BRAID: Bounded Reasoning for Autonomous Inference and Decisions

TL;DR

BRAID introduces a structured prompting framework using Mermaid-based instruction graphs that improves reasoning accuracy and cost efficiency in large language models for autonomous inference tasks.

Contribution

This paper presents BRAID, a novel bounded reasoning framework that enhances LLM reasoning through structured, machine-readable prompts, reducing costs and increasing accuracy.

Findings

Structured prompts significantly improve reasoning accuracy.

BRAID reduces inference costs in autonomous systems.

Effective across multiple GPT model tiers.

Abstract

Large Language Models (LLMs) exhibit nonlinear relationships between performance, cost, and token usage. This paper presents a quantitative study on structured prompting using BRAID (Bounded Reasoning for Au tonomous Inference and Decisions) across multiple GPT model tiers, eval uated on the AdvancedIF, GSM-Hard, and the SCALE MultiChallenge benchmark datasets. BRAID introduces a bounded reasoning framework using Mermaid-based instruction graphs that enable models to reason struc turally rather than through unbounded natural-language token expansion. We show that structured machine-readable prompts substantially increase reasoning accuracy and cost efficiency for agents in production systems. The findings establish BRAID as an effective and scalable technique for optimizing inference efficiency in autonomous agent systems. All datasets and detailed result logs are available at https://benchmark.openserv.ai.