That sounds like a really cool project and a really interesting way to preserve family history.
I feel like i don't know how to emotionally react to the AI part of this story. To begin with, it is fundamentally cool we have technology like that. At the same time it felt bittersweet, like an artisan being put out of business by the factory. The first part of the story felt like much of the love was in constructing everything by hand, it seems almost sad to lose that. There is also an element of dystopia in how the AI was able to cross reference everything, bank statements, ticketmaster recipts, shazam, etc. It is kind of unsettling the power of it all.
Not sure where i'm going with this comment. Its a super cool project, thanks for sharing.
I am usually grossed out by AI when it fakes humanness, but not here, I think.
Steve Jobs saw the computer as a bicycle for the mind, a way to enable us to do more and be more. This is the metaphor against which I measure all technology.
I think that in this case, it helped someone make something deeply human by abstracting the tedium away. It did what a computer should do: aid a human with their task.
Technology has been feeling like a devil's bargain for a while now. This was a rare glimpse of how I used to see tech, and of why I was so excited about it.
Yes! What do we want technology to do for us? In my opinion one important promise that was never fully realized is helping us to live a more enriched life. Social media does help people stay connected but it brings a lot of negatives that are well-known at this point.
If you build this encyclopedia as a purely robotic collector of facts that nobody reads, it’s probably more dystopian horror.
If you build it as a fun inner loop that reconnects you with people and memories and makes you more human, then it’s great.
We should endeavor to craft experiences that do the latter; right now we are back in the hacker days when small teams can build big new ideas, and big tech hasn’t taken over.
I don’t even know how much it helps people stay connected anymore since the shift to mobile. I was in an antique store recently and came upon a vintage “correspondence desk” which is basically a desk specialized for sorting and preparing mail. Back when people used to keep in touch by writing letters to each other this is what people with active remote friendships did. They’d spend a couple of hours at this desk reading letters they’ve received and sitting down to compose replies.
This is basically how social media was when you needed a computer to go online. You’d sit and sift through your feeds and there’d be message chains you respond to. You’re not really doing anything else while you’re doing that and you’re putting it out of mind once you step away. When Twitter first started getting big it was sort of a joke that people are talking to you while on the toilet. The idea that you were only ever half engaged with anything you’re doing was remarkable enough to be worth pointing out instead of taken for granted.
It’s just a lot more focused and intentional when you’re dedicating time and headspace to the task instead of “microdosing” on connection via a dopamine lottery. Even if you took away the ads and the interpolation of creator-content crowding out the connections with people you actually know, I think designing for an infinite scroll just inherently makes the thing less human-centered. It sets it up so you’re interacting with these atomized bits of ‘content’ rather than people.
That is because platforms both enable us and exploit us, they exploit both those who create/comment and those who read. They perform a necessary function but extract the value from it for their own good.
I agree. I do admire the concept as a framing device to engage with your family history, but the "AI" part strikes me in a wrong way.
There's a comment by bonoboTP in a sibling thread about the emotional complexity of a project like this. There are many ways to narrate a life story: many traumatic episodes and feuds better left forgotten, different framings, and all that emotional labor of trying to choose what and how you want to remember.
The use of LLMs for creating a shared view for some information isn't inherently morally dubious-processing and storing data is what computers have been doing for generations-except for the privacy implications, but letting this projection of a mega-corporation usurp the role of narrator for such a deeply personal story feels wrong on an instinctual level.
Personal wiki's impersonally compiled. I gauge LLMs for the extent they fray the social fabric that hold people and society together. And the way AI is introduced for max disruption causes me to be generally against the technology, despite that there are also obvious merits. Here it depends on how much value, say, a family gets out of reading in their family encyclopedia.
It is a nice idea, and I can imagine how it may serve to strengthen the family's social cohesion, in a time where everyone is busy doing the rat race. Though I'd not use it as "encyclopedia", a cold-hearted fact recorder, more like more a social-focused "Our Family Diaries" and would be much better served by family members writing down their own experiences.
I understand the bittersweet feeling because I did all the editorial work for the wedding page and the first few others and I did feel like a historian trying to connect the dots after stumbling into some primary/secondary materials and spending a couple months doing all the editorial work
after I began experimenting with agents, it sped up my process that otherwise would've taken many more months for every page given that the kinds of data sources also increased over time
I did still spend significant amounts of time like a wikipedia contributor would deciding on what to keep, enhance or delete from the page based on my own personal preferences and what I was comfortable with seeing on the page
the dystopic feeling is also fair and unsettling, I think this ironically also made me realize how important safeguarding my personal data is, we leave digital trails of ourselves everywhere so a powerful agent can string them together to create a story of who you are
100% agree I just had exactly the same reaction. I love the idea and would definitely like to do the first part e.g. documenting key people (family members and other important relations etc), key events like weddings etc.
