Why Your Agent Evaluations Will Fail You (and How to Fix Them Before Production)

Anthropic deprecated Sonnet 3.5. Some of Xelix's pipelines migrated smoothly. Others broke — and customers noticed within hours. What separated the two? Evaluation. Paul Solomon and James Price Farr have spent 5+ years building AI systems that process millions of invoices for enterprise customers. In this episode, they share the evaluation-first framework that now saves them every time a model changes, an orchestration layer fails, or an agent picks the wrong tool. Key takeaways: • Evaluation-first, not evaluation-after — Retrofitting evaluation on an agent already in production is painful. Build your eval pipeline before you build the agent. • Monitor tool calls, not just outputs — If the agent isn't selecting the right tools, nothing downstream will be correct. Tool-call monitoring is your leading indicator. • 3 tiers of automation — Not everything needs an agent. Rules-based → single LLM call → agentic system. Pick the simplest tier that solves the problem. • Extended thinking tames token explosion — After migrating to newer, more verbose models, enabling extended thinking (with a budget) moved reasoning out of expensive output tokens and brought costs back under control. • Human-in-the-loop by default — Start with human review on every output, then earn trust toward touchless automation as customers gain confidence. • Pragmatism wins — Use whatever technology works best for the problem. Not every feature needs an LLM. Recorded live at AWS Summit London.With Paul Solomon (linkedin.com/in/paul-solomon), Head of AI Engineering at Xelix ; With James Price Farr (linkedin.com/in/jamespricefarr), AI Engineering Team Lead at Xelix• Xelix — AI-Powered Accounts Payable Platform (xelix.com)Strands Agents SDK — Open Source (github.com/strands-agents/sdk-python)Amazon Bedrock — Managed LLM Inference (aws.amazon.com/bedrock)Amazon Bedrock AgentCore (aws.amazon.com/bedrock/agentcore)Strands Agents — Steering Files and Hooks for Agent Accuracy (Claire Liguori) (strandsagents.com/latest/blog)Amazon SageMaker (aws.amazon.com/sagemaker)Fast.ai — Practical Deep Learning Courses (Book Recommendation) (fast.ai)The Fifth Risk — Michael Lewis (Book Recommendation) (amazon.com/Fifth-Risk-Michael-Lewis/dp/1324002…)Neurosymbolic AI and Automated Reasoning on AWS (aws.amazon.com/bedrock/agentcore)Kiro — AI-Powered Development Environment (kiro.dev)