One person.Many intelligences.

Models change every month. The method doesn't.

The harness is what sits behind everything you see here: a way of working that follows Jorge from one tool to the next. Today there are four: Claude Code, Codex, OpenCode and Pi.

One set of standards

The same rules, the same memory and the same skills, whichever tool you are in.

Each on its own ground

It doesn't force any tool to behave like another: it adapts to how each one works.

Almost a year of daily use

It started as a few files on one computer. Today it is an open project anyone can install.

The intelligences execute. Deciding is still one person's job.

No piece of work starts with “do it all”. It starts with a decision, gets split across several intelligences and comes back for someone to check.

1. DecideWhat gets built, what doesn't, and why. That part is never delegated.
2. ExecuteSeveral agents work at once, each on a bounded part of the problem.
3. VerifyNothing is accepted because an AI says so: it is tested and reviewed first.

What comes out of the workshop. All of it built with AI.

Own products, experiments and client work. Some are in production and some are not yet; here each one says which.

• PhotoHeartEnd-to-end business management for professional photographers.In production
• MezqAssisted review for pharmaceutical quality assurance.In testing
• BubbleChatChatbots you can embed in any website.Built, not launched
• WhsprVoice dictation on Android that never leaves the phone.Working, unpublished
• LevelArcYour habits turned into a role-playing game.Working, unpublished
• BTC labA lab for stress-testing trading strategies with no real money.Experiment
• Video StudioVideos made with code instead of an editor.Own use
• Client workCustom websites and tools for businesses.Private

That is what .si stands for.

Super Intelligence is not a machine that knows everything. It is what one person achieves by working with many intelligences at once and never handing over the decisions.