Prompt chain for A/B test hypothesis generation

[PROBLEM] Our team struggles to generate statistically valid A/B hypotheses quickly. [OVERVIEW] Feeds DAM asset metadata to an AI pipeline that suggests taxonomy tags based on content analysis and brand guidelines. [TRYABLE] Open the JO flow in the canvas, export as JSON, then run through the summarizer prompt included in the template. [SETUP] Target license [IMPACT] Analytics digest reaches 3x more stakeholders with zero additional effort. [LESSONS] Chunk long inputs; the model loses precision past a few thousand tokens. [SAMPLE_OUTPUT] Flow summary: 3 entry conditions, 2 decision splits, 1 wait node, 4 message actions.