Compliance Reporting as a "Skeleton" of the Enterprise Knowledge Graph

I have this hypothesis that compliance reporting and the accounting information systems that generate that compliance report can be the "skeleton" of what people are referring to as the "enterprise knowledge graph". This blog article is my brainstorming about this idea.

X-Ray and Skeleton

In their article, The Financial Statement is a Projection, Puget Research points out that a financial statement is like an x-ray and they point out that "A financial statement is one of civilization’s most successful compression technologies." Paraphrasing, a financial statement is not "the enterprise", it is a compressed summary of information about an enterprise that is the output from a very trusted set of processes called accounting information systems. In compliance reporting, an enterprise does not provide every colorful detail about itself; it provides a compressed, summarized version of information about the enterprise generally following some specific financial reporting framework. This serves a specific purpose; provide certain specific specified information to a regulator about the enterprise in a consistent format that enables comparability across other enterprises that do the same thing. This makes the capital markets "go around" (e.g. operate efficiently and effectively).

But that compliance report is only a "skeleton". Bones. Some of the ligaments and tendons are their I guess, maybe even some of the muscles, maybe even some of the organs are identified. This is not a perfect analogy; but you might get the idea that I am getting at: (a) the information "skeleton" is far from everything, (b) the information "skeleton" is highly trusted because of the nature of accounting information systems, (c) there is a lot more activity taking place and details beyond that "skeleton" that you cannot get from that "skeleton". The primary point being that there is a trusted skeleton.

Digital Twin

At the other end of the spectrum is the notion of the "digital twin". The digital twin of an enterprise is described in the paper, Digital Twins of an Organization for Enterprise Modeling.

Per this paper, a digital twin of an organization (DTO), is essentially a living, machine‑readable mirror of how an enterprise actually works; its processes, people, information flows, and contextual nuances. It extends the familiar idea of digital twins in manufacturing into the far more fluid, human, and information‑rich domain of organizational activity.

At its core, a DTO combines enterprise models which are structured descriptions of processes, roles, systems, and rules with real‑time operational reality. Instead of static diagrams or documentation, the DTO is continuously fed by data traces left in information systems and enriched by contextual information captured from knowledge‑intensive work. This produces a representation that is:

Dynamic: reflecting ongoing activities, deviations, and changes as they happen.
Semantic: built on graph‑based meaning models and ontologies that make relationships explicit and machine‑understandable.
Context‑aware: incorporating the subtle, often unstructured information that normally disappears in traditional systems.
Actionable: enabling simulation, prediction, and scenario exploration to support agility and decision‑making.

The paper emphasizes that the DTO is not just a dashboard or a process map. It is a holistic, navigable environment; supported by technologies like Context Spaces and corporate memory systems; that helps organizations see how work actually unfolds, understand why deviations occur, and experiment with alternative paths. In doing so, it aims to close the long‑standing gap between enterprise modeling and real‑time execution, giving organizations a way to adapt faster, learn continuously, and coordinate more effectively.

And so, at this other end of the spectrum; I would point out that (a) the digital twin of an enterprise is still not everything about the organization, it is still a model; (b) that model certainly contains that "skeleton" related to compliance reporting; (c) there is a lot of other "stuff" related to the activities of an enterprise that goes beyond compliance reporting; and finally (d) much of that additional "stuff" which provides details and color is absolutely connected to that compliance reporting.

Economic Geometry

It is now the case that two different people have pointed out something that I understood but did not have the words to describe. Those words where first provided to me by Andrew Noble of Lodgeit Labs and his mention of the notion of the "manifold" or The Accounting Manifold. And now Puget Research, while not mentioning the manifold directly, do mention the notions of "economic geometry" and "topology". Stated succinctly by Puget Research, "A practical enterprise geometry would be a temporal, typed economic hypergraph with quantities, states, constraints and provenance."

Accounting ledgers are special types of tools. A ledger is not a "log" or a "list". An accounting ledger is a computational substrate. Double entry bookkeeping is a consciously designed and engineered mathematical model. Topology is a branch of mathematics and a manifold is a part of topological space. While these mathematical ideas can be applied to geometry; they can also be applied to accounting.

