Personal AI history · browser only
One habit worth
changing first.
Hand your own assistant export to this page and it names the single improvement to prioritize in how you ask — what that move is worth across your history, how much of it the reading covers, and how far to trust the answer. It is a reading of one person: yours. It is never a score against anybody else.
Your export never leaves this tab. No upload, no sign-in, no account to connect, no storage, and no record that outlives the page. Your prompts are measured and dropped inside one function call, and the result carries counts and rubric identifiers in their place.
This is the personal path and it is separate on purpose: it is not the organizational conversation-export import, it never joins an org chart, and nothing read here becomes part of a team grade.
Before you choose a file
What this does with your history
Every line below is a field of the shipped boundary or a refusal the contract publishes, and each one is checkable rather than a matter of tone.
Read in this browser tab. Not uploaded, not stored, not sent anywhere for analysis. No provider account is connected and no attachment is opened.
Supported files
Which exports this can read
Two shapes, both of which a person can obtain for themselves. A file that matches neither is refused rather than guessed at.
A personal conversation export as JSON, or a personal prompt log as CSV or TSV carrying a date column and a prompt column.
Getting your export
Where to ask for your history
Pick your assistant and this says what to request, what it will arrive as, which file to choose out of it, and what to leave out. You download it yourself: this page never contacts an assistant and connects no account.
Ask your assistant for your own data export — usually under Settings, in a “data”, “privacy”, or “export” section. Most arrive as a ZIP holding a conversations.json; choose the ZIP as it downloaded, or the JSON inside it. A CSV or TSV of your own prompts with a date column works too.
Read your history
What single improvement should I prioritize in my own AI usage?
Choose your export, or build the worked example first to see what a result looks like before you hand over anything of your own.
Chosen here, read here. The file is opened by this tab and by nothing else, and the picker never sends it anywhere.
Nothing has been read
Choose your own export, or build the worked example, and the result appears here.