AI Agent Response Excecution time considerationIssue A custom agent checking the quality of change records is taking more then 4 minutes to complete with the AWS Claude model provider. Facts The response time for LLM is directly dependent on the number of tokens in the response. Release Any Cause Root Cause1. The custom AI Agent generates lengthy responses due to updated instructions requiring detailed Findings and Recommendations.2. The token limit (initially 4096, later increased to 8192) was insufficient to accommodate the expanded response structure, causing truncation and retry loops. 3. Claude's inherent verbosity and linear latency scaling with token count exacerbated execution times compared to previous LLM providers like NowLLM. Resolution Steps to Resolve1. Review and optimize the agent's instructions to reduce response token count, particularly by capping the length of recommendations and explanations. 2. Constrain worknote content to include only essential details instead of duplicating the full section-by-section assessment. 3. Skip Findings/Recommendations for sections meeting your checks criteria to eliminate redundant text. 4. Trim the per-field ratings payload in the tool input schema to reduce token density, limiting explanations to a single sentence and populating suggested values only for failed fields. 5. Monitor token usage and adjust the Response Max Tokens configuration if necessary, ensuring alignment with system limits and LLM performance characteristics.