GateLens: A Reasoning-Enhanced LLM Agent for Automotive Software Release Analytics

TL;DR

GateLens is introduced, an LLM-based architecture for reliable analysis of complex tabular data that uses Relational Algebra as a formal intermediate representation between natural-language reasoning and executable code, addressing the reasoning-to-code gap that can arise in direct generation approaches.

Abstract

Ensuring reliable data-driven decisions is crucial in domains where analytical accuracy directly impacts safety, compliance, or operational outcomes. Decision support in such domains relies on large tabular datasets, where manual analysis is slow, costly, and error-prone. While Large Language Models (LLMs) offer promising automation potential, they face challenges in analytical reasoning, structured data handling, and ambiguity resolution. This paper introduces GateLens, an LLM-based architecture for reliable analysis of complex tabular data. Its key innovation is the use of Relational Algebra (RA) as a formal intermediate representation between natural-language reasoning and executable code, addressing the reasoning-to-code gap that can arise in direct generation approaches. In our automotive instantiation, GateLens translates natural language queries into RA expressions and generates optimized Python code. Unlike traditional multi-agent or planning-based systems that can be slow, opaque, and costly to maintain, GateLens emphasizes speed, transparency, and reliability. We validate the architecture in automotive software release analytics, where experimental results show that GateLens outperforms the existing Chain-of-Thought (CoT) + Self-Consistency (SC) based system on real-world datasets, particularly in handling complex and ambiguous queries. Ablation studies confirm the essential role of the RA layer. Industrial deployment demonstrates over 80% reduction in analysis time while maintaining high accuracy across domain-specific tasks. GateLens operates effectively in zero-shot settings without requiring few-shot examples or agent orchestration. This work advances deployable LLM system design by identifying key architectural features--intermediate formal representations, execution efficiency, and low configuration overhead--crucial for domain-specific analytical applications.

Figures (5)

• Figure 1: Comparison of conventional Chain-of-Thought (CoT) reasoning and its extension using Relational Algebra (RA) statements. (Left) CoT employs unstructured, fused reasoning where multiple analytical concepts are blended together in informal thoughts without clear separation. Operations cannot be independently mapped to executable code snippets, making reasoning steps opaque and difficult to debug in isolation. (Right) RA-based reasoning adopts a structured, compositional approach where each relational algebra operation (filtering, joining, projecting) is a discrete, formally grounded step. Each operation is independently interpretable, reusable, and can be directly mapped to executable code. The color-coding illustrates how RA maintains clear boundaries between distinct analytical steps, whereas CoT conflates multiple operations within single reasoning thoughts. This structural distinction enables transparent, formally grounded reasoning that can be systematically verified and debugged at each stage.
• Figure 2: GateLens top-level architecture: The system processes high-level queries from the end user, generates the necessary data manipulation code using the enhanced reasoning layer with the help of RA, executes it, and outputs the result table as a decision-support resource.
• Figure 3: Overview of the prompt engineering flow and prompt structure within the GateLens system architecture.
• Figure 4: Comparison of the original method and the method without the RA module across different datasets.
• Figure 5: Comparison of GateLens against CoT+SC across different numbers of few-shot examples.