Loading learning content…
Loading learning content…
Reviewing the draft, seeing the exact prompt, and turning it on — one agent or a whole team.
Read through the lesson, mark it complete when the concept is clear, then move to the next lesson in the sequence or jump back to the module map.
The interview (Step 1) hands you a draft — a named agent with a job description, a starting prompt, and a suggested cadence. Launching is the step where you look at that draft, adjust anything that's off, and turn it on.
Nothing is hidden here. You read the exact wording the agent will use before it ever runs. If a line sounds wrong for your business, you edit it, the same as you'd edit a document.
Launch Studio supports two shapes:
Start with one agent if you're not sure. You can always add more later — nothing about the interview or the rubric locks you into a single shape.
Every agent Launch Studio produces ships with cautious defaults: it drafts and flags for your review rather than acting automatically, unless you explicitly change that. That's a deliberate choice — the goal of Step 2 is a working draft you trust, not a black box you have to hope is right.
Before an agent touches real work, you can run it in the on-platform test playground — the same conversation flow it'll have in production, but consequence-free. This is where you catch a tone that's off or an instruction that's ambiguous.
Once it's live, the real question is: is it doing the job well? Step 3 covers grading — how Launch Studio checks the agent's work against the rubric from Step 1, and what to do when it falls short.