Thirty-Two Is Not Too Early to Record a Life. Seventy Is Too Late.

Memory does not fail all at once. The specifics go and the summary stays, so the loss arrives with no sensation attached.

Levine and colleagues built the Autobiographical Interview to compare how younger and older adults describe events from their own lives. Younger adults produced more internal details, the episodic material tied to one time and one place: what the room looked like, what was said, what they were thinking while it was said. Older adults produced more external details, the semantic material that floats free of any particular occasion. The gap held even after the interviewer went back with structured probes asking explicitly for contextual detail (Levine et al., Psychology and Aging, 2002, 17(4), 677-689). The particulars were not being withheld out of modesty. They were not there to retrieve.

The cost lands on the small things. At thirty you can still produce the smell of a first flat's kitchen on a Sunday, what a flatmate said the night before he moved out, what you were wrong about at twenty-four and how many years it took you to notice. At seventy the story is still there, in the version that has been told forty times, worn smooth, stripped of most of what made it that day rather than any day.

A controlled study points at the mechanism. Peripheral perceptual details of a filmed episode showed substantial loss across seven days while central details held up significantly better, and the same study found that retrieving the memory soon after encoding prevented the loss of both kinds of detail over the following week (Sekeres et al., Learning & Memory, 2016, 23(2), 72-82). Film footage in a laboratory is not your life, and I will not stretch the finding past what the finding says. The direction is plain enough. Recall a thing early and the fine grain lasts longer.

There is a softening as well. The intensity attached to unpleasant memories fades faster over time than the intensity attached to pleasant ones, a robust result known as the fading affect bias (Ritchie et al., Memory, 2009, 17(4), 428-444). So the seventy-year-old account of your hardest decade will be kinder than the decade was. Usually that is a mercy. The distortion lives only in the late recording.

Then the finding that gives the rest of this its weight. Koppel and Rubin define the reminiscence bump as the increased proportion of autobiographical memories from youth and early adulthood observed in adults over 40, and Rubin, Rahhal and Poon place that period at roughly ages 10 to 30, reporting that it yields not only the most memories but the ones rated most vivid and most important, along with people's favourite films, music and books (Memory & Cognition, 1998, 26(1), 3-19; Current Directions in Psychological Science, 2016, 25(2), 135-149).

Two concessions, because the finding is softer than it first sounds. Vividness there is how a memory is rated when somebody recalls it decades later, which is not a measurement of how much episodic detail actually survived. And Koppel and Rubin's own argument is that the bump is not one fixed object: different classes of cue produce distinct bumps that vary in size and in where they sit on the timeline. Grant both concessions and the position still stands. The years that dominate a person's life story are the years whose texture has thinned most by the time anybody gets round to recording them.

You are inside those years now. Not approaching them. Inside them.

A child of yours will care less about you at eighty than about you at whatever age they are when they finally get curious.

Suppose she is twenty-nine. She has been passed over for something she wanted, or she has just found out she is pregnant, or she is sitting in a car outside a house she cannot decide whether to buy. What she wants in that car is not your considered summary at eighty-two. She wants to know what you were like at twenty-nine, when you were as unfinished as she is, and whether you also had no idea what you were doing.

That person exists once. You cannot reconstruct him at eighty, because eighty-year-olds edit without intending to. They know how the story ended. They know which of the fears turned out to be nonsense, and they cannot un-know that while describing the fear. A twenty-nine-year-old describing his own uncertainty is reporting. An eighty-year-old describing the same uncertainty is narrating, and narration always arrives with the moral already attached.

There is one thing available only now, and it is the smallest item on the list and probably the most valuable. You can record being wrong. Not a confession. A plain account of what you currently believe about work, money, your parents and your own competence, some of which is going to turn out badly mistaken. Almost nobody records that at seventy, because by seventy it has all been quietly corrected.

