Case study · 01
Rundown University
A hands-on workshop for The Rundown University on launching a personal AI agent with OpenClaw, from how models work to scheduled automations.
300+
11
2h 23m
What we taught
For The Rundown University, Michael Shimeles taught a workshop on setting up a personal AI agent with OpenClaw, from scratch. More than 300 people attended. It starts with how language models work and ends with an agent that has its own tools, memory, and routines, connected to the services people already use.
The workshop is now an on-demand course, Launch Your OpenClaw from Scratch: 11 chapters and 2 hours 23 minutes, built for beginners.
What it covers
- Agent basics
- How language models use tokens, embeddings, context, and prediction, so prompts and workflows are designed around how the model actually works.
- Prompting
- Concise instructions that define the task, the constraints, the process, and the definition of done, so the agent behaves reliably.
- Setup
- Deploying OpenClaw, connecting model access and a way to talk to it, and configuring the core files that shape its identity, tools, memory, and routines.
- Tools
- Connecting services such as Notion, Trello, Google Calendar, search, and email, then checking each one with real actions.
- Automation
- Skills, subagents, and scheduled automations that keep context and handle repeatable work without blocking the main agent.
- Refinement
- Recursive skill refinement, reverse prompting, memory, and lean context management, so the agent keeps getting better after setup.
Why it matters
Setting up an agent is the easy part. Keeping it useful takes clear instructions, tools tested against real work, and routines that improve over time. The workshop teaches that whole loop to people starting from zero, and it's the same hands-on approach we bring to training engineering and operations teams.