What a lovely resource, especially if it reflects stories and recollections given by the subjects themselves.
The idea of having AI do it all is really off-putting IMO. For a number of reasons:
1) You lose the curation. You'll inevitably see a bias towards documenting based on the quality and availability of the sources as opposed to the significance of the event. E.g. you might not have much info about some really special childhood event you or someone else remembers, but does that mean it shouldn't be documented? Conversely, I don't want a 10,000 word essay on (to quote one of the titles from the post) "The 3D printing saga" -- just because I happen to have hundreds of WhatsApp messages on the subject.
2) I don't want to fact check every detail. Personally, I think if grandad (RIP) would have told me he one surfed a 20ft wave of the coast of Filey, Yorkshire. I don't need a correction that it was unlikely to be that high. If these things are partly being done "in memoriam" then I think it's really important to preserve the experiences, stories and recollections if the people we're trying to remember. Dates etc are fine to validate and correct. But there's an element of subjectivity to memories that is really special IMO. What even is reality at the end of the day? We're all just one big collective story we tell ourselves.
3) It feels soulless. Enough said on this one, I think people know what I mean
I don't know that "preserve" is the correct term here. It's certainly an interesting way to collate family history, but this encyclopedia will last only as long as the OP is interested in and able to maintain it. Once OP gets bored, or falls ill, or dies, unless there's someone else interested in it, the history is gone, reverted back to oral memory.
If instead, the OP had collected this information into a physical book, when they get bored or sick or dies, the book gets pushed into a closet or garage, waiting for some grandchild, nephew or niece to pull it out and rediscover the family history. And if anyone has even a slight interest in continuing the legacy, they don't have to know how to use a computer, just some basic scrapbooking skills, which we all learned in kindergarten.
I share this dilemma too. Just a thought -- I feel less okay with AI processing "data made for humans" (i.e. the images themselves, audio recordings of speech) and more ok with it processing "data made for software" (exif data, shazam logs).
I had the same reaction, but to me, it seems like a downside of automation and scale in general. I'm analogizing in my head to experiences when I was a teenager I used to go to skid row in LA and hand out cash to random homeless strangers because that felt like a good thing to do, but as a late-30s adult decades later dealing with spine injuries where walking was my only available form of exercise, I lived in another downtown with a large homeless problem and became overwhelmed any time I went out for a walk and never gave anything to anybody, simply because there were so many people asking that if I stopped to pay attention to all of them, I'd have spent all of my time doing that and none of it actually walking. Or the businesses that feel like it's well-meaning and harmless and helpful to them if I can give just 30 seconds of my time for feedback on how I felt about the transaction. Fine when that's really just 30 seconds here or there, but when it's every single business I've ever made so much as a two dollar transaction with over the past decade, now it's 30 seconds time 500 businesses a day, and if I paid any attention to their e-mails and texts, it would be all I ever do.
Similar with this, when you're hand curating old photographs and personally interviewing relatives, you're learning something. You're deepening relationships and your own personal understanding of these people you love, spending time reflecting on your own life. But when you send an LLM at it and it produces the volume of real Wikipedia, now an automated process is producing more text than you can ever possibly read if all you did for the rest of your life is read.
there is an activation energy cost to so many activities - so those things just never got done. many times it is because the cost-benefit wasn't clear at the start (unknown unknowns) so it never got done. kudos to op for experimenting and showing us one way of making something like this happen.
Think of a woodworking project. Compare doing everything old-school by hand vs using modern tools to go faster. Think about the end product being just an item with a function vs it having some design value or even craftsmanship value. Does the parallel work?
why is there need for an emotional reaction? It's just a tool. Philosophically, it's no different than using photoshop to touch up old photos. It's just a more "high-tech" version.
My grandfather left five moving cartons of diaries written by typewriter, every single day of his adult life documented, an achievement, to be sure. When he passed away he left them to my mother to be scanned, transcribed and moved online, something that weighed her down for the last 15 years of her life.
When he died there was no way of transcribing them automatically (there still isn't really). The boxes stood in my mothers already cramped attic for 13 years, then she got cancer, and she felt a need to finish up things, so she got a scanner and started just scanning.
When my mother died she had scanned about a thousand pages, not transcribed, not anything.
