A field guide for makers with a day job
Turn unfinished ideas into small experiments with AI
You can build a working prototype in an evening now. So why is your folder still full of projects nobody has ever seen? Building got cheap. Choosing what to make, getting anyone to see it and deciding honestly whether it worked did not.
Free while it is being written, or pay what you like from $9.99. PDF and EPUB, no DRM, every update included.
Somewhere on your computer there is a folder. It might be called projects, or ideas, or, if you are being honest with yourself, someday.
Inside are the remains of things you started: a half-built app with a beautiful login screen and nothing behind it, a landing page for a product that never existed, and a domain name you bought at one in the morning and have renewed twice out of guilt.
Most of those projects did not fail. They never got far enough to fail. This book is about getting them that far: small, cheap shots, each one aimed, fired, looked at and decided.
Three in a row? Welcome. You are among friends.
With an AI assistant, the build shrank to an evening. Choosing what to make, reaching people and deciding whether it worked stayed roughly the size they were, so they are now most of the bill.
So the advantage moves from having the single best idea to a habit: many small, cheap, aimed attempts, each one actually looked at. That is the book's argument, and it comes with its conditions. More shots only help when each one is cheap, aimed and read honestly.
When nobody can tell which idea will work, the winner is rarely the best aim. It is the person who takes the most shots and looks where each one lands.
Every shot in the book goes round the same four steps, and the book is built around them, one part each.
Part 1 Open the Folder
Part 2 Aim
Part 3 Fire
Part 4 Look
Part 5 Decide
Know what a shot costs before you fire
Alex is a product designer with a day job, a Sunday podcast and eleven side projects in various states of abandonment. Through the book she takes one idea, a free page that tells podcasters whether their episode's loudness matches Apple's published recommendation, from a scribble in a notebook to a decision.
Pip is her AI coding agent: a small round robot whose catchphrase, "Done! ✓", arrives early and often, usually before anything is done. Fast, cheerful and wildly overconfident.
Both are fictional, and so are Alex's numbers. Everything else, every study and every story about a real company, comes from a published source, listed in the Notes.
New chapters land on Leanpub every Thursday until the finished book on 1 December. Each drop replaces the one before it and carries every improvement made since, so your download is always the whole book so far.
Follow the build on X: @SublimeCoderX
Three questions for you
Reply on X to @SublimeCoderX, or use the contact link on the Leanpub page. Every answer gets read.
A practical, occasionally funny field guide by Rakesh Menon for people who want to make new things often, with AI doing much of the building. It turns one loop (aim, fire, look, decide) into 30 move cards, prompts, templates and a 30-day plan, so the unfinished ideas in your someday folder become small, cheap experiments you can actually judge.
Makers with a day job: developers with a weekend and an itch, designers with a notebook full of "someone should build this", and product people tired of admiring other people's side projects. It assumes you can get something built, with or without AI, and helps you choose what to build, get it seen and decide whether it worked.
Yes, while it is being written. On Leanpub you can get it free or pay what you like from $9.99, and everyone receives every new chapter and the finished book as free updates. The minimum price is planned to rise when the book is finished on 1 December 2026.
One part every Thursday on Leanpub: Part 1 came out on 29 October 2026, Part 2 lands on 5 November, Part 3 on 12 November, Part 4 on 19 November and Part 5 with the appendices on 26 November. The finished book follows on 1 December 2026. Each drop replaces the one before it, so your download is always the whole book so far.
No. You can mine complaints, score ideas, write bet sheets, run a fake door and deliver a service by hand without writing code, and several of the strongest shots in the book involve none. You will learn, a little at a time, how to put a page online, how to count a click or a sign-up, and how to read what your AI assistant says it has done.
PDF and EPUB, both without DRM, from Leanpub. The PDF suits computers and the EPUB suits phones, tablets and e-readers. The finished book adds a move card index and a printable deck of all thirty move cards.
None of the parts are new. Paul Graham, Eric Ries, Rob Fitzpatrick, Alberto Savoia and Annie Duke have written wisely about finding ideas, testing them early, asking good questions and quitting on time, and the book credits them throughout. What is new is the price of a shot: when an AI assistant can build in an evening, their advice stops being something for funded founders and becomes a regular practice for anyone with a laptop.
No. Alex, the designer with eleven side projects, and Pip, her overconfident AI coding agent, are fictional, and so are Alex's numbers; the book says so whenever it uses them. Every statistic, study and story about a real company or person comes from a published source, listed in the Notes at the back.
AI assistants supported the research and drafting, including finding and summarising publications. Cited sources were checked against the originals wherever they could be opened, and the author is responsible for the selection, interpretation and use of all of it. The pages headed "From my own folder" are the author's own experience.
Reply to Rakesh on X at @SublimeCoderX, or use the contact link on the book's Leanpub page. The three questions that help most: where did you stop reading, what one thing would you try this week, and what was confusing, wrong or boring?
Free while it is being written. Get it now and you receive every new chapter, and the finished book.