Deterministic Core, Agentic Shell
DavidKPiano · x · 2026-07-20
The author outlines a system design philosophy developed over the past two years:
- Initially, allow the model to perform reasoning for all steps, observing its behavior to determine correctness.
- Gradually strip away the steps that can be replaced by deterministic logic.
This eventually leads to a "deterministic core + agentic shell" architecture: most reasoning is eliminated, yet the system remains more stable than purely deterministic approaches while handling a wider array of tasks.
More from coding & agent
- Filtering secrets on the write path: why agent memory deserves more suspicion at ingest than retrieval — Classic-Being-7960 · 2026-10-26
- INT21's AI agent swarm rewrote Drama 3's inference runtime in Rust within one hour, beating SGLang-Omni — bingxu_ · 2026-10-26
- After OpenCode key leak: 5 practical rules for keeping keys away from agents — jacobholm_dev · 2026-10-26
- wandb 0.31.0 ships with 2x faster startup, adds JSON output for coding agents — wandb · 2026-10-26
- $150 modular Linux handheld Mecha Comet opens preorders — pocket SSH and Claude Code included — Paimaamu · 2026-10-26
- Running 20 single-job Grok bots to manage calls, roadmaps, and weekly updates — inkko44 · 2026-10-26