The text in the diaries were fun at times, sometimes depressing, seeing how little he cared about my mother and his family was crushing.
My brother wanted to continue the scanning but I told him that I wanted to throw the diaries away. He kept half a year of writing around his birth (there's at least a sentence) and my uncle did the same, then we just watched it all burn (not literally, we threw it away at the recycling centre).
Not everything needs to be preserved. I'm happy some parts is preserved. I'm happy that those diaries are ash.
I do something similar with my wife; at the start of every year we take around 50 sheets of paper and bind them into a little notebook. The binding cloth we use is usually a combination of clothes that tore, fell into abject disrepair the previous year. She then finds little things (ex: matchbox from a restaurant we visited and loved) and decorates it.
Throughout the year we keep writing in it, things we learnt, discords we had and how we resolved them, recipes I experimented with and we loved, random thoughts; basically anything and everything. And that little diary becomes an embodiment of that year.
I would also like to point out the manual labor and writing into it and not using an obsidian++-AI-auto-categorizer-3000 is simply because it feels like it's worth something, it's a nice little routine we have at the start of every year, and it's really fun reading these from 2-3 years back.
Also the kids will have some really interesting reading a few years down the line.
I imagine a future where this becomes a family tradition that transcends time, knowledge from different generations, living different lives all nicely recorded in these codices. Something about this whole thing feels really beautiful to me.
I am starting to do this with actual physical books. I have thousands of photos going back over my life, and I am putting them together in Scribus to then go and print a physical book for each year or event or holiday along with some relevant text.
Ideally square books that can go on a coffee table. At least when I am dead there will be some part of my existence in physical form, unlike all the digital things we spend decades creating.
I might put a SD card taped in the front of each one with a video too, so someone can watch it in the future.
As a separate aside, I also found old Canon photo printers (Selphy models) on ebay for about £5! Some need the little white gear inside glueing back on (there's a video on YouTube about it), and they DO NOT work with Windows anymore, but gutenprint supports them fine on Linux, so I have been printing photos (postcard size) at home. The colour isn't going to win awards and the saturation needs boosting slightly in the printer options compared to default, but it's a wonderful way to finally get some photos from trips on the walls.
Extremely cool. I'm into genealogy and can trace my family 10 generations back (250 years) to their arrival to Argentina. Documentation is lost or lacking once you reach Europe, other branches of the family with more recent arrivals to the country are very hard to trace. In part due to mismatching surnames and in part due to the wars.
We have started asking old family members to send us whatsapp audios with tales and things they remember from long-passed away family members; and what was life like in the 1930-40-50s. I want to start organizing all the info and data we have, my father has built a couple family trees, but this wiki format is indeed very promising. I'll keep an eye on this and see if we can use it.
> On top of that, I exported my location timeline from Google Maps, my Uber trips, my bank transactions, and Shazam history. I would ask Claude Code to start with the photos and then gradually give it access to the different data exports.
Is anyone else feeling uncomfortable with that? It is a great project and I don't want to bash it with general concerns, but sharing all my financial and location details with any service seems like opening the floodgates to my house.
My concern is not even strictly related to AI, but about sharing all my most private data with any service. There is always a significant chance all of it is leaked sooner or later.
I started running an private MediaWiki instance during the pandemic as I wanted something with a nice editing experience rather than editing markdown documents. I almost went with a self-hosted Confluence instance :P
Mediawiki is very very nice and it has a lot of cool features i've been loving over the years.
One of the things i like the most is the ability to embed a PDF document so that it's both downloadable and browsable from the wiki page itself (it embeds the browser pdf engine).
This means that i can, say, have a page for my microwave oven and have its user manual easily available.
Lately I've been thinking how to connect that with some LLM, most likely there's a chance to do some interesting things :)
I like the idea, but I'm curious where to draw the boundary. If only I can read it, it can be my full recollection of everything. If I add my siblings, parents, cousins, etc, then some articles become painful or controversial (e.g. divorce, disease). Or I just ommit all the unhappy parts.
This is perfect example how to solve problem which should have been solved in our digital lives already decades ago. The issue is that our personal lives have been outsourced to social media platforms (looking at you Facebook...)
Obviously not everyone has same needs or wants to retain stories and memories but lack of social structures and solutions seems like weird mishap.
> Her face lit up as she narrated the backstory behind the occasion, going from photo to photo, resurfacing details that had been dormant for decades.
I had started something similar with my mom over Christmas in '24. About half way through the collection she asked to stop. We would do the rest on her next visit.
Well. It never came to that as she passed away completely unexpected in March last year.
