On turning continuing education into a reference asset, and what it changes about how I practice.
Every CPA has a continuing education obligation. Forty hours a year, give or take, depending on the state board. I have been practicing long enough that the obligation is approaching a thousand hours of sit-through time. For most of that, I treated the hours the way almost every CPA I know treats them. I took the courses. I passed the exams. I filed the certificates. And most of what I studied left again within a few weeks, because that is what happens to knowledge that does not have a home to return to.
This time I decided to build the home that wasn’t in my head.
The frustration that started it was specific. A client would ask a question that touched a domain I had studied, sometimes recently, and I would find myself back in the same place I had been before the course, re-reading primary sources and re-building the reasoning chain. I was paying the cost of the learning twice, or three times, because the first pass had never been captured in a form I could return to later.
The compounding was not happening. I was paying for continuing education every year, but the balance sheet was being written in pencil.
Worse than the redundant work was what the pattern implied about the whole model. If the knowledge could not compound, the hours were not an investment. They were rent.
I wanted to see if I could rebuild the cycle so that it actually compounded.
The first structural move was boring but load-bearing. Every course I studied needed to leave behind three artifacts, not one. A set of notes. A set of exam-prep questions with answers I could defend. And a permanent reference brief. The brief is the distilled, cross-cited version of what the course taught, written in my own words, organized so I could return to it cold in two years and know where the authority sat.
That third artifact is the one that changes everything. Notes are personal. Exam prep is exam-focused. A reference brief is a future self’s working copy, written to be read by someone who does not remember the course.
I started with the material I had already taken. Individual tax first, because it was the broadest surface. Then business-entity tax. Then estate, gift, and trust. Retirement and benefits next, then IRS procedure, which covers the adversarial side of a practice: audits, notices, penalties, collections, Tax Court. Financial planning and investments after that, where half the client questions actually live in a mixed advisory practice. Audit, assurance, and controls came next, the attest-side pillar I had let go stale by being an operator for the last decade. Then financial accounting, the preparer-side sibling to the audit skill. Then nonprofits, because tax-exempt is its own discipline and tax-exempt work never waits for someone who half-remembers it. And then ethics and practice management, because Circular 230 and the AICPA Code carry real preparer-penalty stakes.
Ten subject-matter briefs. Each one is twelve to twenty thousand words of cross-cited distillation. Each one cites the authority hierarchy: Code, regs, AICPA SSTS, Circular 230, FASB ASC, the Uniform Accountancy Act. Each one is searchable, portable, and available in under a second to me and to any AI I connect to it. The subject-matter briefs live in one plugin. A second plugin sits alongside it and governs how every new course enters the system.
The second plugin is where the hard-won rules live.
There are twelve of them. They enforce things I will not trust myself to do consistently — authoritative-source citation, upstream extraction fidelity, positive-evidence handoff between notes and exam prep, a seven-category verification pass before I declare a set of notes complete, a candidate-enumeration checklist before I sit the exam, and a sixteen-category classification for every wrong answer cross-cut by a five-class taxonomy for the kind of precondition that failed.
Some of those rules landed in the system the way most rules land. I noticed a pattern once, wrote it down, and it became a rule. But three of the most important ones came from one bad afternoon.
I was working through three forensic-accounting courses in a single session. The notes for one of them ran the PDF through a default extraction pipeline that flattened three content-dense reference tables without warning. The exam prep for another one read the notes in a way that reinforced a wrong answer I had almost caught. A third exam had a compound-option question where my review under-selected by one clause because the candidate list got silently pruned before the verification step ran. Five distinct failures, all on the same afternoon, all passing whatever verification I had in place at the time.
The obvious response would have been to patch each one individually. The right response was to notice that the failures shared a common shape. In every case, the verification step was silently accepting a precondition it could not directly observe. The notes verification assumed the PDF extraction was faithful. The exam-prep verification assumed the notes were correct. The enumeration check assumed the candidate list was complete. The system was checking what it could see without checking what it was standing on.
Three rules came out of that retrospective: upstream extraction fidelity, positive-evidence handoff, and candidate enumeration. They are not about fraud courses. They are about the class of failure where a verification step silently trusts its own inputs. That pattern has a name now, and it has a response.
Most of this sits quietly in the background of my practice. The briefs come into play when the kind of question they know arrives. The workflow runs every time a new course enters the inventory. The rules catch what they are built to catch, without needing to be remembered.
What has changed is that the knowledge I have paid for is now accumulating instead of dispersing. I am not necessarily faster. I have a home to return to.
The reframe is less about efficiency than about category. Forty hours a year of CPE is a compliance expense when it leaves behind nothing you can return to. It is raw material when it leaves behind something that compounds. Same hours, same courses, same subject matter. Different balance sheet.
This is the working idea that underlies most of what I build, turned toward my own profession. An income depends on you. An asset works without you. A knowledge base you carry in your head is an income. It requires you to be there, it degrades when you are tired, and it dies when you do. A knowledge base that lives in a place your AI can read, a partner could inherit, and that gets denser every year is an asset.
There is something to the word frequency here that I am careful with, because I do not want the word to carry the meaning it usually carries. What I mean is the literal one. The thing I know is something I transmit. If I transmit it only into my own memory, it goes where I go. If I transmit it into a system that compounds, and that someone else could operate, it outlives me.
The durable asset, applied to what I know.
The workflow plugin holds the process. The subject-matter plugin holds the corpus. Both are versioned, both are source-controlled, both are designed to be picked up by a successor advisor or by a future version of me and operated without any of my current context.
I am not done. There are domains the library does not cover well yet — international tax, cost segregation, qualified opportunity zones, and forensic accounting as its own brief rather than a route-out to an adjacent one. Each new course will enter the system in the same shape. A course comes in. Notes, exam prep, brief. The brief goes on the shelf. Next time the domain shows up in a client file, the brief is where the question meets the answer.
It is the same principle I built 444 Growth Partners around.
If it depends on you, it doesn’t transfer.
I have spent a long time building things for other people that do not depend on them. It was time to build a version of that for the thing I know.