Because forgetting is the default and remembering is a product feature. Models do not retain anything between conversations on their own; the app around the model decides what to save, and each app saves it only for itself.
Inside one product, memory features carry some things across chats: ChatGPT saves selected facts, Claude projects keep instructions, Gemini keeps saved info. Between products, nothing persists at all. Every new tool meets you as a stranger because your context is stored in the last tool’s silo, and silos do not talk.
Re-explaining yourself burns the first ten messages of every new relationship with an AI. Worse, it quietly locks you in: after a year of teaching one assistant who you are, switching to a better one means starting broke. Your own history becomes the reason you stay with a worse product.
The fix is to stop storing yourself in their products and start carrying one file of your own. A pupil is that file: built from a free four-minute read, pasted into any AI in about two minutes, curated line by line, revocable in one tap with offline proof. New chat, and it already knows how to meet you.
Take the free four-minute read, get your word and your one-of-a-kind eye, and add your pupil to ChatGPT or Claude in about two minutes. Revoke it all with one tap, anytime.
Take your read, free Watch the revoke demoContext windows reset per conversation, and persistent memory is a deliberate product layer with real privacy weight. Companies ship it walled because that is both safer for them and stickier for you.
Plain pasted text works everywhere today, and consent-scoped grants like Sign in with Pupul layer verification and revocation on top for apps that integrate it.