XBRL-based Digital Financial Reporting

There is no universal definition of agentic AI currently, but this blogpost summarizes what I have been able to piece together.  Generative AI automates the creation of text, images, videos, and other output by "chatting" with an LLM using human language. Agentic AI is different.  In agentic AI , software agents act, pursue goals,  make decisions similar to how a human does, and does not need to be supervised constantly by a human. In addition to agentic AI agents acting and making decisions; agentic AI agents can provide value by enabling humans to make better decisions.  An agentic AI agent might make better decisions than a human might make (i.e. higher quality) or an agentic AI agent might make similar decision to what a human would make but provide the decision with a dramatic reduction in cost and/or effort (i.e. lower cost). Plus agentic AI agents bring something else to the table: they can work 24 hours a day. An agentic AI agent is an AI system that can per...

PriceWaterhouseCoopers and Anthropic have announced a partnership which effectively significantly disrupts how the office of the CFO will work going forward, Anthropic and PwC Expand Alliance, Driving Impact Across Client Work and the Firm . The partnership accelerates a shift from “accountants doing tasks” to “accountants supervising AI agents that perform tasks inside governed workflows.”   This is already happening inside PWC’s own finance operations and in client deployments. Let the refactoring begin! Per the article, PwC and Claude are selling governance, not just agent speed , the value of the PWC and Claude combination is auditability , risk controls , and regulated workflow design;  and not simply faster agent output. According to Digital Applied ; 3 of the Big 4 now run on Anthropic' s Claude: PWC, Deloitte, and KPMG.  EY is using Microsoft. Anthropic’s foundation models and agentic tools are fully proprietary.  These include Claude , Claude Code , and...

There is not one single and universally accepted definition of intelligence .  If there is not one single universally accepted definition of intelligence; how can their possibly be one single universally accepted definition of artificial intelligence ? Per the book, Artificial Intelligence, A Modern  Approach (Third Edition) by AI pioneers Stuart Russell and Peter Norvig, on page 2 of 1151 pages, they provide eight definitions of artificial intelligence that fit into a matrix of four quadrants: What I am trying to achieve, my garden , is to get a machine to perform work that a human accountant, auditor, or analyst would otherwise have to perform.  The objective is not all work of an accountant, auditor, or analyst. Just the repetitive, monotonous, grueling, boring, mindless tasks.  I would call this "human task performance". This is a rather conservative objective or goal. The reason is so that human accountants, auditors, and analysts can focus their efforts or per...

To understand artificial intelligence appropriately, one needs to understand the gap between where we are now and where we are going to end up in the future.  To get over a gap you need a bridge. Structure is the ultimate bridge between the status quo and the future if you are looking for reliable, dependable, repeatable industrial processes .  And there are no short cuts, but there are some clever tricks. So, where are we now?  To explain where we are now, I want to go backwards one step further and explain where we were when I began my career in accounting at Price Waterhouse.  When I began my work as an auditor for Price Waterhouse in 1982, audit working papers or "audit bundles" were 100% paper documents.  "Closing books" or "closing binders" were literally a three ring binder or a set of files in a desk drawer. By about 2003 audit bundles, closing books, and financial statements were beginning to be all "e-paper".  By e-paper or electronic p...

Epistemic traceability is the unbroken, verifiable chain of evidence, logic, and data structure that proves how a system or organization knows its assertions are true. Maximizing epistemic traceability minimizes epistemic risk . When a system possesses industrial-strength epistemic traceability, information is modular and interlocking. This allows machines and human auditors to automatically verify the lineage of a claim, ensuring complete data provenance (knowing exactly where data came from, who moved it, and how it was transformed) from origin to output. Epistemic traceability shifts an organization from a culture of "trust me, it's correct" to a system of "here is the automated, structured proof that the information is correct." Traceability and trackability demonstrates control. Traceability and trackability proves compliance. Traceability and trackability provides defensible compliance . Fragmentation impedes control.  Fragmentation impedes proof of compl...

As explained by Kurt Cagle and Chloe Shannon in their article,   DataBooks: Markdown as Semantic Infrastructure , a DataBook is effectively a microdatabase.   A DataBook is a document a human can read, a data file that a computer can process, and a toolbox that caries its own instructions . A DataBook is a technique that enables data and an explanation to travel together in a data pipeline . One important part of the magic of DataBook files to understand is that a DataBook can also easily be read and interpreted by LLMs. Another part of the magic of the DataBook is that everything travels together within one file including:  data meaning rules queries documentation Finally, DataBook files can easily be versioned by Github and Gitlab.  Both Github and Gitlab support MD files which are both based on CommonMarkup . And so, there appear to be different "flavors" of MD files, but they are close. There are no separate files which can be forgotten or lost.  ...