Blueprint First, Model Second: A Framework for Deterministic LLM Workflow

TL;DR

This paper introduces the Source Code Agent framework, which separates workflow logic from LLMs to enable deterministic, reliable, and efficient autonomous agents for structured environments with strict procedural requirements.

Contribution

It proposes a novel decoupled architecture that uses source code-based blueprints and deterministic execution, improving reliability and performance over traditional LLM-based agents.

Findings

Achieves 10.1% higher average Pass score on tau-bench benchmark.

Dramatically improves execution efficiency.

Establishes a new state-of-the-art in deterministic LLM workflows.

Abstract

While powerful, the inherent non-determinism of large language model (LLM) agents limits their application in structured operational environments where procedural fidelity and predictable execution are strict requirements. This limitation stems from current architectures that conflate probabilistic, high-level planning with low-level action execution within a single generative process. To address this, we introduce the Source Code Agent framework, a new paradigm built on the "Blueprint First, Model Second" philosophy. Our framework decouples the workflow logic from the generative model. An expert-defined operational procedure is first codified into a source code-based Execution Blueprint, which is then executed by a deterministic engine. The LLM is strategically invoked as a specialized tool to handle bounded, complex sub-tasks within the workflow, but never to decide the workflow's path. We conduct a comprehensive evaluation on the challenging tau-bench benchmark, designed for complex user-tool-rule scenarios. Our results demonstrate that the Source Code Agent establishes a new state-of-the-art, outperforming the strongest baseline by 10.1 percentage points on the average Pass^1 score while dramatically improving execution efficiency. Our work enables the verifiable and reliable deployment of autonomous agents in applications governed by strict procedural logic.