AI agents

Hybrid search, reranking, and context budgets

• Retrieval-Augmented Generation Explained for Builders
• A Rewrite Budget per Query: How Many Facts Before Retrieval
• Citing Spans, Not Chunks: The Gate Gets Sharper
• Context Window Budget: How Many Chunks to Send and When to Stop
• Extractive Answers: Sending the Passage, Not the Document
• How Many Chunks to Send: A Context Window Budget for Answers
• Hybrid Search and Reranking: When and How to Combine Them
• Mapping Retrieval Budgets to Answer Quality Targets
• Retrieval Budgets for Agent Memory: How Much to Fetch Per Turn
• Setting a Retrieved-Token Cap per Query Class
• Trimming Retrieved Chunks Before the Cap
• A Cancellation Path When a Rewrite Overruns Its Fact Budget
• A Context Budget per Turn: Deciding What Raw Text Survives
• A Quality Target in Versioned Config: Fields, Owner, Test
• A Recall Loss Budget per Rule: Setting the Number Before You Ship
• A Support-Bot Answer Built From Two Sentences
• Answering From One Span: A Pipeline From Query to Citation
• Choosing a Fact-Count Cap Versus a Token Cap: Which Binds First
• Compaction Triggers: Fold, Trim or Retrieve Again
• Cutting Candidate Rows With Metadata Before the Cap
• Dynamic Chunk Counts: Setting k per Query Type
• Fitting a Cap to the Context Window
• Metadata Filters in Retrieval: Scoping Answers by Team and Date
• Prompt Compression for Long Contexts: Methods and Trade-offs
• Query Rewriting and Expansion Before Retrieval
• Scoping Retrieval by Tenant and Team: Metadata Filters
• Sending the Passage Instead of the Whole Chunk
• Serving a RAG Answer Without Sending the Whole Document
• Setting a Retrieved-Token Cap per Query Class: A Method
• Sizing a Re-Ask Budget per Query Class: A Method
• Trimming Retrieved Text to the Answering Sentence
• When a Tool Returns Too Much: Drop, Summarize or Re-Call