Pareto 20, No. 10: Chandler Nguyen, for cancelling the AI subscription you are paying for out of habit

He published his AI bills, his uptime spreadsheet, and the day he cancelled a $200 plan after $1,892 in fees. His stack has changed three times this year, which is the point.

Seat 10 on The Pareto 20 is Chandler Nguyen, and it is the seat for the most common tool decision a solo operator makes: keep paying, or cancel.

Who he is

Nguyen describes himself as an advertising executive with 18 years in the industry who still runs global media operations at a major agency he does not name, and who taught himself to code at 40. He builds AI products on the side, including a podcast generator and a marketing platform, and he writes up the costs. His blog publishes several times a month; the most recent piece we confirmed is from August 11, 2026.

What to read him for

The uptime spreadsheet. I'm Dropping My $200 Claude Code Plan After Two Weeks with Codex (March 31, 2026) compiles 90 days of the two vendors' own status pages into a table and concludes: "Over 90 days, Anthropic's services experienced roughly 8-10x more downtime than OpenAI's," with a named March 25 incident. It is his reading of public status pages, not an audit, but it is the kind of evidence almost nobody bothers to collect before choosing a subscription.

The cancellation, with the receipt. Why I Cancelled Claude Max After 13 Months (April 3, 2026): "I cancelled Claude Max after 13 months and US$1,892.38 in subscription fees." Cost, reason, replacement, in that order.

The bills. My AI Bill Went From $200/Month to $45 (August 11, 2026) publishes two months of invoices, $57.01 and $33.09, alongside 786 commits shipped in the same window, after moving to a cheaper model through an open-source coding tool. His conclusion: "I am not going back to $300/month tool chains unless something forces me to." He also notes, to his credit, that the cheaper vendor has announced price increases, so the figure has a shelf life.

Splitting the job across models. Opus for Design, Codex for Execution (May 25, 2026) is a practical division of labour between two tools that most people run as one.

How to read him

His primary stack changed three times in five months: one vendor in March, a second in April, a third by August. That is not a flaw in the writing, it is the writing. Read the most recent post first and treat earlier verdicts as history. The bills are his own screenshots; nobody has audited them.

Start here

The March 31 post, then jump to August 11 to see where the experiment landed.

What he sells, and what that means for you

An AI-Native Media Operations course with a coached practice layer, credit packs for his podcast generator ($4.99 to $19.99), and a marketing platform. We found no affiliate links or sponsorships. The unnamed agency employer is the thing to hold in mind: his tool choices are made inside a large organisation's constraints as well as his own, and readers cannot check which is which. His course page names its audience as CMOs and heads of media alongside aspiring builders, so the register runs a little larger than one person, but the bills are one person's bills.

Not for you if

You want a stable recommendation. He will have changed his mind by next quarter, and he will publish the invoice when he does.

---

Part of The Pareto 20, twenty creators who get a solo operator most of the way on AI decisions. How we chose them. Facts above were checked against the creator's own pages on September 12, 2026.