On the bench 4 public experiments What would an eleven-star version of one moment look like? Take one customer moment out to eleven stars, then work backwards to a six or seven-star version a real team could build, with the benefit, bet, tradeoff and smallest test written down.
Input One ordinary customer moment
Output A buildable six or seven-star version Open Beyond five stars ↗ How much agent work can your team review? Replay an eight-hour workday. Change the workload and review capacity, then see what finishes and what is still waiting.
Input Five workload and review assumptions
Output Completion, review queue & waiting time Open Agent Review Lab ↗ start check gate ask 0.98 0.91 Interactive experiment Can an agent make a decision inside a policy? Give a public pull request a defined decision path. Inspect the classification, the legal steps, and where human judgment is needed.
Input Public GitHub pull request
Output Disposition, change type & risk Open Merge Graph ↗ Meta Wellness JEV · SYSTEM ONE hook pain angle trust risk watch Interactive experiment
What becomes visible when ads share a structure? Explore public ad examples through a structured lens. Compare hooks, angles, and risk with visible classifications you can inspect.
Input Public ad library & fixtures
Output Hook, angle & risk classification Open Ad Spy ↗ 
