Where to draw the determinism boundary in LLM pipelines
aronchick · x · 2026-09-26
In a thread on LLM pipeline architecture, the author argues every pipeline team must answer one question: "Where do we draw the determinism boundary?"
His heuristic: if an edge case requires a human to sort out, that's a great spot for an LLM to judge — but only to opine, not decide. He also notes the pattern repeats outside the hot path: the biggest LLM-assisted refactors quietly lean on deterministic tooling underneath. The judgment layer always floats on a deterministic substrate.
More from coding & agent
- Jev vs BigQuery's built-in AI functions: a scaling test — rseroter · 2026-09-26
- Drone Wireless Charging Cradle One-Shotted With AI CAD Tool dingcad — yacineMTB · 2026-09-26
- GPT-6 Sol hits #6 on Agent Arena with +7.7% net improvement at 56% less cost per task — arena · 2026-09-26
- "If the app works, the code's perfect": dev says AI ended his code hygiene — banteg · 2026-09-26
- New trick encodes source map data in background colors so AI agents can locate code lines from screenshots — jasonkneen · 2026-09-26
- Claude-generated video explains model-based formal verification and state machines — DavidKPiano · 2026-09-26