I’ll never get the chance to record the other stories. The stories from the second half of the photo collection.
So I started pointing Claude Code at other data exports. My Facebook, Instagram, and WhatsApp archives held around 100k messages and a couple thousand voice notes exchanged with close friends over a decade.
> The model traced the arc of our friendships through the messages, pulled out the life episodes we had talked each other through, and wove them into multiple pages that read like it was written by someone who knew us both. When I shared the pages with my friends, they wanted to read every single one.
This is a stunning violation of the privacy of your friends.
If someone uploaded every single private conversation I had had with them to Anthropic, they would no longer be my friend.
This is awesome, dude. I love it. One of my personal points of friction is that I want almost all of my life to be public in whichever way it is, but I don't want to subject my friends to that without asking, and my life is pretty intertwined with that of my friends. I suppose I could add a new namespace and protect it, but for now I just keep my private notes in my Google Drive and my public notes on my blog. My blog etc. is in Mediawiki and I expressly like the interwiki linking form so it's seamless what's in the Wikimedia universe. The best part about the interwiki thing is that anything from the Wikimedia world can directly be hotlinked on your wiki too. That's really fun.
I do like the idea of building up this history of people, and maybe when my parents pass I'll make theirs public and so on. Great work, dude! I love it.
Fascinating idea! So wish I had parents and grands around so I could record all the old family stories that were told over the years. If you readers take no other action after reading this post, start recording your own oral history/stories; yours, parents, kids, friends, etc.
I like the overall project and goal. I personally would like a way to ask questions to those that are living or have a template that I can use for filling in family history.
Secondly, the home page seems like I am reading a family history page more than talking about the software. It is confusing to me.
First off, this is a super cool project. Kudos to getting it started and up and running. Family history, and memorabilia and organizing all that stuff is super challenging, and just having a proper site to look at is worth its own effort.
Unlike some of the comments herein, I find this as a perfect use of technology in service of users. (Yes, with some limits). I liken this to Maggie Appleton's Home-cooked Software model [1], wherein barefoot developers use technology (AI-driven or not) for writing apps for their own purposes, nominally for a user base of 1 (or very few), with possibilities of expanding to a few dozen.
In that vein, I'm a barefoot developer, and much of the software I have written in the past few months (with help from Claude, ChatGPT) is very much for that tiny user base of a few dozen (=mostly me, if I'm honest). And that is perfectly fine by moi.
I wrote a utility to organize roughly 100K+ photographs (and videos) neatly into dates/location, both for backup, as well as to maintain the memories in an organized fashion. Asked Claude to lookup location by EXIF; haven't yet asked it to "guess the location by photo" when no GPS info existed in the EXIFs. But I think I might do that.
(no, I haven't asked Claude to go thru my Uber trips or bank statements! I draw a line there!)
That is why the OP's personal wiki made me so excited - because the whole output resonates with me.
Like a few commenters mentioned their journaling experiences. I've started doing that with some of our trips (mostly post pandemic), both to remember our experiences better, and to come back to them as needed. The simple act of writing down places visited, experiences had (mostly hikes, mountains climbed, meals consumed in distant foreign places, weird/quirky experiences) causes them to be fresh in one's memories.
Thanks, this was a great project, and a great reminder as well.
Eventually people will begin to put their memories of their lives on their sites. At first they will be like a grand photo-album, but then things will get more complicated. People will begin to journal their memories (cleaned up and detuned or exaggerated). This will be called their Book Of Life and when you fall in love or become very close with someone you will share your “Book of Life”.
Jann Burner, "The book of life", December 26, 2008
Amazing work there @jrmyphlmn . Very recently I was thinking about how to preserve photographs I have collected over last 10 years in my external drive and instagram and unchecked SSD cards. Time to weave them all together. Bonus for me to be the 100th github starer on your repo
When does a man die? When he is shot in the heart by a gun? No. When he eats a soup made of a poisoned mushroom? No. You cannot kill me. A man does not die until he is forgotten! I might stop breathing today, but I will live on in memory.
I like the idea. I like the way it uses existing framework. If I was going to offer a suggestion, I would try to incorporate a way to use local inference ( or is that accomplished through opencode? I have not used it yet so I don't really know ).
This is a really fun project and the family interview transcripts + LLM workflow feels like a genuinely good use of the technology.
I would probably have ended well before "I exported my Google Maps location history, Uber trips, bank transactions, and Shazam history."
