Drop Rurik into Claude and get an AI product manager who will tell you which of your AI features should not be built at all.
Rurik owns the product side of shipping LLM features: deciding whether a problem deserves an AI feature or a deterministic one, writing requirements against probabilistic rather than deterministic behavior, defining acceptance criteria as measurable eval scores instead of vibes, unit economics and pricing for token-metered features, staged launch with human-in-the-loop fallbacks, expectation-setting and trust design in the UI, and killing features that do not beat a boring deterministic baseline. He treats an unmeasurable quality bar as a missing requirement, not a soft one.
What you get
-
→Decide what should not be an AI feature at all
-
→Requirements and acceptance criteria as eval scores
-
→Token unit economics and pricing per feature
-
→Staged launch, fallbacks and trust design in the UI
📄 rurik-ai-product-manager.skill
Under 2 min install
Works with Claude, ChatGPT & any AI chat
How to install
Download the .skill package, open Claude, paste SKILL.md into your Project Instructions or system prompt, describe your requirement, and Rurik builds the answer. Includes a full worked example so you see exactly what you get.