A long interview, in small pieces
The MemoriesBox guided interview is 380 questions across 21 chapters of a life, each read aloud, each with follow-ups that dig into the answer, run in short sessions across months and answered in the person's own voice, so a life is covered in depth rather than in fifty-two paragraphs.
Facts on this page verified
What it is
MemoriesBox asks questions and records the answers. That is the whole mechanism, and everything else in the product is built on top of it.
What makes it an interview rather than a questionnaire is the follow-up. Every one of the 380 questions carries its own smaller questions underneath — the ones a curious grandchild would ask next — and they are put one at a time, after the answer, in the same voice. "What did your father do?" is followed by "What did his hands look like?", not by the next item on a list. That is how an ordinary good conversation goes, and it is the difference between an hour of material and a paragraph.
Sessions are short and there are many of them. Twenty minutes is a reasonable first session. Most stories are not told in one sitting, and the ones worth having are almost never told in the first one.
How the follow-up works
Three things happen in sequence, and none of them is complicated.
- The question is read aloud, in a natural voice, and the person answers out loud. Nobody types. Nobody writes an essay. If the person did not catch it, they say "repeat" and it is read again; "next" or "skip" moves on. Hands-free, because a hand holding a phone is a hand not free to gesture at a memory.
- The answer is kept as audio and turned into text. Both stay. If what was said was "sorry, what?" rather than an answer, the question is put again instead of being filed as answered.
- The follow-ups come next, one at a time, written in advance for that question: the details a child would want, asked in the order a child would ask them. Then the session ends before anyone is tired, and the next one picks up at the next unanswered question.
What the interview does not do — yet — is compose a new question from the content of an answer. The follow-ups are written, not generated, and one session does not remember what was said in the last. We say that plainly because the difference matters: a written follow-up is the same for everyone, and it never invents a detail to ask about that the person did not say. What does look back across sessions is the hundred-answer account: every hundredth answer, the interview stops and tells the person, from their own transcripts, what they have actually built — what is now kept, and where they seemed to want to go deeper. No praise, no score. A reading of their own words back to them.
The other half of the method is what does not happen. There is no weekly deadline and no counter showing how many chapters remain. A person who records four sessions in a fortnight and then nothing for two months has not failed at anything; the one thing that reaches them is a single emailed question, at most once a week, which they can stop with one click. It is described on /features/companion.
Why fifty-two emails produce different material
The dominant format in this category is a question a week for a year, printed at the end. It works, it has made a lot of good books, and it has a specific limit: each answer is written alone, so nothing follows up.
| Weekly prompt services | A long guided interview | |
|---|---|---|
| How the question arrives | Email or SMS, one a week | Asked out loud, in a session |
| How the answer is given | Written, or recorded and transcribed | Spoken, and kept as audio |
| What follows the answer | Next week's unrelated question | The follow-ups written for that question, one at a time |
| Typical length of one answer | A few paragraphs | As long as the person keeps talking |
| Who does the editing | The person, while writing | Nobody. The hesitations stay |
| What exists at the end | A hardcover book | Hours of the person's own voice, plus the transcripts |
| Can it be asked something new later | No | Yes |
The examples, with prices verified on 17 August 2026: StoryWorth sends weekly email or SMS prompts and prints a hardcover, from $59 a year. Remento is $99 a year, includes a book, and rewrites spoken answers into written prose with its Speech-to-Story feature. Storii telephones the person, including on a landline, and works through more than 1,000 fixed prompts at about $99 a year.
Be fair about two of them. Storii's phone calls reach people who will never open an app, and that is a genuine advantage until our device ships in 2027. HeritageWhisper, at $39 a year, runs a live adaptive AI voice interviewer called Pearl, so it is not a prompt list either, and it is a third of our price. If the interview itself is all you want, HeritageWhisper is the cheaper way to get one. What none of the eleven services we compared does is keep the person's own voice as a model and let the family ask it something later. That comparison is on /compare.
What it will not do
- It will not write the story for you. The transcripts are what was said. We do not smooth spoken language into literary prose, because the way someone actually talks is most of what a recording is for.
- It will not push. No streaks, no notifications, no daily count. One emailed question a week at most, to someone who stopped, with a one-click opt-out. See /features/companion for why that restraint is deliberate.
- It will not interview someone who has not agreed. The person has to know they are being recorded and take part on purpose. See /trust/consent.
- It will not insist on chronological order. People remember by association, not by decade, and an interview that fights that gets short answers.
- It will not check facts. If a date is wrong, it stays wrong, and the family can note it later. An archive of what someone said is more useful than a corrected version of it.
- It will not replace a human interviewer. For a difficult relationship, a reticent parent, or a person who is very ill, a person in the room is better. That is what VIP Complete is, from $790, and it is also what trylifestory sells at $595 for a two-hour human video interview.
What it's for
The situation is specific. Somebody is 78 and perfectly well. There is no deadline, which is exactly why it never happens. 53% of Americans cannot name all four of their grandparents, and 51% say they want stories about what their ancestors were like when they were young (OnePoll for Ancestry, March 2022, n=2,113). The stories are not lost in a dramatic event. They are lost by nobody asking.
It is also for the harder version. After a diagnosis, sessions get shorter and the order changes: voice first, then the things only that person knows. /guides/not-much-time covers that properly, without a sales pitch.
StoryCorps has recorded more than 700,000 people since 2003 and its archive sits at the Library of Congress. The format it proved is two people, a microphone, and a conversation that is allowed to go where it goes. That is the format, not a form.
Questions people ask
How long does a MemoriesBox interview take?
There is no fixed total. There are 380 questions across 21 chapters, sessions are as long as the person wants, and they accumulate over months. About 30 minutes of recording is enough to build a good voice model, usually reached in the first two or three sessions, long before the story is done.
Can the interview be done in Russian, Ukrainian or Spanish?
Yes. The 380 questions exist in English, German, Russian, Ukrainian and Spanish, each set written for its language rather than translated. The site runs in the same five languages, and the digital self answers in all five.
What if my father runs out of things to say?
That is the normal case, and it is what follow-up questions exist for. When someone says there is nothing to tell, the problem is almost always the question. /questions/when-they-deflect is written entirely about that.
Do I have to be there while my mother records?
No. She can record alone, and many people prefer it because they talk more freely without an audience. You can also sit with her, which changes what gets said, usually for the better.
Related: What MemoriesBox does · How the voice model is built · How the sessions become chapters · Consent, and who agreed to be recorded · Where your recordings live · The full comparison table
If you want to start while there is no deadline, begin here. $139 a year, first 14 days free, cancel anytime.