On whether any of this does a child any good, the evidence is suggestive and thin, and I would rather say so than lean on the finding. Duke, Lazarus and Fivush found that children who knew more about their family's history scored higher on measures including internal locus of control, self-esteem and family functioning, and lower on anxiety (Psychotherapy, 2008, 45(2), 268-272). That is a small brief report, it is correlational, and it cannot separate the stories from the kind of family that tells them. Take it as a reason to treat the question seriously, not as proof of anything. Written from the child's side, what your child will want to ask you at thirty goes through the questions people actually put to their parents once they are old enough to want the answers.

Nothing on this page is an argument from risk. You are very likely to be alive in forty years. On the current Australian Life Tables, of those alive at exact age 32, about 83 in 100 men and 89 in 100 women are still alive at 72 (Australian Government Actuary, Australian Life Tables 2020-22). Any company in this category that reaches for the alternative to close a sale is selling badly, and the ones that aim that pitch at people in their thirties will deserve every bit of the backlash coming for them.

The real argument is duller and much harder to get out of. You are going to change, more than you currently believe, and the change does not run backwards.

Quoidbach, Gilbert and Wilson surveyed more than 19,000 adults aged 18 to 68 about their personalities, values and preferences. People at every age reported that they had changed a great deal over the previous decade, and predicted that they would change relatively little over the next one. The authors called that the end of history illusion: at every age, people treat the present moment as the point at which they have finally become who they are going to be (Science, 2013, 339(6115), 96-98). Everybody thinks they have just arrived at their final self. Everybody is wrong, you included, and so am I.

The movement underneath is measurable. A meta-analysis of 92 longitudinal studies found that social dominance, conscientiousness and emotional stability increased most between the ages of 20 and 40, and that four of six trait categories showed significant change in middle and old age as well (Roberts, Walton and Viechtbauer, Psychological Bulletin, 2006, 132(1), 1-25). You do not settle at thirty. You keep moving, and the twenty-year stretch you are standing in is the stretch where the movement is fastest.

So the useful question is not whether you will still be around to record this later. Ask instead whether the person you are now will still be around, because he will not. Nothing bad has to happen for that to come true. Thirty-two becomes fifty-two through an ordinary process that erases nothing dramatically and everything gradually.

A life recorded at thirty holds material that a life recorded at seventy no longer contains. Recording young is not premature. Recording late is lossy.

Starting at thirty does not mean finishing at thirty. The first layer goes down at thirty and every later year stacks on top of that one.

This is where the age objection comes apart. Almost everybody pictures a single sitting: a camera, a list of questions, three hours, done. Judged that way, waiting is obviously the sensible move, because at seventy you would have more to say. Build a Persona in short sessions across years instead and the arithmetic reverses completely. Start at thirty-two and by sixty you are holding twenty-eight years of layers, each one recorded by the person who was actually living that year.

The difference between those two objects has nothing to do with length. One late session gives you a single narrator, with a single set of opinions, describing every decade from the same chair. Layers give you the twenty-nine-year-old on twenty-nine and the forty-one-year-old on forty-one, including the places where the two of them flatly disagree. Those disagreements are the best material in the whole record, and they are only available to somebody who started early enough to have them.

There is a practical gain as well, with nothing to do with legacy. Recording a few memories a month while they are recent means the detail is still present to be recorded, which is the Sekeres finding applied to a life rather than to a film clip, and that transfer is my inference and not a result. Waiting preserves nothing. Waiting is the mechanism by which the detail goes.

Start building your legacy Free build, 50 memories, no card required.

The discomfort is real, and calling it vanity would be a cheap answer. What has happened is a category error about who is in the room.

Recording yourself feels presumptuous because you have imagined a public. You picture strangers encountering this, weighing the significance of your life, finding the life unremarkable. Gilovich, Medvec and Savitsky ran five studies showing that people substantially overestimate how much others notice and remember their appearance and behaviour, because they anchor on their own vivid internal experience and adjust insufficiently for the fact that nobody else is paying that much attention (Journal of Personality and Social Psychology, 2000, 78(2), 211-222). They called that the spotlight effect.

