Typeface

Walk through SWE-bench, AgentBench, GAIA, and TAU-bench: what they measure, their shared blind spots, and how to interpret agent benchmark numbers.

Three concrete situations where fine-tuning is the wrong tool

Which of these are valid reasons NOT to fine-tune?

Order the five steps of the canonical agent loop from first to last

Explain the LLM vs runtime split in an agent and why the distinction matters for debugging

Which component is 'the agent' in an LLM-based agent system?

How does LangGraph's StateGraph enable agent loops that a standard DAG pipeline cannot express?

What LangGraph feature allows the agent loop to cycle back to an earlier node rather than terminating?

What conditions produce an infinite agent loop and what runtime safeguards prevent one?

Which runtime safeguards prevent an agent from looping indefinitely?