Linux Foundation Launches Appia Foundation and DocLang Working Group: Two AI Infrastructure Standards Initiatives Explained

In June 2026, Linux Foundation organizations announced two distinct AI infrastructure standards initiatives that together address both the data quality and governance layers of the AI stack: the Appia Foundation, formed under the Joint Development Foundation (JDF), and the DocLang Specification Working Group, launched by the LF AI & Data Foundation.

Below is a breakdown of each initiative, their founding members, the problems they solve, and how they interconnect.

---

Appia Foundation: Standardized AI Conformity Assessment

Announced: June 17, 2026
Hosted under: Joint Development Foundation (JDF), a Linux Foundation organization

Founding Members:

Tier	Members
Steering	Arm, Ericsson, Google, Mastercard, Microsoft, Omron, OpenAI, Schneider Electric, Siemens
General	Armilla AI, Mitsubishi Electric, Nemko
Contributor	Naaia

The Appia launch materials identify a broad cross-industry founding coalition, and Nemko separately described itself as a founding member of the initiative.

Problem it solves

Appia addresses fragmented AI conformity, assurance, and trust practices across the global AI value chain. Organizations today face inconsistent approaches to demonstrating that their AI systems meet regulatory and consumer expectations. Appia aims to replace ad hoc methods with standardized, modular assessment infrastructure.

What it produces

The initiative is intended to establish modular open specifications and standardized conformity assessment frameworks for AI models, systems, and applications across the AI value chain. The focus is on practical assessment infrastructure that can support AI conformity, assurance, and trust across jurisdictions and organizational roles.

---

DocLang Specification Working Group: AI-Native Document Format

Announced: June 9, 2026
Hosted under: LF AI & Data Foundation, a Linux Foundation organization

Founding Members:

Role	Members
Founders	IBM, NVIDIA, Red Hat
Contributors	ABBYY, HumanSignal

Multiple sources list these five organizations as founding or contributing members, with some also crediting Forgis.

Problem it solves

DocLang addresses the fact that many enterprise document formats (PDF, HTML, Markdown, LaTeX) were primarily designed for human rendering rather than AI-native processing. The working group's stated goal is to provide an interoperable document-processing standard for AI and agentic workflows.

What it produces

DocLang is described as a constrained XML format built from the ground up for LLM tokenizers, with a 1-to-1 mapping between DocLang tokens and model tokens and minimal token count. Every component carries semantic role, geometric bounding box, and reading order. The specification complements the open-source Docling project.

---

How These Initiatives Fit Together

Both initiatives sit within the broader Linux Foundation ecosystem but address different layers of the AI stack, making them complementary rather than overlapping.

Layer of the AI stack they address:

• DocLang operates at the data/document layer, standardizing how documents are represented for AI and agentic workflows.
• Appia Foundation operates at the governance and conformity layer, developing specifications and assessment frameworks for AI conformity across the value chain.

How they connect:

• DocLang ensures document inputs are more structured and interoperable for AI systems.
• Appia provides a framework for assessing whether AI systems and value-chain participants meet relevant conformity expectations.
• Together, they suggest a fuller AI infrastructure pipeline: reliable document data → assessable AI systems.

It is important to note that both initiatives were announced in June 2026 and their specifications and frameworks are still under development. The Appia Foundation is accepting additional members, and the DocLang working group is also open to new contributors.