 · · EN | DE

    What Is Agentic AI? From One Prompt to a Workflow

    Why one prompt is often not enough, what an agentic workflow does, and the first agent I designed: a job matcher.

    Read in English →Auf Deutsch lesen →

  2. Lesson 2 of 6 · · EN | DE

    Workflow or Agent? The Spectrum of Autonomy

    From fixed steps to agents that choose their own tools.

    Read in English →Auf Deutsch lesen →

  3. Lesson 3 of 6 · · EN | DE

    The Four Design Patterns

    Reflection, tool use, planning and multi-agent collaboration.

    Read in English →Auf Deutsch lesen →

  4. Lesson 4 of 6 · Coming soon

    Task Decomposition: From a Goal to Steps

    How to get from "find me a job" to six clear steps.

  5. Lesson 5 of 6 · Coming soon

    Evaluating an Agent

    Is 70 the right cut-off? Does the same job get the same score twice?

  6. Lesson 6 of 6 · Coming soon

    Building the Job-Match Agent, Step by Step

    The build log, including what goes wrong.