Keep what works.
Throw the rest.

Own your taste.

Stop copy-pasting prompts between tabs like an animal. Ask once. Trinity fans it out to Claude, GPT, and Gemini. Shows you where they split. Makes the call the way your taste would.

Just an MCP and a Chrome extension — no new app, no cloud, no API key. Your transcripts never leave your machine. Day one, you run councils: ask once, all three labs answer in parallel, and the chairman hands you one synthesized verdict — plus exactly where they split. That works the moment you install.

Then it compounds. Your lens builds quietly in the background from the transcripts already on your disk — the pattern in how you rephrase, judge, and decide. A first pass takes minutes, not days, and you never wait on it: councils work without a lens, and it upgrades them once it's ready, sharpening every time you use it. Once it's there, every council runs through your lens — the chairman stops returning the generic answer and starts returning the one you would have picked.

And when a new model lands, score it against your taste. Not a generic leaderboard — a per-axis number on your own hardest questions: "Opus 4.8 gave me what I wanted 84% of the time." A benchmark no lab can publish, because only the layer above the labs sees your judgment across all three.

• Run a council across all three labs — day one
• See exactly where the models split
• Your lens builds from transcripts already on your disk
• Then every council runs through your lens — the answer you'd pick
• Score any new model against your taste the day it lands
• Every override sharpens the local router
• Embedding model + index stay local