How easy it is to move from XBRL 2.1 to the new OIM Taxonomy specifications will significantly impact its adoption.

Introduction

This is the first in a series of articles that examines the modernisation initiative being undertaken by XBRL International.

Project Tavi is the current code name for the next generation XBRL specification. The first public working draft of this specification was published in September 2026.

This specification, part of the Open Information Model (OIM) initiative, represents one of the most important architectural developments in the evolution of digital business reporting since the introduction of XBRL over 25 years ago.

Yet despite the significance of the initiative, the transition to the new OIM based framework could actually prove considerably less disruptive than many organisations might initially assume.

Rather than completely redesigning XBRL, the emerging model clarifies and modernises concepts that already exist within today’s XBRL ecosystem.

The result is a framework designed to support modern data pipelines, AI-assisted tooling and processing, more consistent validation and implementation flexibility. While at the same time, still preserving compatibility with existing reporting semantics and regulatory infrastructures.

The challenge of today’s XBRL environment

Over the last two decades, XBRL has become the global standard for structured business reporting.

Regulatory systems worldwide such as the SEC’s EDGAR platform in the United States and ESEF reporting in Europe, to name but two of the most visible, depend upon XBRL reporting frameworks.

However, practitioners across the XBRL ecosystem experience some persistent challenges:

• XBRL semantics are defined in XML and distributed across multiple structures
• Taxonomies, contexts, dimensions, and linkbases can be difficult to interpret collectively, especially when they exist in large numbers
• Validation logic often exists outside the XBRL reporting model itself
• Many jurisdiction-specific filing rules require extensive external processing layers
• Integration with modern data architectures can be cumbersome

As highlighted in the OIM Taxonomy presentation materials, the issue is not that the XBRL model is inadequate — quite the contrary — just that the underlying reporting model has never been expressed explicitly enough.

The central premise behind Project Tavi is therefore remarkably conservative:

XBRL already contains the model we need — but it needs modernisation.

Therefore, the objective is not to redefine reporting semantics, but to refine the existing multidimensional reporting model, explicitly define the underlying objects in an implementation-independent manner resulting in a model that is easier to understand and to process.

What is OIM ?

The Open Information Model (OIM) is an ongoing initiative to modernise XBRL. It is an abstraction layer that defines XBRL semantics separately and independent of any single syntax or document representation.

Historically, XBRL has been tightly associated with XML syntax and document-oriented reporting. OIM changes this perspective by defining a conceptual reporting model independent of serialisation format.

Under the proposed framework:

• XML becomes just one possible representation rather than the core model itself
• JSON serialisation becomes a primary reference implementation
• Reports, taxonomies, validation rules, and dimensional structures become unified components of a single explicit model
• Facts are defined as typed data values located within a multidimensional cube, identified by dimensional coordinates

This conceptual shift is significant because it aligns XBRL more closely with modern data engineering practices.

Making the existing cube model explicit

One of the key insights of the model expressed by Project Tavi, is that current XBRL reporting already behaves like a multidimensional cube structure.

Every fact in an XBRL report is already defined by the “built-in” dimensions of:

• Concept — the “what”
• Entity — the “who”
• Period — the “when”
• Unit — the “how”

Together with any additional dimensions that are defined by the taxonomy.

Today, those relationships are largely described using contexts and dimensional definitions, which are not always particularly easy to understand and follow.

The proposed model formalises this structure and provides consistency between those dimensions that are “built-in” and inherently available and those that are defined by model authors.

This approach introduces clearly defined objects that make up the model. Additionally, for the large majority of these objects, there is a direct mapping between existing XBRL definitions, as defined in XML, and these new components.

The list below outlines the correspondence between the major model objects and existing XBRL definitions

• Cube — hypercube definition, notAll relationship
• Cube dimensions — all relationship, hypercube-dimension relationship
• Domains — dimension-domain relationship
• Members — domain-member relationship
• Networks — ELR networks such as presentation and calculation
• Groups — ELR and sections
• Relationships — to/from linkbase relationships
• Layouts — table, breakdown and axis definitions
• Properties — attributes, both built-in and user defined

The first important observation is that most of these objects represent, or include, existing XBRL constructs. In many cases, current implementations of XBRL products will already have implemented objects very similar to these.