There is no public in the room. Your audience is two or three people who already love you, one of whom will one day want to ask you something specific. They will not be grading whether the life was significant. They will be checking whether the answer sounds like you. Those two tests are nowhere near each other, and only the first one asks you to be impressive.

The presumption objection also smuggles in an assumption worth dragging into the light: that a life earns the right to be recorded through achievement. You apply that standard to nobody else. You would not tell your mother her stories are not worth the tape. You would not say that to a friend across a dinner table. A rule you enforce against yourself and nobody else is usually not a rule about worth. If that knot is the thing keeping you out, the case against thinking you are not important enough to leave anything behind takes the knot apart properly, and it is the parent argument this page sits underneath.

If the residual feeling is closer to grim than to self-important, that is a different objection, and whether building a Persona is morbid answers that one without trying to talk you out of the feeling. If the fear is that nobody would ever want to talk to your Persona in the first place, who actually ends up talking to a Persona is the page for that. And if what really worries you is the version of you that survives turning out to be a flattering summary rather than the real thing, read on being remembered wrong.

Record one memory a month, on your phone, for five minutes, and put the date on the file. That costs nothing and it is most of the value on this page.

Do it like this. Pick something from the last few weeks rather than a life event: an argument you half won, something a child said that you will otherwise lose by spring, whatever nearly made you quit your job, if something did. Talk for five minutes without planning first. Say the date out loud at the start. Save the file somewhere that is not only your phone.

Recent beats important, which is the one counterintuitive instruction here, and the part of it that comes from the research is narrow. In the Sekeres study, retrieving a memory shortly after encoding prevented the loss of both central and peripheral detail, while peripheral detail left alone for a week was largely gone and the central gist held up. Applying that to a life rather than a film clip is our inference rather than a result, and it points one way. The gist of a wedding will probably still be there in thirty years. The texture of last Tuesday is already going.

Then, once a year, do the harder one. Write down what you currently believe about your work, your money, your parents and your own capability, in plain language, without editing for whoever reads this later. Date the file. Do not reread the previous years before you write the new one. After five years you are holding something no single late recording can produce, which is a record of the changing rather than the changed.

Method counts too, and at thirty you already have plenty of it and no sense that you are carrying anything unusual. The things only you know how to do covers the practical knowledge that never gets written down because whoever holds it assumes everybody does. If you want structure rather than a blank page, how to interview yourself sets out a question sequence that gets past the polished anecdotes, and it works with nothing more than a voice recorder. Nothing in this section requires an account with anybody, including us.

Everything above holds whether or not you ever sign up for anything. Here is what we built, stated narrowly, because a promise made loosely by a company talking about fifty-year horizons is worse than no promise.

Idy, the guide, does the interviewing. Sessions are short and conversational rather than a questionnaire, and each one adds to what is already there, so starting at thirty-two costs you an evening rather than a project.

Your voice is cloned from your own recordings, so your Persona answers in your voice rather than through a synthetic substitute. Cloning is available on every plan, including the free one. Listening is the metered part: the free build includes three synthesised listens a month, and paid plans raise that ceiling. Where you never covered a subject, your Persona will not manufacture a fact to fill the space, and answers from what is actually there instead.

The limits, which come from building this rather than from any study. The free build is a one-time budget rather than a monthly allowance: 50 memories and 100 chats in total, after which building continues on a paid plan. At 50 memories a Persona is thin and answers narrowly, because the material you have given is the only thing there is to answer from, and the range comes from returning rather than from the first sitting. A narrow answer is what a missing subject looks like from the outside, because your Persona speaks from what you handed over and lets the person asking bring more, rather than reading out an inventory of what is absent. The voice clone is only as good as the audio, so a session recorded in a noisy kitchen produces a rougher result than one recorded in a quiet room. And nothing you say is checked against outside records, so a memory you have misremembered goes in exactly as you told the story. That is a deliberate trade, because correcting you would mean overwriting you, and you should know the trade exists before you rely on the result.

