QuestionQ8

VerifiedDesign AI-powered business solutions

A company uses a Microsoft Foundry agent that summarizes customer feedback and suggests products to customers. The agent accesses data from multiple knowledge sources.

Users report slow agent response times.

Telemetry indicates that the agent frequently hits its token-usage limit.

You need to recommend a solution that reduces token usage without reducing the quality of generated responses.

What should you recommend?

Explanation

Chunk documents during indexing so retrieval returns smaller, relevant passages rather than entire or oversized documents. In RAG and agentic retrieval, retrieved passages are included in the model input; appropriate chunking helps meet input-token limits and prevents truncation while retaining relevant grounding content.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!