BEFORE YOU PAY FOR A BIGGER PLAN
Hitting your limit? Find the one thing eating your plan before you pay for a bigger one.
I once burned 20% of my plan before lunch. The fix wasn't upgrading. It was checking the meter. I got tired of checking by hand, so I wrote the check down as a routine. You type "why am I hitting limits," it reads your usage, and it names the one job eating the plan. It's one of seven files in The Setup.
30-day refund · nothing to install · no terminal · Claude Pro/Max or ChatGPT Plus
WHAT THE ROUTINE DOES
Five words in. One change out.
- Reads the two usage meters most people never open
- Names the one thing eating it: the model, the long chat, the file you re-upload every morning
- Tells you not to upgrade unless the numbers actually say so
The plan was never the problem. The setup was.
Someone upgrades to the bigger plan, and two weeks later the tool still forgets their work, still burns through the limit, still needs the same explanation every morning.
So the limits routine comes inside a folder with the other fixes. Drop it into one Project, type start, answer three questions, and it writes your memory file and your voice file itself.
"Edit this." Anything you wrote with AI comes back sounding like you.
- Strips the twelve tells: the big words, the tidy last line, the "I hope this finds you well"
- Keeps your facts, your numbers, and your hedges. "I think" and "probably" are how a person sounds
- Two modes: a light pass that keeps your words, or a full rewrite from your voice file
"Brief me on X." A short, sourced answer before you decide anything.
- Under 300 words: the short answer, what matters, what's unsure, what I'd do, sources
- Every number sourced or cut. It says "not sure" instead of guessing
- Every brief adds an entry to a knowledge file, so the tenth one starts from what you already found
Once a month: what still works, what broke, what it cost.
- One check-in a month, three questions, one line each
- It fixes whatever broke first, then notes what changed
- That's how the setup still works on week 12, not just on night one
"Remember that." It writes it down, and never makes the same mistake twice.
- Every job ends with one question: anything to remember?
- Corrections go in a "things you got wrong" list it reads before every answer
- Hand it anything you actually wrote and it refreshes your voice file from your real sentences
THE SETUP
$49
- The folder: 7 files and the one page you read
- The interview that writes your memory file and voice file for you
- The editor, the research brief, and the limits routine, ready on day one
- A monthly check-in that keeps it working: what still works, what broke, what it cost
- A live start page with the exact clicks for Claude, ChatGPT, and Gemini, kept current
- Sample files so you can see what "done" looks like before you start
- 30 days, full refund, no questions
Works with Claude Pro or Max, ChatGPT Plus, or Gemini. Nothing to install. No terminal. The $149 version of this idea exists; this is the pieces that matter.
QUESTIONS PEOPLE ASK ME
Will this raise my limit?
No. Nothing can. It shows you what's burning the plan you have and the one change to make. If the numbers say you need the bigger plan, it tells you that too.
Do I need Claude Max?
No. Pro is fine. The interview runs once on the best model, then the cheaper model does the daily jobs. The start page tells you which to pick.
I'm on ChatGPT.
Same folder. The start page has the ChatGPT clicks. Regular Chat with a Project is enough.
Is this just prompts?
No. A prompt pack gives you 500 things to paste and nothing that knows you. This is seven files your AI reads so it knows you and sounds like you. You paste one line, once.
How long does it actually take?
Twenty minutes the first night, and most of that is you answering three questions. After that, nothing. You just ask for things.
Check the meter before you upgrade.
Get The Setup, $4930-day refund · nothing to install · Claude or ChatGPT
See everything in the folder →
Chris