I have done this once before. I turned gaming from passion into content, community, and eventually business. Now I am doing it again with fitness. Content, partnerships, Team Haraldsen, and DailyBase are all part of the same direction: helping people become better.

AI makes building easier. It does not make people care. Creators understand attention, trust, storytelling, and distribution. Operators understand systems, execution, business models, and hard decisions. A lot of interesting companies will come from the middle 🚀

Claude Design just dropped and it's amazing. A solo operator can now sign 40 local clients at $2K/mo retainer and hit $960K ARR by year end. The build is free. The money is in distribution.

Everyone shipped the same product this week. Read it as one release, not three. Anthropic's Opus 4.7 can stay on a task for hours. OpenAI's Codex picked up computer use. Perplexity shipped something designed to live on a Mac Mini and operate your machine directly. Three companies, same week, same bet. AI is moving from something that helps you work to something that does the work. Most people will try the new models, screenshot the benchmarks, and move on. That misses the point. The real change is that the input to AI is no longer a prompt. It is a goal and the setup to reach it. The skill moves to context engineering. The leverage is in knowing what to hand off, how to structure it, and when to check back in. I had Claude Code scan three of my codebases today. Small refactors, quality fixes, general cleanup, nothing glamorous. Auto mode, no permission loops. It kept spinning for hours, shipping improvement after improvement while I got on with the rest of my day. The best hours were the ones I did not touch it. If you are still copy-pasting back and forth with a chat window, you are using 2024 tools for 2026 work. The upgrade is not the model. It is how you set the work up. Links 👇

OpenAI — Codex computer use dub.sh/cv4X0CO