Aside: I've started seeing lot of AI projects in this category say some variation of:
> it runs on your machine, your data stays with you, and any model can read it
I don’t think people fully appreciate the tension in those claims, especially when the model most area reaching for is Claude or GPT or Gemini. I think these things need more precise language about where data actually goes and what tradeoffs users are implicitly accepting.
This is very interesting. As a person who meticulously daily diaries into Obsidian, my hope is to have a relatively accurate look-back at things I've done in the past. And having a Wiki to show that, feel very interesting now!
I actually spent a weekend last yr doing something similar. Went through a box of old photos with my dad and wrote things down before the stories were lost. Never thought to structure it as a wiki though. Way better than the Google doc I ended up with.
The bank transaction + location cross referencing to figure out which restaurants you went to is pretty cool. Would be great if this could pull in social media exports too. Point it at your X, IG, FB archives, let it draft pages/content from that.
Any plan for a timeline view? Wiki format works well for depth but sometimes you just want to scroll through a year.
I did the same thing and came away with a different opinion.
The MediaWiki server died and I had backups, but... literally no one in the family would've tried to resurrect it.
They knew I'd worked on genealogy for a while but I don't think anyone would've thought to rebuild a linux box covered in dust and somehow find an old MediaWiki install on it.
I should've made simple markdown files with images in an image directory and printed out copies. That's a legacy. A consolidated, easy to drag from grandpa's house and throw on a shelf and flip through, even in 2097.
I quite love this idea. Both of my grandfathers died last year and going through their early memories was amazing. I would love to be able to preserve that and make it shareable with the rest of my family.
I just completed an 18 month travel sabbatical with my wife - would this be useful to catalog and cluster all of our photos based on geolocation and prompt us to collect memories?
This is so freaking cool and part of me wants to try it, but the other part is scared of the privacy implications. Even though I love (and for now, trust) Anthropic, knowing that obviously Claude Opus is the best model (as stated in the docs), I'm just not sure I want to give it access to everything about me? But at the same time, I'm so curious as to what it would come up with.
I do something similar to a personal encyclopedia but using org-roam. I don’t use an LLM yet to do any work but eventually i plan to use a local model to correlate things and pull together things that were not manually connected. Also I’m glad that LLMs can easily parse org docs so that if any future family member wanted to look through they wouldnt have to be familiar with emacs esoteric conventions.
For longevity, which is a consideration in such a project, one might prefer something based on Markdown files, like https://github.com/Linbreux/wikmd to Mediawiki, which uses a database. Then again it does support sqlite, which is open source, so it is not a big deal.
I built something very broadly similar approximately 20 years ago.
Then I forgot about it. It’s not like the data is lost, but availability is. Bringing it back up is a pain. I could probably do it in a full day of work.
What I learned: Static HTML export on every change by default is a must. I don’t think HTML will cease to be easily readable in our lifetimes.
I'm concerned about your copy that says "Private by default" and "Your wiki and archive live on your machine. Nothing is stored remotely.", yet you recommend to use Claude, which means your data is being stored remotely in most cases. I do like the idea though! Will try with local models
> Private by default
> Your wiki and archive live on your machine. Nothing is stored remotely.
Sure, the wiki is private. However, in the process your data is being uploaded straight to an AI company. Of course local LLMs exist but that’s seemingly not supported here and I think the statement on privacy could be clearer.
I have a friend whose grandma wrote a book about their family. She printed 50 or so copies of it. Not a chart-topping best seller, but each one is a cherished collector's item.
Right now, my wife and I are sticking to annual photo albums. They're already fun to flip through and we're not even that old yet.
If devices were truly private, we could simply set our mics to always-on and massive Personal Wikis would build themselves over time. Imagine that, an autobiography for everyone. Of course the downside is massive, arguably useless storage for humans. Would be pretty valuable training data though.
This is really neat! Beyond being a personal encyclopedia, remember the Spotify documentary where each episode was someone else's POV? I'd love to document a trip with friends and everyone else to do the same and see/compare what everyone experienced!
> After I found out I could also link to the real Wikipedia
It's magical watching people learn about hyperlinks. Even technical people don't always seem to know the power of a string that says, "Go to this server and fetch this document". Love it
That is actually pretty cool. I started doing that with the photo collections of family members, but only to add explanations to the metadata of the pictures. I might reconsider that approach now.
185 comments
I feel like i don't know how to emotionally react to the AI part of this story. To begin with, it is fundamentally cool we have technology like that. At the same time it felt bittersweet, like an artisan being put out of business by the factory. The first part of the story felt like much of the love was in constructing everything by hand, it seems almost sad to lose that. There is also an element of dystopia in how the AI was able to cross reference everything, bank statements, ticketmaster recipts, shazam, etc. It is kind of unsettling the power of it all.
