Inspect the Warehouse Brain Terminal in Your Browser
Experience the exact command-line interface produced by python3 warehouse_laya.py --all.
See how Laya evaluates all 20 real-world warehouse orders in real time using a pure neural semantic router—no hardcoded if/else rules, no API keys, and calibrated probabilities for every single decision.
Zero Setup, Live in Browser
Users and evaluators can inspect exactly how the terminal command executes without having to clone the repository, set up a Python virtualenv, install torch or transformers, or download model checkpoints.
Pure Semantic Neural Routing
In traditional systems, urgency is hardcoded with hundreds of fragile if/else rules. Laya replaces all of them with a single forward pass over an URGENCY_DECISION_SCHEMA, projecting orders into high-dimensional embedding space.
Calibrated Probability Mass
Every row outputs genuine calibrated confidence (e.g., 89.9%). Unlike traditional LLM token sampling, Laya produces exact normalized probabilities across candidate criteria, enabling automated confidence gating.