Experimentation
Pipelines of staged experiments
Decide the order and the combination a set of experiments runs in, and let the schedule arm each stage in turn.
A pipeline is an ordered run of stages, and a stage is a set of experiments that run at the same time. It is a sequence of parallel groups rather than a general graph, because order and combination is exactly what a CRO programme needs to express.
The stages are the exclusion mechanism, and that is the main thing building one buys you. Experiments that would collide go into different stages and each runs at full traffic in turn, rather than being made mutually exclusive at runtime and splitting the audience between them.
Each stage advances on a rule you pick: when every experiment in it has concluded, which is the default and the only one that cannot cut anything short; when the first conclusive result arrives, which stops the rest; or when a person says so.
Two separate switches govern how much the engine may do on its own, because auto-stopping the wrong experiment and auto-deploying a loser are not comparable mistakes. Both default to off, and with automation off the engine still evaluates the board and logs what it would have done. That dry run is the same function with the same inputs rather than a second code path, which is what makes it an honest preview.