Addison W. Brown — A personal history

Chapter 01 2021 — 2022 Books, then an algorithm

What makes
things move.

Right after high school I set out to start a quantitative trading firm. I read the books, I programmed the algorithm — and I ran straight into everything I didn’t know yet.

The first thing I ever studied on my own — nobody assigning it, nobody checking — was the stock market. I finished high school and fell into it completely.

The plan, and at nineteen it felt entirely reasonable, was a quantitative trading firm. So I read — dozens of books, one after another: the value investors, the macro traders, the market histories. The one that rewired me was Ray Dalio’s. He treats the world as a machine — the same causes producing the same effects, over and over, for anyone patient enough to map them. I believed him. I still do.

So I acted on it the only way I knew how: I taught myself just enough code to program a trading algorithm. That sentence is most of the triumph of this chapter. The firm never followed. Between my ideas and my abilities sat everything I hadn’t learned yet — real programming, real statistics, the slow fundamentals — and no amount of conviction could stand in for them.

The market didn’t look random to me. It looked like cause and effect, at scale.

Dozens of books on markets — value, macro, and the people who beat them

0 algorithm programmed, in code I was learning as I wrote it

0 firms launched — this ledger keeps its zeros

Blueprint · 2022 · Never built

The machine I couldn’t build

If Dalio was right — if the world really is a machine of causes and effects — the lens didn’t have to stop at stocks. The next idea was a tool that mapped why things work: point it at any decision and it would show you the causes behind the winners, so that anyone could choose the way the best investors do. Three of the questions it was meant to answer:

Nº The question Domain

01 Where does a coffee shop actually work? Retail

02 Which house on the street is underpriced — and why? Real estate

03 Why does one video pull ten million views? Attention

“Everything” was the problem. I needed one arena where cause and effect kept score in public.

YouTube was that arena. Every experiment runs in the open, every outcome is a number anyone can read, and the machinery underneath — psychology, packaging, the promise a video makes — repeats exactly the way Dalio said causes do. I never built the tool; I didn’t have the hands for it yet, and this time I knew it.

So I did something smaller, and more honest. I opened a spreadsheet.

What it taught me

The ideas were never the problem. The fundamentals were.

At nineteen I wanted a firm before I could really program, and an edge before I knew statistics. The rest of this record is me going back for the fundamentals — on purpose, one language at a time.

Chapter 02 2022 — 2023 A spreadsheet, then MrBeast

What makes
things spread.

I taught myself why videos go viral — badly at first, then well enough that the biggest YouTuber on the planet paid for my ideas.

It started with a question I couldn’t put down: why does one video pull ten million views while a nearly identical one pulls ten thousand?

I went looking for patterns — thumbnails, titles, the first ten seconds, the promise a video makes and whether it keeps it — and I kept score in a spreadsheet. The spreadsheet, if I’m honest, sucked. The tools to build what I could see in my head didn’t exist yet. But the thinking underneath it held up, and it rewired how I watched everything.

I stopped watching videos and started watching what they did to people.

Contract · MrBeast · 2022 — 23

The credits

That way of seeing got me contract work with the biggest YouTuber on the planet. My job was ideas — take what I understood about psychology, virality and the algorithm, and turn it into videos built for a hundred million people. Five of the ones I worked on:

Nº Video Views

01 $1 vs $1,000,000,000 Yacht! 566,000,000

02 50 YouTubers Fight For $1,000,000 485,000,000

03 100 Kids Fight Parents For $500,000 347,000,000

04 Real Life Spider‑Man Vs Skyscraper 284,000,000

05 I Survived 50 Hours In Antarctica 261,000,000

Combined, across the five 0 Views · counted August 2026

Idea work is strange: your fingerprints end up everywhere and nowhere. What I kept wasn’t a credit line — it was proof, at the largest scale there is, that attention follows rules.

The next thing I bought wasn’t a camera. It was a lawn mower.

What it taught me

Virality isn’t luck. It’s pattern recognition.

The spreadsheet was crude, and the question it asked was everything: why this and not that? I never stopped asking it.

Chapter 03 2023 — 2025 Lawn care, then a little more

Bigger wasn’t
better.

Two years running a one‑man lawn care company. It paid for community college, and it quietly rewired how I think about money.

A red Ford Ranger parked behind a fleet of red Ferris stand‑on mowers and a Toro on a concrete pad — Fig. 01The original setup, early 2025. A Ranger, three Ferris stand‑ons, a Toro, a trailer — and a lot of overhead.

