XBRL-based Digital Financial Reporting

This blog post is a response to a LinkedIn post by Paul Wilkinson effectively calling for a literacy project . The literacy project relates to the notion of " XBRL for all " or "AI-powered personal finance reporting using XBRL — the same structured reporting language used by Fortune 500 companies, scaled for every individual." I agree with both things Paul is saying.  I agree that  personal finance can benefit from the same things Fortune 500 companies can benefit from.  And I also agree that what is needed to make that a reality is a literacy project, as opposed to a technology project. What is going on in the world seems to be misunderstood by many, if not most, people. Two key aspects of what is going on is the move to "digital" and harnessing the power of  "artificial intelligence". Very specifically, "digital" and "artificial intelligence" as applied to the following as stated by Paul: "Today, GAAP and IFRS are lang...

This blog post introduces you to what I am referring to as a global open industry standard Digital Closing Book .   Here is my working proof of concept digital closing book  that I will be referring to in this blog post. To understand what a digital closing book is; it is helpful to first understand the problem that I am trying to solve. The consultancy Gartner estimates that currently the typical Fortune 1000 company used more than 800 electronic spreadsheets to prepare its financial statements for regulatory reporting. The current approach becomes increasingly problematic when you know that a 2024 multi‑university study found that 94% of spreadsheets used in business decision‑making contain errors , many of them critical. This is only one of many, many studies that point out spreadsheet error pervasiveness in operational environments. The third thing to understand is that electronic spreadsheets are effectively documents and LLMs have a really hard time understan...

I used to think of knowledge representation approaches as a spectrum.  Now, I look at them as a pipeline . In her article, Systems for Organizing , Jessica Talisman refers to these sorts of systems as knowledge organization systems (KOS). When you are talking about artificial intelligence, it is very important that you understand two key details and distinguish between (a) the type of artificial intelligence you are talking about and (b) the problem that you are trying to solve.  The point here is that not all artificial intelligence is the same and not all problems are the same.  If you do not understand this, I highly recommend that you read Jessica Talisman's article (i.e. System for Organizing ). As Talisman points out in her article; metadata, taxonomy, thesaurus, schema, ontology, and knowledge graph work together as she describes: " Metadata establishes identity, taxonomy imposes hierarchy, the thesaurus maps equivalence and association, the schema enforces s...

Effectively, what financial reporting standards do is prescribe standard "buckets" into which information about business events needs to go.  There are relationships between some of the "buckets" and the "buckets" need to be organized in certain specific ways. You don't really need "accounts" to create a financial statement.  But you do need to understand which standard financial report line item needs to contain the information about the business event. But accounting systems tend to not track this business event information.  As such, the information has to be added manually. That is why accounting systems cannot output proper cash flow statements and statements of changes in equity. The business event information does not exist in the accounting system. Traditionally, in the world of accounting, these "buckets" are formally known as " accounts ".  Companies create these accounts, they organize the accounts into what is...

The Kimball Method is a dimensional modeling approach for building a data warehouse that builds the data warehouse from business‑focused data marts, using fact and dimension tables to create a fast, intuitive, and scalable analytical environment.  The Kimball method is based on a series of key principles that define how data should be structured and organized to facilitate its analysis and exploitation. These principles form the basis of dimensional modeling, providing a clear and systematic framework. The Kimball Method was conceived during the 1980s by Ralph Kimball and other colleagues at Metaphor Computer Systems. Since then, it has been successfully utilized by thousands of data warehouse and business intelligence project teams. Kimball and his colleagues did not invent the idea of multidimensional data or even the earliest forms of star schemas; what they did was formalize, popularize, and systematize them into the modern dimensional modeling method used worldwide. Kimball’s...

In a prior blog post I mentioned the limitations of traditional business intelligence and business analytics. One of the more significant limitations of traditional business intelligence is the lack of support for semantics by traditional business intelligence and business analytics platforms.   To address this limitation, traditional business intelligence is adding what is being referred to as a " semantic layer ".  A semantic layer is the governed business meaning layer that defines the " things ", the " associations between things ", and other stuff necessary for a analytics systems and artificial intelligence systems speak the same language. An example of adding a semantic layer can be seen by checking out Google Looker . Looker uses something called LookML or Looker Modeling Language .  Looker Modeling Language is described as the language that is used in Looker to create semantic data models. Looker says that a "semantic model is the foundat...