Reverse engineer the software you need, personalized for you.
For example: , , , , ,
12 internal tools, each with a drawing, a parts list, a data model and a build prompt.
We pay for a giant CRM and use about ten percent of it.
Every month-end we spend two days arguing over the commission spreadsheet.
Writing a custom quote takes an hour, every single time.
Our content plan lives in four different docs and nobody trusts any of them.
Our best customer quotes are buried in emails and call notes.
Client onboarding lives in one person's head, and things fall through the cracks.
Purchase requests get lost in Slack, and nobody knows who approved what.
Every Monday I rebuild the same status report from scratch.
Nobody knows how many software tools we're actually paying for.
I can never tell who's out next week until they're already gone.
We're hiring for three roles and tracking candidates in a spreadsheet.
Customer feedback is scattered across email, calls and support tickets.
Every blueprint follows the same format, so you can compare tools and know what you're getting before you build.
One sentence from the person who has it. For example: purchase requests get lost in Slack.
A labelled drawing of the main screen, so you can see where each part of the tool lives.
The records and fields behind the tool, so the build starts with the right structure.
A prompt you can paste into Claude, ChatGPT, Cursor or any other AI builder, plus common variations.
Most business software is built for thousands of companies at once, so every team ends up adjusting its process to fit the tool, and paying per seat for features it never uses.
That trade-off made sense when custom software took months to build. With AI, a small internal tool now takes an afternoon. The hard part is knowing exactly what to build.
This site collects plans for those tools. Each one starts from a real problem and is written up so your team can build its own version, with whichever AI it already uses.
Sent on Sundays. Each email covers one tool: the problem, the drawing, the data model and the prompt.