A ready-to-use prompt to design a strategy for freshness-sensitive retrieval, producing a structured draft you can act on directly rather than a vague suggestion.
RAG and Retrieval
Prompts for designing and evaluating retrieval-augmented generation systems.
Helps you design a cold-start strategy for a new RAG system with no usage data quickly, while still surfacing the open questions a real version of this work would need answered.
A structured prompt that helps you design a strategy for retrieval over structured knowledge graphs by asking the model to show its reasoning, not just its conclusion.
Turns a short description into a concrete plan to design an A/B test plan for comparing two retrieval strategies, with assumptions called out instead of silently guessed.
A ready-to-use prompt to design a strategy for summarizing long retrieved documents before generation, producing a structured draft you can act on directly rather than a vague suggestion.
Helps you design an access-control-aware retrieval strategy quickly, while still surfacing the open questions a real version of this work would need answered.
A structured prompt that helps you design a strategy for handling conflicting information across sources by asking the model to show its reasoning, not just its conclusion.
Turns a short description into a concrete plan to design a strategy for retrieval over code repositories, with assumptions called out instead of silently guessed.
A ready-to-use prompt to design a feedback loop from user ratings into retrieval tuning, producing a structured draft you can act on directly rather than a vague suggestion.
Helps you design a strategy for multilingual retrieval quickly, while still surfacing the open questions a real version of this work would need answered.
A structured prompt that helps you design a caching strategy for repeated queries by asking the model to show its reasoning, not just its conclusion.
Turns a short description into a concrete plan to design a context window budgeting strategy for retrieved passages, with assumptions called out instead of silently guessed.