You nominate the people who may reach your Persona, that list is yours, and you can change the list any time while you are alive. How many names the list holds depends on the plan: the free build allows one trusted contact and one executor, and paid plans allow more.

Release runs through Executor Lock™. A nominated person starts the request and uploads evidence of death, that evidence is checked against the details we already hold and can be routed to manual review, and a seven-day hold runs from the moment the report is filed. During the hold you can revoke the whole thing if you are alive, which returns your Persona to standby. At the moment the lock triggers, the snapshot holds your Persona exactly as you left things. Nothing is pruned or re-weighted afterwards, no part of our system mutates the snapshot after that point, and the snapshot carries a tamper-evident hash.

While you are alive, all of this is private by default. Nobody on your nominated list has access before the lock, and there is no shared feed of what you record.

For a reader in their thirties, the part that matters is the interval. Build at thirty-two and lock at eighty-two and there are fifty years of dated layers in there, so the twenty-nine-year-old and the sixty-one-year-old are both still present rather than blended into one narrator. Accuracy deserves pressing on, and what happens if your Persona gets something wrong sets out where the failure modes actually sit. For any company promising a fifty-year horizon, the answer on continuity should be a document rather than a reassurance: commitments on continuity, data portability and what happens to your material if the company ceases trading sit in the legal framework and trust pages, and the written commitment is the thing to judge rather than the tone of this paragraph.

Your first fifty memories are free, there is no card, and the free build does not expire, so a first layer laid down at thirty-two is still there at fifty-two. What the paid plans add is worth reading later, or not at all.

Is thirty too young to start recording a life story?

Thirty is not too young. The years that dominate most people's life stories are ages 10 to 30, and those are the years whose fine detail has thinned most by the time people usually sit down to record anything. Starting at thirty means recording those years from inside them, rather than describing them at seventy from summary knowledge.

What if I record all this and then change into somebody different?

That is the expected outcome and the reason the dates matter. Personality keeps changing between 20 and 40 and beyond, so the recordings you make now become the only surviving account of a person who is gradually replaced. Layers are not contradictions to be cleaned up. Later sessions sit alongside earlier ones rather than overwriting them.

Will my life at thirty be boring to listen to?

Probably to a stranger. Your audience is not a stranger. The test being applied is not whether the life was remarkable but whether the person is recognisably you, and small ordinary detail passes that test far better than achievements do. Our own observation, offered as an observation rather than a finding, is that the stories families keep retelling tend to be the small ones rather than the impressive ones.

What does it cost, and what happens if I stop paying at thirty-five?

Current pricing sits on the pricing page rather than here, because prices change and this page is on a different review cycle. On the substance: what you have built belongs to you, and the design intent is that a lapse in payment does not destroy your material. Read the current terms before relying on that, and hold us to what they say.

Can anyone see this while I am alive?

Not unless you decide otherwise. Your material is private to you by default. The people you nominate have no access before Executor Lock™ is triggered and verified, and nothing you record is published or shown to other users.

Can I delete everything later?

While you are alive, yes. You can delete individual memories as you go, and you can delete your account along with everything the account holds. One exception is worth naming: if you have acted inside somebody else's Executor Lock, by filing a report or raising a dispute, those records have to outlive the person who filed them, so we block the deletion and tell you rather than starting a delete that cannot finish. You also control what is ever released: photos, memories and live chat are separate switches, and you can turn any of them off before the lock. What does not exist today is a setting where you decide in advance who may delete a Persona after the lock, so do not build a plan around that. If post-lock deletion control matters to you, check the current terms and ask us before you rely on anything.

Start building your legacy Free build, 50 memories, no card required.

