Yourself First

Pauses easy lookups and everyday decisions before they reach ChatGPT, and counts the tokens you kept.

Yourself First AI was built for hard problems: research, analysis, writing, code, and work that needs real judgment. A lot of everyday use has drifted somewhere else. People ask it to plan the day, pick dinner, decide what to wear, or look up a fact that a short search would answer. Over time those habits thin out the skill of thinking for yourself. Yourself First is a pause before that habit. When you are about to send a prompt to ChatGPT, the extension classifies it on your device. If the prompt needs real work — analysis, research, synthesis, or material you brought — it goes through. If it is a quick lookup or an everyday decision at the start of a chat, Yourself First holds it. You can look the answer up yourself, decide it yourself, or still send it after a short pause. Follow-up questions inside a conversation that is already underway can still go through. Nonsense and abusive language are stopped either way. Keep the hard problems for AI Use the model where it earns its place: long analysis, drafting from your own material, technical work, research that has to be put together rather than copied from the first search result. Questions you can answer with a search, a definition, or a minute of your own thinking stay with you. That is how the skill stays sharp. Spend tokens on work that needs them Every easy prompt still costs tokens. Yourself First keeps those for the tasks that actually need a model, and shows you an estimate of what you held back. When a prompt is something you can find on the web, the extension can open a normal search so you get the answer without spending a generation on it. Also on Google Google search often puts an AI answer above the results. Yourself First can hide that block, so a lookup stays a lookup — links and sources, not another generated summary. Settings include a simple way to switch your default search to Google Web, which does not show that overview. Private by design Classification runs on your device. Prompt text is not saved and is not sent to a Yourself First server. What is stored are counts: prompts held, prompts you chose to send anyway, and estimated tokens kept. Those counts can sync with Chrome when you are signed in.