Imagine the utility of a queriable manifold with important details about business events that has the same trust level as accounting information systems.

Business Events

The notion that an enterprise is made up of business events or business contracts have been around for years, including in the Theory of Accounting and Control and Resource, Events, Agents (REA). This idea has been standardized via ISO/IEC 15944-4:2015 Accounting and economic ontology.

So, imagine a ledger, a business events ledger, which was the mechanism for collecting information about an enterprise. Business event information comes from documentation. That documentation is the source of information that goes into the business events journal. From the business events journal the business events ledger is projected; this is somewhat similar to how a general journal, general ledger, and general ledger trial balance works today. You get all the traceability/trackability and provenance capabilities. You get all the details about resources and obligations as explained by Dave McComb and Cheryl Dunn in their book, The Future of Accounting. McComb's and Dunn's contribution to REA is the detailed explanation of the core patterns of business events (which I have modeled here and prototyped here).

It is the business events ledger that feeds the "enterprise geometry". It is that information that enables the accounting manifold.

Populated Using Artificial Intelligence

This is a lot of detail. This detail will be populated with the assistance of artificial intelligence which provides much of the labor. No, this is not about AI hallucinating about business events. This is about maximizing the potential of artificial intelligence by adopting an AI-first, Digital-first, Graph-first mentality.

One feature of the business events ledger manifold is that the source of every piece of information is distinguished as being an actual measurement or observation, a fact that is derived using deductive reasoning, information that was provided per some probability-based process, or some other source.

The work is already being done today using traditional systems and processes. But the existing infrastructure was not constructed with artificial intelligence in mind. As a result, the foundations of these systems are inappropriate and tend to have flaws. Today's systems have lots of fragmentation. That fragmentation causes gaps. Those gaps require work arounds. Those work arounds cause inefficacies, errors which must be remediated, and other undesirable characteristics.

An improved approach might be to invest in fix processes and foundations rather than live with suboptimal foundations that are not fixable, continue remediating errors or dealing with the consequences of failures.

Recognizing the Real Role of Accounting

In the era of artificial intelligence, accounting will be recognized for what it really is. As the thesis of The Financial Statement is a Projection states, "The financial statement should remain the X-ray. It should stop being mistaken for the body." When double entry bookkeeping best practices were documented in 1494, Luca Pacioli contemplated a tool which has been, up to a few years ago, pretty much ignored. That tool, which Pacioli referred to as the Ricordanze, predicted future business events based on historical business events. Think Financial Planning and Analysis (FP&A).

ACTUS is a global open standard which has enables the implementation the notion of the Ricordanze for financial institutions. The business events of financial institutions are all financial contracts. Those financial contracts were specified as machine interpretable algorithms, thus the title Algorithmic Contract Types Unified Standards (ACTUS). The idea came from Willi Brammertz, a founder of the company Ariadne. Financial institutions use this technique today for systems integration and business analysis.

Financial contracts and trade contracts of enterprises can be represented in this same way enabling forecasts of future business events.

Business Outcomes

Refactoring accounting information systems of enterprises is a big investment. But there are opportunity costs for not leveraging artificial intelligence that also must be considered.

What specifically is in this for business? The economics of human memory are reconfigured. Significantly more hard won knowledge stays with the enterprise. Massive improvements in enterprise information quality will result. Because everything is digital, techniques such as Lean Six Sigma's Poka Yoke mistake proofing can be leveraged. But the biggest advantage is understanding the enterprise and how it actually operates better.

Being honest, most existing systems that have been created were not consciously designed, they basically evolved over time. But there really was not an over arching plan. The typical enterprise implemented software and systems to meet their immediate needs. And, we did not have the knowledge of the technologies that would even become available. So what exists typically tends to be a kludge.

Accounting & Audit by Design (A&AD) Framework proposes a different approach. Per the A&AD consistency is expected and enforced, truth is not assembled because it is inherent, errors are prevented from ever entering the system, the system is perfect per a "fenced boundary" enforced by declarative machine interpretable rules. The result is a virtuous cycle; a causal chain that yields trust.

Additional Information:

Search This Blog

XBRL-based Digital Financial Reporting