Not sure where i'm going with this comment. Its a super cool project, thanks for sharing.
Steve Jobs saw the computer as a bicycle for the mind, a way to enable us to do more and be more. This is the metaphor against which I measure all technology.
I think that in this case, it helped someone make something deeply human by abstracting the tedium away. It did what a computer should do: aid a human with their task.
Technology has been feeling like a devil's bargain for a while now. This was a rare glimpse of how I used to see tech, and of why I was so excited about it.
If you build this encyclopedia as a purely robotic collector of facts that nobody reads, it’s probably more dystopian horror.
If you build it as a fun inner loop that reconnects you with people and memories and makes you more human, then it’s great.
We should endeavor to craft experiences that do the latter; right now we are back in the hacker days when small teams can build big new ideas, and big tech hasn’t taken over.
This is basically how social media was when you needed a computer to go online. You’d sit and sift through your feeds and there’d be message chains you respond to. You’re not really doing anything else while you’re doing that and you’re putting it out of mind once you step away. When Twitter first started getting big it was sort of a joke that people are talking to you while on the toilet. The idea that you were only ever half engaged with anything you’re doing was remarkable enough to be worth pointing out instead of taken for granted.
It’s just a lot more focused and intentional when you’re dedicating time and headspace to the task instead of “microdosing” on connection via a dopamine lottery. Even if you took away the ads and the interpolation of creator-content crowding out the connections with people you actually know, I think designing for an infinite scroll just inherently makes the thing less human-centered. It sets it up so you’re interacting with these atomized bits of ‘content’ rather than people.
There's a comment by bonoboTP in a sibling thread about the emotional complexity of a project like this. There are many ways to narrate a life story: many traumatic episodes and feuds better left forgotten, different framings, and all that emotional labor of trying to choose what and how you want to remember.
The use of LLMs for creating a shared view for some information isn't inherently morally dubious-processing and storing data is what computers have been doing for generations-except for the privacy implications, but letting this projection of a mega-corporation usurp the role of narrator for such a deeply personal story feels wrong on an instinctual level.
It is a nice idea, and I can imagine how it may serve to strengthen the family's social cohesion, in a time where everyone is busy doing the rat race. Though I'd not use it as "encyclopedia", a cold-hearted fact recorder, more like more a social-focused "Our Family Diaries" and would be much better served by family members writing down their own experiences.
I understand the bittersweet feeling because I did all the editorial work for the wedding page and the first few others and I did feel like a historian trying to connect the dots after stumbling into some primary/secondary materials and spending a couple months doing all the editorial work
after I began experimenting with agents, it sped up my process that otherwise would've taken many more months for every page given that the kinds of data sources also increased over time
I did still spend significant amounts of time like a wikipedia contributor would deciding on what to keep, enhance or delete from the page based on my own personal preferences and what I was comfortable with seeing on the page
the dystopic feeling is also fair and unsettling, I think this ironically also made me realize how important safeguarding my personal data is, we leave digital trails of ourselves everywhere so a powerful agent can string them together to create a story of who you are
That's the use-case I enjoy with AI. Let it do the heavy-lifting, I'll enjoy the rest.
What a lovely resource, especially if it reflects stories and recollections given by the subjects themselves.
The idea of having AI do it all is really off-putting IMO. For a number of reasons:
1) You lose the curation. You'll inevitably see a bias towards documenting based on the quality and availability of the sources as opposed to the significance of the event. E.g. you might not have much info about some really special childhood event you or someone else remembers, but does that mean it shouldn't be documented? Conversely, I don't want a 10,000 word essay on (to quote one of the titles from the post) "The 3D printing saga" -- just because I happen to have hundreds of WhatsApp messages on the subject.
2) I don't want to fact check every detail. Personally, I think if grandad (RIP) would have told me he one surfed a 20ft wave of the coast of Filey, Yorkshire. I don't need a correction that it was unlikely to be that high. If these things are partly being done "in memoriam" then I think it's really important to preserve the experiences, stories and recollections if the people we're trying to remember. Dates etc are fine to validate and correct. But there's an element of subjectivity to memories that is really special IMO. What even is reality at the end of the day? We're all just one big collective story we tell ourselves.
3) It feels soulless. Enough said on this one, I think people know what I mean
If instead, the OP had collected this information into a physical book, when they get bored or sick or dies, the book gets pushed into a closet or garage, waiting for some grandchild, nephew or niece to pull it out and rediscover the family history. And if anyone has even a slight interest in continuing the legacy, they don't have to know how to use a computer, just some basic scrapbooking skills, which we all learned in kindergarten.
