The class
Two hours. One useful decision at a time.
This is a working lunch, not a software demo. The room uses your real recurring work to decide where an AI agent helps, where it creates risk, and how to start safely.
What you leave with
Decisions, not inspiration.
- 01
A map of your own week
Your recurring work sorted into what an agent can carry now, what still needs judgment, and what should not be touched.
- 02
A safe first project
A narrow, visible place to begin—small enough to test and useful enough to matter.
- 03
Five questions for any vendor
Scope, access, approval, monitoring, and what happens when the workflow breaks.
The agenda
Every minute has a job.
Your actual work
Each owner names one recurring task. That list becomes the spine of the session.
What an agent actually is
Instructions, access, memory, and boundaries in plain language—then one short live build.
Where agents pay off
Useful patterns sized for businesses with one to twenty people.
Working lunch: Map Your Week
Sort your recurring work on paper, then get a straight answer on your top item.
How this goes wrong
Five failure modes and the boundaries that prevent them.
Your first agent, done right
Choose a narrow project with low risk, visible payoff, and thirty days to earn trust.
Next steps and private questions
A short recap, an optional free workflow audit, and time for the questions people ask privately.
Who it is for
A small room. Deliberately.
Who runs it
Practical judgment, not AI theatre.
Robert builds useful systems for real work—the kind that still has to run correctly in month six when nobody is watching.
The class compresses the same questions used to scope an actual project: what it is for, what it may touch, where a person approves, and how you know when it broke.
The work is at rblurrr.cloud ↗. Judge it yourself.