The systems for running a one-person business on AI: what to automate first, how the pieces fit together, and the workflows behind real products. Written for solopreneurs and freelancers who want results they can ship, not theory.
Before I sold a single prompt, I ran all 75 of them against the Claude API. Not skimmed. Not “reviewed.” Ran, with realistic inputs, checking each output against pass criteria I wrote down first. Three of the prompts that survived are in this post, complete, and you can run them today. Here is the problem…
Claude is five products wearing one name. This playbook maps which surface fits which job in a one-person business: chat, Projects, connectors, Cowork, or the API, with dated 2026 costs and the five builds to start with.
A plain-English make.com for beginners guide: what scenarios, modules, and credits are, how to build your first no-code automation, and what Make costs in 2026.
The exact AI workflows I run my one-person business on: the five systems, the real monthly stack, honest costs, and what to automate first.
The third time in one quarter I rewrote a “sorry to chase, any update on that invoice?” message, I stopped and counted. Eleven of them. Six onboarding emails asking new clients for the same five details I always need. And zero testimonial requests, because by the time a project wrapped I’d moved on to the…
Claude Cowork for solo creators, tested: I ran 4 real no-code tasks and show what it did, where it broke, and what it costs in 2026.
The real Claude prompts for automation behind the content systems I built and sell, plus four rules that keep output on-voice and on-brand.
I am shipping Newsletter Engine: the Notion + Make.com + Claude pipeline from my last post, packaged as a one-time-purchase product. $67 founding price for the first 50 buyers, $97 after.
The 4-layer system I use to turn old blog posts into a weekly newsletter in 10 minutes. Real architecture, real output, runs on Make’s free tier.
How to build production AI workflows in Make.com with Claude. Skills, MCP, prompt caching, and the cost-control patterns most guides miss.
60 minutes, one plain-text file, works in Claude, ChatGPT, or Gemini. The Voice DNA framework: 12 questions you answer about yourself once, no sample uploads, and AI stops sounding like AI.
Discounting because AI made you faster is the most expensive instinct you have in 2026. A solo creator’s pricing framework, what the data actually says, and where the advice stops working.