Working and Building with AI
What a PM learned building real products with AI coding tools: Claude Code vs Cursor, how to check code an AI wrote, MCP and agents, and the services that came out of it.
Read the idea-to-launch post first, then move on to the tool comparison and the quality principles. The projects below are live: you can open them and try them.
Start here
- 01How I Build Products Alone with AI: From Idea to LaunchFrom idea to launch, alone
- 02Claude Code vs Cursor: A Builder's Deep DiveTwo AI coding tools, compared from the builder’s seat
- 03How to Trust AI-Generated Code: 8 Verification Rules for Solo BuildersEight checks before you trust code an AI wrote
Jump in
All posts
- Speed in the AI Era: How I Built Three Products in Weeks10min
- Beyond Copilots: Why AI Agents Changed My Daily Workflow8min
- MCP Will Become the USB-C of AI Applications10min
- The End of Prompt Engineering?9min
- Why Every PM Should Learn AI Automation9min
- AI Didn't Help Me Pass PMP. It Changed How I Learn5min
Projects
- AttractiveWebAIA website analysis tool: paste one URL and it reads the page’s metadata, links, structure, images and technology stack, then turns them into an IA (information architecture) spreadsheet you can hand to a client. DeepAnalyzer crawls several pages to map a whole site. Free to try without signing up.Project page
- First RunAn onboarding RPG for new hires and people joining a project. You join Make Feelbetter, a fictional IT company, pick one of five roles (developer, PMO, publisher, planner, designer) and make real work decisions in quests given by mentor NPCs. Free to play in the browser, no sign-up.Project page
Glossary
- AI Agent
An autonomous software entity that perceives its environment through sensors, makes decisions, and executes actions using tools.
- MCP
Model Context Protocol: An open standard created by Anthropic that enables secure integrations between LLMs and external data sources or tools.
- Prompt Engineering
The practice of structuring, refining, and optimizing textual inputs to instruct generative AI models to produce desired outputs.
- RAG
Retrieval-Augmented Generation: A technique to improve LLM accuracy by retrieving relevant documents from an external source before generating a response.