But, by defining them purposely as objects, the definitions and the semantics become explicit, machine-readable and independent of the implementation or serialisation format.

Why the transition may be easier than anticipated

At first glance, a new “OIM Taxonomy” specification may look and sound like a major platform migration. But in practice, the transition is overwhelmingly evolutionary, designed to preserve ecosystem continuity.

The meaning of XBRL does not change

The most important message is simple:

The proposed model does not change the meaning of what XBRL is describing.

As noted above, current concepts, dimensions, facts, calculations, and relationships continue to exist. The existing reporting semantic model remains intact.

The proposed specification updates and improves how the model is expressed, but not what the data means.

This distinction is critical because it dramatically reduces migration risk.

Existing taxonomies can potentially be converted automatically

One of the strongest indicators of the feasibility and manageability of transition is the fact that automated conversion prototypes already exist.

A major part of the specification development process is to prove its viability in an incremental manner as it advances. To this end, a significant investment has been made in providing both conformance tests together with a migration tool for existing taxonomies.

These provide a high degree of confidence that the model is coherent and can be validated and that existing taxonomies can be converted to the new model without information loss.

The OIM Taxonomy model and Technical Introduction presentations made recently during the Digital Reporting Europe conference in Paris noted that:

• Existing taxonomies can be transformed using generic conversion approaches
• Core taxonomy structures map easily and naturally into the new framework
• Different taxonomy styles may benefit from different conversion strategies
• It is, however, recognised that more complex table linkbase implementations may require additional work

This suggests that for most organisations the conversion of existing taxonomies can be straightforward, therefore avoiding a taxonomy redesign from scratch.

In this case, the transition could resemble a normalisation and modernisation process layered on top of existing taxonomy assets.

Validation moves closer to the model

One of the most important practical improvements involves validation architecture.

Today, many regulatory validation rules sit outside the reporting model and are not defined nor implemented in XBRL:

• Regulatory rules concerning the construction of extension taxonomies
• Filing rules and constraints, such as which parts of a report are to be filed
• Data quality rules, especially those for accuracy and precision
• Jurisdiction-specific business validations, formats and profiles

With the proposed model, many of these constraints, the majority of which require specific software development, can be embedded directly within cube definitions and model structures. The model offers a much richer set of constraints and filters as compared to what exists today.

This has several advantages:

• Validation becomes more portable
• Reporting semantics become clearer
• Data quality rules become easier to manage
• Tool interoperability improves
• Jurisdictional consistency becomes easier to achieve

For software vendors and regulators, this could substantially reduce implementation complexity over time.

Document independence enables modern architectures

Another major advantage of the OIM framework is document-source independence.

Current XBRL processing is still heavily influenced by XML document structures and Inline XBRL rendering requirements.

Whereas the proposed taxonomy model builds on this separation of reporting semantics from data representation syntax.

Under the new architecture, the same reporting model could theoretically consume or map data from:

• XML filings
• Inline XBRL
• APIs
• Databases
• HTML
• PDF
• Tabular systems

This flexibility aligns naturally with modern enterprise data pipelines and cloud-native architectures.

It also positions XBRL more effectively for integration with AI systems and analytics platforms.

JSON as a reference serialisation

A particularly important practical development is the inclusion of a JSON-based reference serialisation.

The OIM Specification Working Group, who are responsible for this specification, recognises that an implementation-independent conceptual model alone is insufficient for interoperability.

The proposed JSON representation therefore becomes a practical bridge between conceptual architecture and software implementation.

This may significantly lower barriers for developers unfamiliar with XML-heavy processing environments.

Modern software ecosystems increasingly favour:

• JSON
• REST APIs
• Cloud-native pipelines
• Streaming architectures
• Schema-driven validation

The OIM initiative aligns XBRL with these technologies far more naturally than traditional XML-centric implementations.

