ULM and PD Engine
Definition
A natural-language interface for non-narrative, invariant-constrained knowledge across domains.
What it does
ULM-PD Engine accepts natural language as input, translates it into invariant-governed execution plans, and executes those plans deterministically across domains.
Results are returned when invariants are satisfied. Execution halts only when invariants are undefined or violated.
Execution model
- Natural language is parsed
- Results are symbolic and law-bound
- Boundary states are returned as data
- Refusal has one meaning only: invariant violation
What this system is
-
Natural language interface to a deterministic symbolic maths engine
(requires an online ChatGPT API key; execution runs locally) - Deterministic
- Invariant-governed
- Symbolic and auditable
- Explicit about boundaries and assumptions
What this system is not
- A chatbot
- A tutor
- An optimiser
- A simulator
- A best-effort inference system
See also: What PD Engine is not
Declared domains
- Mathematics Core — Available
- Building Services Engineering — Foundation & specialty packs
- Structural Engineering — Declared
- Mechanical Engineering — Declared
- Electrical Engineering — Declared
- Public Health Engineering — Declared
Only declared capabilities are supported.