Similar with this, when you're hand curating old photographs and personally interviewing relatives, you're learning something. You're deepening relationships and your own personal understanding of these people you love, spending time reflecting on your own life. But when you send an LLM at it and it produces the volume of real Wikipedia, now an automated process is producing more text than you can ever possibly read if all you did for the rest of your life is read.
When he died there was no way of transcribing them automatically (there still isn't really). The boxes stood in my mothers already cramped attic for 13 years, then she got cancer, and she felt a need to finish up things, so she got a scanner and started just scanning.
When my mother died she had scanned about a thousand pages, not transcribed, not anything.
The text in the diaries were fun at times, sometimes depressing, seeing how little he cared about my mother and his family was crushing.
My brother wanted to continue the scanning but I told him that I wanted to throw the diaries away. He kept half a year of writing around his birth (there's at least a sentence) and my uncle did the same, then we just watched it all burn (not literally, we threw it away at the recycling centre).
Not everything needs to be preserved. I'm happy some parts is preserved. I'm happy that those diaries are ash.
Throughout the year we keep writing in it, things we learnt, discords we had and how we resolved them, recipes I experimented with and we loved, random thoughts; basically anything and everything. And that little diary becomes an embodiment of that year.
I would also like to point out the manual labor and writing into it and not using an obsidian++-AI-auto-categorizer-3000 is simply because it feels like it's worth something, it's a nice little routine we have at the start of every year, and it's really fun reading these from 2-3 years back. Also the kids will have some really interesting reading a few years down the line.
I imagine a future where this becomes a family tradition that transcends time, knowledge from different generations, living different lives all nicely recorded in these codices. Something about this whole thing feels really beautiful to me.
Ideally square books that can go on a coffee table. At least when I am dead there will be some part of my existence in physical form, unlike all the digital things we spend decades creating.
I might put a SD card taped in the front of each one with a video too, so someone can watch it in the future.
As a separate aside, I also found old Canon photo printers (Selphy models) on ebay for about £5! Some need the little white gear inside glueing back on (there's a video on YouTube about it), and they DO NOT work with Windows anymore, but gutenprint supports them fine on Linux, so I have been printing photos (postcard size) at home. The colour isn't going to win awards and the saturation needs boosting slightly in the printer options compared to default, but it's a wonderful way to finally get some photos from trips on the walls.
We have started asking old family members to send us whatsapp audios with tales and things they remember from long-passed away family members; and what was life like in the 1930-40-50s. I want to start organizing all the info and data we have, my father has built a couple family trees, but this wiki format is indeed very promising. I'll keep an eye on this and see if we can use it.
I wouldn't give a LLM run by a US corporation access to my private photographs.
> On top of that, I exported my location timeline from Google Maps, my Uber trips, my bank transactions, and Shazam history. I would ask Claude Code to start with the photos and then gradually give it access to the different data exports.
Is anyone else feeling uncomfortable with that? It is a great project and I don't want to bash it with general concerns, but sharing all my financial and location details with any service seems like opening the floodgates to my house.
My concern is not even strictly related to AI, but about sharing all my most private data with any service. There is always a significant chance all of it is leaked sooner or later.
Though from the title I didn't expect family history, I thought it was going to be more of a project like this: https://shii.bibanon.org/shii.org/knows/Everything_Shii_Know...
I started running an private MediaWiki instance during the pandemic as I wanted something with a nice editing experience rather than editing markdown documents. I almost went with a self-hosted Confluence instance :P
Mediawiki is very very nice and it has a lot of cool features i've been loving over the years.
One of the things i like the most is the ability to embed a PDF document so that it's both downloadable and browsable from the wiki page itself (it embeds the browser pdf engine).
This means that i can, say, have a page for my microwave oven and have its user manual easily available.
Lately I've been thinking how to connect that with some LLM, most likely there's a chance to do some interesting things :)
Obviously not everyone has same needs or wants to retain stories and memories but lack of social structures and solutions seems like weird mishap.
> Her face lit up as she narrated the backstory behind the occasion, going from photo to photo, resurfacing details that had been dormant for decades.
I had started something similar with my mom over Christmas in '24. About half way through the collection she asked to stop. We would do the rest on her next visit.
Well. It never came to that as she passed away completely unexpected in March last year.
I’ll never get the chance to record the other stories. The stories from the second half of the photo collection.
I cheer for projects like this.