• What would I even say to an AI of myself, for the practical version of the problem, which is sitting down and having no idea where to begin.

• What a personal AI is, and what one is for, which sets out the general case, including the uses that have nothing to do with anybody dying.

• The case against thinking you are not important enough to leave anything behind, the parent argument to this page, for readers whose objection is about worth rather than age.

The claims about memory and personality come from peer-reviewed primary sources, read directly rather than through secondary coverage: the original Autobiographical Interview work on episodic and semantic recall, the reminiscence bump literature, a controlled forgetting study on detail types, the fading affect bias research, the end of history illusion survey, and a meta-analysis of longitudinal personality studies. Every one is listed below with a PubMed link, so you can check the wording against the abstract. The survival figure comes from the survivor column of the current Australian Life Tables, also linked below.

What is not settled: whether recording your life story does your family any good. The family-history finding is a small correlational brief report and cannot show cause. Nothing here has been tested on people building a Persona, because that evidence does not exist yet. The forgetting study used film footage in a laboratory rather than autobiographical memory, and the transfer to your own life is an inference rather than a result. Where a source has been stretched, the page says so in the line where the stretching happens. The description of our own product, including the limits above, comes from building and running the thing rather than from any source, and sits on a shorter review cycle than the research does, because product behaviour changes and published findings do not.

Found something wrong, or a source misread? Send it through the contact page and we will correct the page and note the change.

1. Levine, B., Svoboda, E., Hay, J. F., Winocur, G., & Moscovitch, M. (2002). Aging and autobiographical memory: dissociating episodic from semantic retrieval. Psychology and Aging, 17(4), 677-689. PubMed 12507363

2. Rubin, D. C., Rahhal, T. A., & Poon, L. W. (1998). Things learned in early adulthood are remembered best. Memory & Cognition, 26(1), 3-19. PubMed 9519693

3. Koppel, J., & Rubin, D. C. (2016). Recent advances in understanding the reminiscence bump: the importance of cues in guiding recall from autobiographical memory. Current Directions in Psychological Science, 25(2), 135-149. PubMed 27141156

4. Sekeres, M. J., Bonasia, K., St-Laurent, M., Pishdadian, S., Winocur, G., Grady, C., & Moscovitch, M. (2016). Recovering and preventing loss of detailed memory: differential rates of forgetting for detail types in episodic memory. Learning & Memory, 23(2), 72-82. PubMed 26773100

5. Ritchie, T. D., Skowronski, J. J., Hartnett, J., Wells, B., & Walker, W. R. (2009). The fading affect bias in the context of emotion activation level, mood, and personal theories of emotion change. Memory, 17(4), 428-444. PubMed 19358015

6. Quoidbach, J., Gilbert, D. T., & Wilson, T. D. (2013). The end of history illusion. Science, 339(6115), 96-98. PubMed 23288539

7. Roberts, B. W., Walton, K. E., & Viechtbauer, W. (2006). Patterns of mean-level change in personality traits across the life course: a meta-analysis of longitudinal studies. Psychological Bulletin, 132(1), 1-25. PubMed 16435954

8. Gilovich, T., Medvec, V. H., & Savitsky, K. (2000). The spotlight effect in social judgment: an egocentric bias in estimates of the salience of one's own actions and appearance. Journal of Personality and Social Psychology, 78(2), 211-222. PubMed 10707330

9. Duke, M. P., Lazarus, A., & Fivush, R. (2008). Knowledge of family history as a clinically useful index of psychological well-being and prognosis: a brief report. Psychotherapy (Chicago), 45(2), 268-272. PubMed 22122420

10. Australian Government Actuary (2024). Australian Life Tables 2020-22. Survivors per 100,000 at exact age 32 and exact age 72: males 98,523 and 81,332; females 99,169 and 88,513. PDF

Last reviewed: 30 July 2026. This page carries no legal, regulatory or platform-policy claims, so it sits on the 90-day review cycle rather than the 30-day one.