High Signal: Data Science | Career | AI Episode 28: From Context Engineering to AI Agent Harnesses: The New Software Discipline
120 snips
Nov 13, 2025 
Lance Martin, a machine learning engineer at LangChain, dives into the evolving landscape of AI engineering. He emphasizes the importance of context engineering and how traditional ML rules are becoming obsolete. The conversation covers why adaptable systems thrive, the architectural advantages of 'agent harnesses,' and the shift towards in-app user feedback for evaluating AI systems. Lance also shares insights on balancing autonomy in agents with human oversight and techniques for managing costs and performance in complex AI tasks.
AI Snips
Chapters
Transcript
Episode notes
Use A Harness To Manage Tool Calls
- Use a harness to execute tool calls, manage message history, and package tool outputs back to the model.
- Let the harness handle execution, context updates, and tracing to keep agents reliable and observable.
Workflows Versus Agent Autonomy
- Workflows follow predetermined code paths while agents let LLMs choose tools and order dynamically.
- Autonomy distinguishes agents and suits open-ended tasks like research or iterative coding.
Choose Workflows For Predictability
- Use workflows for predictable, repeatable tasks and agents for open-ended, adaptive tasks.
- Embed agents inside workflows when you need autonomy in a controlled step.