I started the way most people start: with the biggest truck I could justify and the most expensive equipment on the lot. It looked like a business. It felt like a business. It was also the least profitable version of the business I would ever run.

So I changed what I measured. Not revenue — profit per hour worked. Every decision after that came back to the same number: which machines to keep, which clients to let go, how to route a day so the truck spent less time on the road and more time earning.

Freshly striped lawn in front of a stone and siding house, mower handle in the foreground

Wide lawn with long, clean mowing stripes running toward a distant house

Striped lawn beside a concrete walkway under large shade trees

Figs. 02 — 04The refined setup, spring 2025. Less equipment, tighter routes, better lawns.

0 years running the business

0 % of community college tuition, paid

Every mower sold for more than I paid — after a full season of use

Most operators treat mowers like an expense. I treated them like inventory. Buy right, maintain obsessively, sell at the right moment — and a year of hard use cost me nothing. That was the moment buying and selling stopped being a side hobby and became the thing I was actually good at.

Project · May — June 2025

A little alchemy

Toward the end, I stopped thinking like a landscaper and started thinking like an investor who happened to own a shovel. The question changed from what does this lawn need to what is the cheapest set of pieces that makes this property worth more than their sum?

Brick ranch house with black trim and bare, weedy foundation beds

Gray house where mature shrubs are being dug out of the front bed

A trailer behind a red pickup, loaded with full‑grown shrubs

Addison in a blue sun hoodie with a shovel, planting shrubs along the front of the brick house

The finished house: full shrubs and fresh mulch along the entire front bed

01/05

01 · May 17 · Before
A brick ranch with good bones and empty beds. Solid house, zero curb appeal.
02 · May 30 · The find
A homeowner wanted a row of full‑grown shrubs gone. Most people pay to have that done. I offered to dig them out for free.
03 · May 30 · The haul
Dug, balled and loaded in a single afternoon. Mature plants, for the price of a shovel and a few hours.
04 · May 31 · Planting
Set, backfilled, mulched and edged the next morning.
05 · June 5 · After
Under two days of work. About three hundred dollars in materials. An estimated fifteen thousand added to the home’s sale value.

$0 in materials

<0 days on site

~$0 estimated value added

0× return on materials

It felt a little like alchemy — assembling pieces that cost almost nothing into something worth far more than their sum.

What it taught me

The most profitable version of a business is rarely the biggest one.

And value doesn’t live in the parts. It lives in knowing how to put them together.

Chapter 04 2021 — 2026 Wake Tech, then Chapel Hill

The language
of business.

I went back to community college for a single accounting class. I left with a degree, honors, and a place at UNC⁠–⁠Chapel Hill.

College and I got off to a slow start. One summer term at nineteen, a C in precalculus — and then two years of not going back.

What pulled me back wasn’t a deadline. It was a sentence. Warren Buffett calls accounting the language of business — and I was running a business every day: pricing hours, flipping machines, watching money move. So in the fall of 2024 I re‑enrolled at Wake Tech for exactly one class, Principles of Financial Accounting.

I wanted to be fluent in the language I was encountering every day.

Table 01 The record, term by term — typeset straight from the Wake Tech transcript. Nothing left out.

Term Hours GPA

Summer 2021 First try, nineteen years old. An A in computing fundamentals; precalculus: C. Then two years of not going back. 8 3.00

Summer 2023 Back for a summer while the business ran. Statistics, and precalculus retaken — straight A’s — plus a one‑credit class called College Transfer Success. The plan was already forming. 9 4.00

Fall 2023 Enrolled in two classes, withdrew from both. The business had my full attention. 0 W

Fall 2024President’s List Came back for one class — Principles of Financial Accounting — and stayed for thirteen hours: microeconomics, writing, communication. 13 4.00

Spring 2025President’s List Managerial accounting, research writing, philosophy, Spanish. 14 4.00

Summer 2025 One summer class. Trigonometry: D — retaken the next spring for a B. 4 1.00

Fall 2025President’s List Fifteen hours — macroeconomics, American government, user‑experience design, Spanish. 15 4.00

Spring 2026Dean’s List Biology, public speaking, intermediate Spanish — and trigonometry made right: B. Degree conferred May 8. 15 3.73

Total Associate in Arts, with honors — May 8, 2026. 70 3.94

Summer 2026 Calculus I, added after the degree was already in hand: C. It stays on the record — a cleaned‑up ledger isn’t a ledger. 4 2.00

The class did exactly what Buffett promised. Debits and credits turned out to be the grammar of everything I’d been doing by feel — the mower flips, the routes, the bushes. I stayed, and the record above is what staying looked like.

