Each mind is built from one body of work: books, letters, notebooks, poems, talks. Choose a question. Each answer comes with its source references and its Revive Score.
Two or three minds take one question. Each one answers the others from its own work. Every line comes with its Revive Score.
Most AI personas copy a style and invent the rest. Revive starts from the person's own words and keeps them in view.
Each mind starts from one body of work. For Franklin, it is his Autobiography and Poor Richard's Almanack. For Emily Dickinson, it is 629 poems.
Revive searches that work for the passages closest to the question.
An AI model writes the reply from those passages. Tap Source References under any answer to see them. Each reference names the work and shows a short excerpt, at most 40 words. Words in gold appear in the source word for word.
Revive writes one answer for each question, as it would for you, and a second model scores it. Only an answer that passes, or just misses the pass mark, appears here. Revive does not write several answers and pick the best.
Tap Revive Score under any answer to see its checks. An answer must pass all six checks:
Then it must score 80 or more of 100 raw points. Half of the points come from how well the pages support each claim. The rest come from a direct answer, the person's voice, and the person's own words.
Revive shows an answer that passes on a scale from 95 to 100. A raw score of 80 shows as 95, and a raw score of 100 shows as 100.
A near miss clears all six checks but scores from 71.8 to just under 80 raw points. It shows from 90 to 95, marked "Near miss". Revive does not show any other answer.
The minds are AI. They are not the people, and they do not speak for them, their families or their estates. The voices are synthetic. Some are cloned from recordings, and some are stand-ins.
On iPhone, tap Share, then Add to Home Screen. On Android, open the browser menu and tap Install app.