Harness Engineering for Predictable Agentic Systems: An Empirical Study of Deterministic Execution Constraints
Source
Saransh Dhage
Author
Saransh Dhage
Date
Terms in this piece · Glossary
determinism — Whether the same input reliably produces the same output — something LLM systems mostly lack, which changes how you test and debug them.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
Why it matters
If your AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition →'s tool and state sequences are already consistent, the unconstrained free-text planning step is the dominant remaining source of run-to-run variance — constrain the plan schema, not just tool selection.
