Your goal: design an agent that can take on new tasks, say how it stays controlled, read messy
pilot results, and make a call on how much autonomy to grant.
How this works (about 30 minutes)
- Felix asks which role you are interviewing for (product, engineering or other) and pitches his
follow-ups to it. The question stays the same.
- One design question, taken deep. Ask him about the team, the tools and the risks. He hands over
what he has, one line at a time, and later a pilot results sheet. The details are made up.
- Many follow-ups, one curveball, then a trade-off turn: ship, hold or change?
- If you want to sketch, say "can I use the notepad?" and Felix puts it up.
- No hints and no scores during the call. Your report comes after.
What scores well
- A job and a context before you draw: the tasks, the tools, what is irreversible
- A clear loop, tools with permissions, a plan that can change, memory with limits
- Control: budgets, an autonomy ladder, approval on irreversible actions, a kill switch
- A way to verify results and to measure the agent by task type, with repeated runs
- Reading results without being fooled, and a call with a line that would change it
- Answers that say where they are going and land in a minute or two
Tips
- Take Gen AI Coach · Agents and Tools first.
- Ask what the agent can touch before you say what it can do.