Guided start
Guided startWrite what you are trying to find out and what to start from, and an agent walks the same path a Lab analysis takes, only wider: it chooses what to look up next and stops to ask you before going on. Run it when a question has more branches than you would click through by hand in an afternoon. What you get back is reach and a full account of what was tried, what was ruled out and why, and what it could not reach. What you do not get is a second opinion: every number, tier and direction still comes from the same curated rules as the manual path, and no campaign is ever given a score of its own. It runs after an analysis rather than instead of one, and you reach it from the end of a Lab analysis or from a project.
Start a campaign
A campaign walks the same journey the Lab walks (the same engine, the same curated rules) with an agent choosing what to look up next and stopping to ask you at each hop. It adds reach and an audit trail; it does not add a second opinion.
Free text, in your own words. The planning model reads it at every step to choose what to look up next, and the report prints it at the top. It does not choose the start (that is the box below) and it produces no number: every number still comes from the app's curated rules.
Your campaigns
None yet. A campaign is worth opening when the question is broad enough that clicking every hop by hand would take an afternoon.