Thoughtprint
AI insights, grounded in what you actually said
Thoughtprint does not try to define you from one entry. Anything it shows is tentative, labelled as its own, and kept apart from your words — and it should link back to the entries behind it.
On-device first
A model on your phone — not a server — handles things like a follow-up question about what you just said. Nothing leaves your phone for that.
Cloud only if it's worth it, and only if you say so
The on-device result has to clear its own bar for quality first. Only below that bar, and only once you've turned cloud processing on, does a request go to Thoughtprint's server instead. Cloud AI is off until you turn it on, and you can delete your cloud copy anytime.
Insights can quote you
Insights can link back to the entries they came from. When Thoughtprint notices something that may keep coming back, look for View evidence to read what you actually said, in your own words.
What it won't do
It won't score you, diagnose you, or guess what you feel without pointing at the entry that says so. Generated observations are worded as tentative — "may" or "might" — until you confirm them.
Still being tested
Deeper theme-tracking across months of entries is still being tested on real archives and isn't part of the beta yet. What's described above is live today.
Not therapy
Thoughtprint resurfaces your own voice reflections. It is not therapy, counseling, medical advice, or a diagnosis.
Questions people ask
- Does the AI make things up about me?
- No. Generated observations are tentative and labelled as Thoughtprint's own, kept visually apart from your words, and should link back to the entry they came from.
- Where does the AI run?
- On your phone first. Cloud processing is off until you turn it on, and you can delete your cloud copy anytime.
- Can I ask my journal a question directly?
- Yes — on-device search answers from sentences you've actually written, each one you can open and check against the entry it came from.