Transduction is All You Need for Structured Data Workflows

ICLR 2026 Conference Submission20491 Authors

Keywords: Large Language Models, Agentics, Agent, Structured Data, Software

TL;DR: Introducing Agentics, a framework that simplifies building agentic AI systems through logical transduction, allowing for declarative data modeling and improved scalability in wide range of structured data workflow tasks.

Abstract: This paper introduces Agentics, a functional agentic AI framework for building LLM-based structured data workflow pipelines. Designed for both research and practical applications, Agentics offers a new data-centric paradigm in which agents are embedded within data types, enabling logical transduction between structured states. This design shifts the focus toward principled data modeling, providing a declarative language where data types are directly exposed to large language models and composed through transductions triggered by type connections. We present a range of structured data workflow tasks and empirical evidence demonstrating the effectiveness of this approach, including data wrangling, text-to-SQL parsing, and domain-specific multiple-choice question answering.

Primary Area: infrastructure, software libraries, hardware, systems, etc.

Submission Number: 20491