Back to Catalog
Language Models
Runtime Patterns
Stochastic-Deterministic Boundary (SDB)
A four-part contract (proposer, verifier, commit, reject) that specifies how LLM output becomes system action.
Intent & Description
📋 Context
Production LLM agents are built on a stochastic core composed with deterministic systems. The SDB provides a contract for managing this boundary safely.
Real-world Use Case
Production agents that need reliable action execution from LLM outputs.
Source
Advantages
- Explicit boundary specification
- Runtime safety checks
- Clear failure modes
Disadvantages
- Additional complexity
- Requires careful verifier design
- Performance overhead
Implementation Example
# SDB Implementation
proposer = llm.propose()
verifier = deterministic_check(proposer)
if verifier.pass: commit(proposer)