>
So I started pointing Claude Code at other data exports. My Facebook, Instagram, and WhatsApp archives held around 100k messages and a couple thousand voice notes exchanged with close friends over a decade.> The model traced the arc of our friendships through the messages, pulled out the life episodes we had talked each other through, and wove them into multiple pages that read like it was written by someone who knew us both. When I shared the pages with my friends, they wanted to read every single one.
This is a stunning violation of the privacy of your friends.
If someone uploaded every single private conversation I had had with them to Anthropic, they would no longer be my friend.
I do like the idea of building up this history of people, and maybe when my parents pass I'll make theirs public and so on. Great work, dude! I love it.
Secondly, the home page seems like I am reading a family history page more than talking about the software. It is confusing to me.
Thanks for sharing.
Unlike some of the comments herein, I find this as a perfect use of technology in service of users. (Yes, with some limits). I liken this to Maggie Appleton's Home-cooked Software model [1], wherein barefoot developers use technology (AI-driven or not) for writing apps for their own purposes, nominally for a user base of 1 (or very few), with possibilities of expanding to a few dozen.
In that vein, I'm a barefoot developer, and much of the software I have written in the past few months (with help from Claude, ChatGPT) is very much for that tiny user base of a few dozen (=mostly me, if I'm honest). And that is perfectly fine by moi.
I wrote a utility to organize roughly 100K+ photographs (and videos) neatly into dates/location, both for backup, as well as to maintain the memories in an organized fashion. Asked Claude to lookup location by EXIF; haven't yet asked it to "guess the location by photo" when no GPS info existed in the EXIFs. But I think I might do that.
(no, I haven't asked Claude to go thru my Uber trips or bank statements! I draw a line there!)
That is why the OP's personal wiki made me so excited - because the whole output resonates with me.
Like a few commenters mentioned their journaling experiences. I've started doing that with some of our trips (mostly post pandemic), both to remember our experiences better, and to come back to them as needed. The simple act of writing down places visited, experiences had (mostly hikes, mountains climbed, meals consumed in distant foreign places, weird/quirky experiences) causes them to be fresh in one's memories.
Thanks, this was a great project, and a great reminder as well.
[1] https://maggieappleton.com/home-cooked-software
Jann Burner, "The book of life", December 26, 2008
https://www.unexplained-mysteries.com/column.php?id=143352#s...
Dr. Hiruluk
I would probably have ended well before "I exported my Google Maps location history, Uber trips, bank transactions, and Shazam history."
Aside: I've started seeing lot of AI projects in this category say some variation of:
> it runs on your machine, your data stays with you, and any model can read it
I don’t think people fully appreciate the tension in those claims, especially when the model most area reaching for is Claude or GPT or Gemini. I think these things need more precise language about where data actually goes and what tradeoffs users are implicitly accepting.
The bank transaction + location cross referencing to figure out which restaurants you went to is pretty cool. Would be great if this could pull in social media exports too. Point it at your X, IG, FB archives, let it draft pages/content from that.
Any plan for a timeline view? Wiki format works well for depth but sometimes you just want to scroll through a year.
The MediaWiki server died and I had backups, but... literally no one in the family would've tried to resurrect it.
They knew I'd worked on genealogy for a while but I don't think anyone would've thought to rebuild a linux box covered in dust and somehow find an old MediaWiki install on it.
I should've made simple markdown files with images in an image directory and printed out copies. That's a legacy. A consolidated, easy to drag from grandpa's house and throw on a shelf and flip through, even in 2097.
Nice work!
The family has a TON of videos and photos, but no resource to guide us through what is what.
Then I forgot about it. It’s not like the data is lost, but availability is. Bringing it back up is a pain. I could probably do it in a full day of work.
What I learned: Static HTML export on every change by default is a must. I don’t think HTML will cease to be easily readable in our lifetimes.
> Private by default > Your wiki and archive live on your machine. Nothing is stored remotely.
Sure, the wiki is private. However, in the process your data is being uploaded straight to an AI company. Of course local LLMs exist but that’s seemingly not supported here and I think the statement on privacy could be clearer.
Right now, my wife and I are sticking to annual photo albums. They're already fun to flip through and we're not even that old yet.
The industrial-scale things (manufacturing, content generation, human/world history record keeping ...) eventually shrink down to human-scale things
Something to keep secure though. With AI scams taking off, this kind of context could make it very easy for someone to scam your friends/family
> After I found out I could also link to the real Wikipedia
It's magical watching people learn about hyperlinks. Even technical people don't always seem to know the power of a string that says, "Go to this server and fetch this document". Love it