The degree came in May 2026, paid for by the lawn care business. And it did what that little 2023 class always said it would: this fall, I start at the University of North Carolina at Chapel Hill.

What it taught me

Nobody sent me back. That’s why it worked.

The first time, college was something I was supposed to do. The second time, it was a tool I picked up on purpose — and a tool you choose is one you actually use.

Chapter 05 March 2026 yoconjugo.com

The language
of people.

The journal a chapter back shows fifteen hours for Spring 2026, and one line of it says intermediate Spanish. This is what studying for that class turned into.

In March 2026, between accounting problem sets and Spanish homework, I built Conjugo — a language app with one conviction at its core: technology should put two people face to face, not stand between them.

It had the serious machinery — spaced repetition on the same algorithm as Anki, AI‑written sentence cards read aloud in a synthesized native voice, a knowledge map covering all eighteen lessons of the Wake Tech Spanish curriculum. But the part I loved was the battles: head‑to‑head flashcard races over live video, and a matchmaking mode that paired you with a native Spanish speaker learning English. You teach yours, they teach theirs. Both people walk away better.

It started as a little script that stuffed AI‑generated sentences into my own Anki deck. Eleven days and eighty‑seven commits later, it was a real product at yoconjugo.com, with a team of four — my girlfriend Katie designed it, and two friends, Reece and Amos, ran marketing and translation. I wrote all of the code.

No AI tutor will ever replace a real conversation.

Conjugo practice-deck dashboard with XP rank card — Fig. 05The practice desk. Decks due for review, and a rank ladder from Recruit to Master. The sidebar’s closing line: “You are one session away from a better you.”

Conjugo review card graded correct, with Again, Hard, Good, and Easy buttons — Fig. 06A card, answered. You type the Spanish, it grades you, and the algorithm decides when you’ll see it again — in a minute, a day, or three.

Conjugo battle screen for creating or joining a flashcard race — Fig. 07Battles. Each player picks the words, then you race through the deck head‑to‑head — live video of your opponent in the corner, because the point was always the other person.

Conjugo knowledge map showing curriculum units as a trail — Fig. 08The knowledge map — the whole curriculum drawn as a trail. The server died when the semester ended; these photographs are the codebase, exhumed from GitHub and run again on my desk in August 2026.

0 days from first commit to a live product

0 commits every one of them mine

0 lessons the full Wake Tech Spanish curriculum, mapped

When the semester ended I stopped paying for the server, and yoconjugo.com went dark. Some projects end because they fail. This one ended because the next one started.

It stayed dark for five months — until the August night I wrote this chapter, when I pulled the code back off GitHub and gave it a corner of Kachieve’s server. It’s alive again at yoconjugo.com — no account, no gate. See for yourself.

What it taught me

The tool was never the point.

Conjugo’s whole thesis was that software should put two people face to face, not replace one of them. I still believe that — even now that I build with AI every day.

Chapter 06 April 2026 — present Kachieve, still shipping

The next
language.

Accounting was the language of business. AI is the language of whatever comes next — so I’m learning it the way I learn everything now: by building something real and keeping score.

A month after Conjugo, Katie and I were building again. Kachieve runs on one idea: any question you can’t answer is just a stack of easier questions you can.

Upload your notes and they become practice problems that keep quizzing you until you actually know the material. And when a question stops you cold, the app breaks it down — recursively, easier and easier — until you find a rung you can stand on. Then it walks you back up, step by step, to the question that stopped you.

The concept is mine, and so is every line of code — I came up with it, prototyped it, and built it myself, putting more than a hundred million tokens through Claude Code across Kachieve and the side projects orbiting it. Learning AI was the point. Somewhere along those hundred million tokens, I got good.

You don’t learn a language by reading about it. You learn it by using it badly, daily, in public.

0 M+ tokens of Claude Code — the price of learning by building

0 users twenty‑seven accounts, minus the six that were me

~0 hours of real studying on the app — my own time excluded

As I write this, the dashboard says exactly one user is active today. Good. Every number I’ve ever cared about started out small and real — and I trust small and real over big and imagined, every time.

The app ships weekly. This chapter doesn’t get an ending yet.

What it’s teaching me

Small and real beats big and imagined.

Twenty‑one real users teach you more than a million imaginary ones — and a language only sticks if you use it every day.