Tech Giants Are Rebuilding Documents Just for AI

IBM, Nvidia, and Red Hat are creating DocLang, a new open standard for documents designed specifically for AI to read and understand.

A new working group backed by IBM, Nvidia, and Red Hat is creating an open standard called DocLang. Hosted by the Linux Foundation's LF AI & Data project, the initiative aims to create a universal document format designed specifically for artificial intelligence. Today, AIs struggle to efficiently process documents like PDFs and reports that are made for human readers. DocLang seeks to solve this by building a new format from the ground up that is optimized for how large language models (LLMs) process information. The goal is to establish a common, machine-readable structure for business documents that AI systems can understand natively, without complex and error-prone parsing.

This matters for any organization building or using AI to analyze internal information. Preparing human-centric documents for AI is currently a major bottleneck, requiring significant time, cost, and engineering effort. Inaccurate data extraction from these documents can lead to unreliable AI outputs, especially in Retrieval-Augmented Generation (RAG) systems that rely on company knowledge bases. By standardizing the document format for machines, DocLang could dramatically simplify data pipelines. This would make AI implementations faster, cheaper, and more reliable for developers, CTOs, and IT teams. For founders and business leaders, it promises a more direct path to leveraging AI on their proprietary data.

As an open standard under the Linux Foundation, DocLang aims for wide adoption rather than becoming a proprietary tool. The project is still in its early stages, but its high-profile backers signal serious intent to solve a fundamental problem in enterprise AI. Companies should monitor the development of the DocLang specification, as it could influence future strategies for document creation and management. A successful standard might lead to a future where businesses maintain two versions of important documents: one designed for people and a corresponding DocLang version structured purely for machines.