By Rafi Moshe, founder of StoriedLife

Most consumer AI is rewarded for one thing: keeping you in the conversation a little longer. It counts sessions, streaks and minutes. Humane AI starts from the person, not the metric. It is AI that improves a life rather than replacing the people in it, that adds value and enrichment rather than fear, and that leaves something real behind. We built Eva, our AI biographer, to that standard: she turns a parent’s stories into a printed book, in their own words and their own voice, something a family can hold and hand down. And the book is not a goodbye. The memoir stays alive. Your parent can keep talking with Eva, keep adding stories, and the book keeps growing. What Eva never does is drag a conversation out for the sake of another session. So ask one question of any product you are considering for your parent. What is it trying to make? If the honest answer is nothing, if it simply wants another session tomorrow, you have learned the most important thing about it. Full disclosure before we go further: Eva is the product my company makes. Test everything below against products I had no hand in building. The test still works. That is rather the point of it.
The humane AI incentive test
Every AI product has a number it lives or dies by, and that number quietly decides how the product treats the people who use it. Here is the test I walked through recently on the AI for Founders podcast, where host Ryan Estes and I talked about how we built Eva differently. Three questions. They take about a minute. What does it count as success? If the answer is time spent, messages sent or days in a row opened, the product succeeds when your parent keeps talking and fails when they stop. If the answer is something completed, a chapter, a book, a job crossed off, the incentives point the other way. What is left when it ends? A chat log sitting on a server is not something a family can gather around. A finished thing your family can hold, read and hand down is. Ask what actually remains after a month of use. Does it know how to stop? A tool aimed at a finish line can say: we have enough for this chapter, let us wrap up. A tool aimed at engagement can never say that, because stopping is the one outcome it is built to avoid. That is the whole framework. One question about incentives, asked three ways.
Why companion apps fail it
The fastest growing corner of consumer AI is the companion app: a friendly character who messages back, remembers your name, and is always, always available. I understand the appeal, and I want to be fair about it. Some lonely people genuinely enjoy these apps, and if an adult chooses one for themselves with open eyes, that is their business. But look at the job the app is hired to do, because its business depends on the conversation never reaching a natural end. A companion that helped you finish something and then waved goodbye would be a companion that lost a user. So the design nudges the other way: the good morning message, the “I was just thinking about you,” the nudge when a streak is about to break. None of that is sinister. All of it is aimed at tomorrow’s session, not at anything your family gets to keep. Now imagine handing that same app to your mother, when what your family actually wants is her stories told properly and kept. The app will happily talk with her for a year and leave you with nothing you can put on a shelf. It did its job well. It was just never hired to do yours.

What a finish line does to a conversation
Give an interviewer a book to fill and the questions change. Eva does not ask a parent to talk in general. She is building chapters, so she asks about the house they grew up in, and when they mention the bakery their father ran, she follows up on the bakery: who woke first, what the mornings smelled like, why it closed. A follow up like that is not small talk. It is an editor’s instinct, pointed at a page that needs filling. Because there is a book at the end, Eva also keeps track of the whole story. She remembers across conversations, so a detail mentioned in March can be picked up again in May, and a story told in two halves can be finished properly. And when a chapter has what it needs, she wraps up. Sessions end. The conversation does not. Eva remembers where your parent stopped, and when they come back next week or next month, she picks up the thread and the story keeps growing. That sounds like a small thing until you have spent time with software designed never to end. What comes out the other side: the stories are written up as chapters, Eva narrates each one aloud, and the finished book is printed with a QR code beside every story. Scan it and you hear the storyteller’s own voice from the original conversation, telling it themselves. My father is 90. He has been talking with Eva for more than a year, and the book on our shelf is his voice as much as his words. You can see how a conversation becomes a book on our How It Works page. Eva also talks with people in their own language, in 50 languages, and that matters more than it might sound. People remember differently in the language they grew up speaking. A lot of the memories live there.
Before you hand your parent any AI
Run the test on whatever you are looking at, including us. Ask what it counts as success, and listen for whether the answer is a finished thing or your parent’s attention. Ask what your family keeps. Ask whether it can stop. Two more things are worth checking, from the builder’s side of the table. First, who else gets to be part of it? A memoir is a family project at its best. The question a granddaughter wants asked, the photograph someone finds in a drawer: these belong in the process, and a good tool makes room for them instead of trying to own the relationship with your parent. Second, does the tool treat your parent as the author? Their stories, their words, their call on what goes in. The moment an AI starts smoothing someone’s life into generic uplift, it has stopped serving the teller. And if a product passes the test, start small. One conversation. See what your parent makes of it, and what it makes.
The conversation behind this post
I worked through this framework at more length with Ryan Estes on his podcast, AI for Founders, in an episode called “The Humane AI Playbook for Founders Building Consumer AI.” If you are weighing AI options for a parent and want to hear how a builder thinks about the incentives underneath the friendly surface, it is worth your time: listen to the episode. And if you would rather judge Eva by using her than by reading about her, you can try a free conversation and apply the test to us directly.
By Rafi Moshe, founder of StoriedLife.
Questions families ask us
What is humane AI?
Humane AI is AI designed around a human goal with an end point, rather than around engagement. It is judged by what it finishes and what it leaves behind: a completed book, a completed task, a person free to get on with their day. The incentive test in this article, what it counts as success, what remains, and whether it can stop, is how we define it in practice.
Is an AI biographer the same as a companion app?
No, and the difference is the goal. A companion app is built to keep a conversation going indefinitely, because ongoing attention is how it succeeds. An AI biographer like Eva is built to complete something: a person’s stories, gathered properly and turned into a book. The underlying technology overlaps. The incentives, and so the behaviour, point in opposite directions.
What does my parent actually end up with?
A finished memoir. Eva turns their conversations into written stories, narrates each story aloud, and the stories are collected into a printed book. Every story in the book carries a QR code that plays the storyteller’s own voice from the original conversation, so the family keeps both the words on the page and the voice that told them.
My parent is not comfortable with technology. Can they just talk?
Yes. Eva works by conversation, the way a good interviewer does: she asks, she listens, she follows up, and she keeps track of the story across sessions so nobody has to start over. She converses in 50 languages. Many families sit in on the first conversation or help get it started, then let their parent take it from there.
Does this replace our own family conversations?
It tends to start them. A parent who has been telling their story comes to the table with more of it at hand, and the finished book gives the whole family something to gather around, add to, and pass on. Eva does the patient interviewing work. The family still gets the stories, and each other.
How can I try Eva?
You can try a free conversation with Eva and hear how she interviews. Judge her by the test in this article: is she working toward something finished?
Sources and method
This post is written by Rafi Moshe, founder of StoriedLife, and describes the design standard Eva was built against. The framework is the author’s own, set out in the podcast episode linked below. No quotes from the episode are used. Product details were checked against storiedlife.ai on November 3, 2026. In line with our content plan, this post states no prices; current plans are on the pricing page.
- AI for Founders, “The Humane AI Playbook for Founders Building Consumer AI,” hosted by Ryan Estes, released October 26, 2026: https://podcasts.apple.com/us/podcast/the-humane-ai-playbook-for-founders-building-consumer/id1798265052?i=1000798237288
- StoriedLife, How It Works: https://storiedlife.ai/how-it-works/
- StoriedLife, About: https://storiedlife.ai/about-us/
Checked November 3, 2026.