Overview
Research area: ML Authors: Anoushka Vyas, Aarushi Dhanuka, Sina Khoshfetrat Pakazad Published: 2026-06-19 arXiv: 2506.14970
Summary
Production data integration is bottlenecked by repeated, lossy handoffs between data owners, engineers, and analysts who must collaboratively discover, structure, and query enterprise data. This paper presents Data Intelligence Agents (DIA), a system of three agents (Data Interpreter, Schema Creator, and Query Generator) that compresses this workflow by treating autonomous coding agents (ACAs) as a first-class abstraction.
Unlike conventional approaches, these agents do not merely emit text. Instead, they generate, execute, validate, and repair concrete artifacts, draw on a shared memory for experience reuse, and surface each result for review by domain experts. DIA is deployed in production for enterprise customers.
Query Generator Evaluation
The research team studied the Query Generator in depth and evaluated it in fully autonomous mode across seven SQL benchmarks, spanning four task categories and four SQL dialects. It matched or surpassed the best published results on all seven benchmarks.
Key Takeaways
- DIA compresses the data integration workflow (discover, structure, query) into a three-agent system.
- Agents are grounded in execution: they produce, run, validate, and fix concrete artifacts rather than just generating text.
- A shared memory enables experience reuse across tasks, with outputs reviewed by domain experts.
- The Query Generator achieves state-of-the-art or better results on all seven SQL benchmarks tested (four task categories, four dialects).
- The results show that an execution-grounded architecture built on ACAs and shared memory generalizes across data intelligence workloads, with adaptation confined to natural-language instructions.
Original Abstract
> Production data integration is bottlenecked by repeated, lossy handoffs between data owners, engineers, and analysts who must collaboratively discover, structure, and query enterprise data. We present Data Intelligence Agents (DIA), a system of three agents (Data Interpreter, Schema Creator, and Query Generator) that compresses this workflow by treating autonomous coding agents (ACAs) as a first-class abstraction: rather than emitting text, the agents generate, execute, validate, and repair concrete artifacts, draw on a shared memory for experience reuse, and surface each for review by domain experts. DIA is deployed in production for enterprise customers. We study the Query Generator in depth and evaluate it in fully autonomous mode across seven SQL benchmarks spanning four task categories and four dialects. It matches or surpasses the best published results on all seven, demonstrating that an architecture grounded in execution, built on ACAs and a shared memory, generalizes across the data intelligence workload with adaptation confined to natural-language instructions.
*Auto-collected on 2026-06-19*