Features for more expressive modelling

The proposed model introduces a cleaner, richer and more coherent organisational structure.

Cubes

Cubes become the central organising structure for facts and dimensional constraints and, ultimately, validation and queries.

All facts in a report must belong to one or more cubes.

Domains and Members

Domains formally define a set of allowed dimensional values, while members define explicit reporting categories to be used as explicit values of domains, removing the need to use special types, as is often the case today.

This improves dimensional consistency and validation clarity.

Groups

Groups provide an organisation mechanism to associate related objects into logical assemblies:

• Calculations
• Presentations
• Layouts
• Cube associations
• Related reporting structures

This effectively modernises and rationalises the role currently played by Extended Link Roles (ELRs) and addresses some of the limitations that exist today; in particular, those related to naming and ordering.

Tooling becomes simpler

The move to a well-defined object model coupled with a JSON serialisation format provides the opportunity for both simpler and more diverse technologies.

The JSON format is well supported in all current programming environments and is also readily exploitable by AI tools. This applies equally to tooling for taxonomy generation through to the processing of XBRL reports and data.

Why regulators should support the direction

For regulators, the features described within the OIM Taxonomy address several long-standing operational challenges.

Benefits include:

• Clearer semantic consistency
• Improved comparability across jurisdictions
• Easier implementation of validation rules
• Stronger interoperability
• More efficient data ingestion pipelines
• More mainstream technology and skill sets

Most importantly, the transition approach explicitly emphasises gradual evolution rather than forced replacement.

A staged approach minimises disruption while allowing the ecosystem to mature organically.

AI and the future of structured reporting

One particularly forward-looking aspect of the proposed model, is its compatibility with AI-assisted tooling.

Current XBRL implementations can be difficult for machine-learning systems to interpret because key pieces of the semantic model are scattered across multiple taxonomy structures. To the point that sometimes it can feel as if current XBRL is actually “AI-resistant”.

An explicit model offers several advantages:

• Cleaner semantic relationships
• Clearer validation logic
• Easier graph representations
• Improved machine interpretability
• Better support for automated data extraction and analysis

As AI-driven regulatory analysis and financial intelligence systems become more important, these structural improvements may become increasingly valuable.

Some of this is already borne out by the experiences of using AI assisted tooling during the specification development. It has been used to test out the model and also to convert existing taxonomies and filer extension taxonomies to the new format. Sample conversions have been made for US-GAAP, IFRS and ESEF taxonomies together with converted reports.

Remaining challenges

Despite the positive outlook, it cannot be ignored that the transition will still involve real work.

Areas likely to require additional effort include:

• Complex table linkbase implementations
• Legacy rendering software, such as viewers and report preparation tools
• Regulator-specific validation workflows and processing
• Tooling modernisation
• Training and ecosystem education

In addition, organisations with deeply customised XBRL pipelines may need to revisit portions of their architecture.

However, because the underlying reporting semantics remain stable, these challenges are evolutionary rather than revolutionary.

To conclude

The proposed specification and OIM framework represent a modernisation of XBRL rather than a replacement for it.

The initiative recognises that XBRL already possesses a powerful multidimensional reporting model. The new specification makes that model explicit, implementation-independent, and easier for modern systems to process consistently.

For preparers, regulators, software vendors, and data consumers, this should ultimately reduce complexity rather than increase it.

Most importantly, the transition path appears intentionally designed to preserve ecosystem continuity:

• Existing reporting semantics remain valid
• A very large proportion of existing taxonomies will be convertible, most likely in an automated manner
• Regulatory infrastructures can evolve gradually
• Current filing systems can coexist alongside newer implementations

The new model will become the bridge to allow XBRL to move from its finance-oriented XML-centric origins into a more flexible, API-driven, AI-enabled reporting future — without abandoning the enormous investments already made across the global reporting ecosystem.

The author is David Bell of AM2 Limited

This article was originally published on Medium.com as part of the Digital Reporting Made Simple publication.