How I ended up running my own server because I have too much shit to remember.
I used to love Evernote.
Seriously. There was something almost magical about the idea that I could save an article, photograph, document, or random thought, throw it into a digital filing cabinet, and actually find it again years later. Before Evernote, my organizational system was essentially a combination of browser bookmarks, folders named “Misc,” and the increasingly unreliable assumption that I’d remember where I put something.
Evernote changed that.
I used it for years. I even paid for the premium version for a while, although I eventually settled into the free tier, which was generous enough for my needs. I rarely bumped up against the monthly upload limit, and it became one of those services I didn’t think much about because it simply worked.
It was a digital extension of my memory. And as someone with a lifelong tendency to collect information about absolutely everything, I found that enormously appealing.
Then things got weird.
Evernote changed hands, its pricing and free-tier restrictions became increasingly aggressive, and the product that had once seemed like an indispensable little utility started feeling like a hostage negotiation.
I understand that software costs money to develop and maintain. I’m a developer. I’m not expecting people to work for free, surviving on good intentions and the occasional compliment.
But there was something particularly irritating about watching a service build a loyal following around an accessible product, then dramatically change the terms of the relationship.
Even worse was the reaction from certain corners of the Evernote community. Anyone unhappy with the changes was quickly informed that they were freeloaders who had no business complaining.
Well, excuse the hell out of me.
I wasn’t demanding a lifetime of free software. I was questioning whether I wanted to keep trusting my personal archive to a company that could fundamentally change the arrangement whenever it pleased.
Suddenly, Evernote wasn’t nearly as cool as it used to be.
The Notion Years
Fortunately, Evernote let me export my data.
That turned out to be important.
I eventually found Notion, which could import my Evernote material and offered something even more interesting: not just notes, but interconnected pages, databases, tables, and the ability to build something resembling a personal wiki.
I was hooked.
Over time, Notion became the place where I kept track of projects, research, technical notes, ideas, and all sorts of other information that seemed worth preserving.
And believe me, I can find an astonishing number of things worth preserving.
A useful article about a programming language I might learn someday? Save it.
Research into a historical event? Save it.
Instructions for fixing some obscure software problem that I may never encounter again? Absolutely save it.
An interesting fact about something completely unrelated to anything I’m currently doing?
You know the drill.
The problem is that collecting information and organizing information are two very different hobbies.
I had become remarkably good at the first.
The second? Well.
My Notion workspace was gradually evolving into a digital equivalent of one of those storage units where somebody opens the door and gets buried beneath seventeen years of Christmas decorations and exercise equipment.
Except everything was searchable, which made it feel organized.
And to Notion’s credit, I was happy with it. I still think it’s an excellent product.
But a little voice in the back of my head kept asking an uncomfortable question.
Hadn’t I been here before?
Notion’s free offering was almost suspiciously generous. I wasn’t convinced that the economics of providing that service indefinitely would always work in my favor.
Maybe nothing would change. Maybe Notion would continue being wonderful for another twenty years.
But I had already experienced what happens when a company changes the rules around a service that has become part of your daily life.
I didn’t particularly want to do that again.
Fine. I’ll Build My Own Cloud.
For years, I’d entertained the idea of running my own server.
Not because I had some grand ideological objection to cloud computing. I use cloud services all the time. They’re incredibly useful.
I just liked the idea of having a machine that belonged to me, running software I chose, storing data I controlled.
And gradually, the list of things I wanted to do with such a machine became long enough that I could finally justify buying one.
Thus began the adventures of AndrewServer3000.
Yes, that’s its name.
No, I’m not taking suggestions.
One of its first assignments was to become the permanent home of my personal knowledge base.
The requirements were straightforward: I wanted something like Notion, hosted on my own hardware, accessible from my desktop, laptop, phone, or whatever other device I happened to be using.
One authoritative collection of information. Multiple ways to access it.
Preferably without paying somebody a monthly subscription for the privilege of remembering what I did in 2017.
How hard could that be?
Experienced self-hosters are laughing right now.
The Obsidian Detour
My first serious candidate was Obsidian.
And I can understand why people love it.
Obsidian stores notes as ordinary Markdown files, which is fantastic from a portability standpoint. It’s extensible, powerful, and has a passionate community. It seemed like a strong contender for replacing Notion.
So I kicked the tires.
And then I discovered a fundamental mismatch between how Obsidian normally works and what I wanted to build.
Obsidian is fundamentally a local-first application. Your vault lives on your device, and synchronization keeps copies of those files available across your devices.
That’s a perfectly sensible approach for a lot of people.
It just wasn’t mine.
I didn’t want my phone maintaining a local copy of my entire digital brain. I wanted my server to hold the information and my phone to access it when I needed it.
Think of the difference between having a library you can visit and being required to carry a copy of every book in the library everywhere you go.
I already carry enough unnecessary shit.
And considering that my long-term ambitions included not just notes but documents, photographs, research archives, and other media, the distinction mattered.
There were workarounds, of course. There are always workarounds.
But I didn’t want to spend the next several years engineering around a fundamental architectural decision that didn’t suit my needs.
Obsidian was a good application.
It just wasn’t the application I was looking for.
Enter AFFiNE
Eventually, I discovered AFFiNE.
AFFiNE is an open-source workspace application that combines documents, knowledge organization, databases, and other tools in a package that feels much closer to Notion.
More importantly, it offered a self-hosted deployment.
Now we were talking.
I could run the server myself, store the data on my own hardware, and access the workspace from other devices.
The software wasn’t merely a place to write notes. It could become the foundation for the personal knowledge system I’d been imagining.
So I installed it.
And, naturally, everything worked perfectly on the first attempt.
Just kidding.
There were Docker containers, configuration files, networking issues, synchronization questions, and enough troubleshooting to remind me why normal people buy appliances instead of building their own infrastructure.
But eventually I had a working AFFiNE instance.
And then came the fun part.
Moving the Junk Drawer
Getting AFFiNE running was one thing.
Moving my existing Notion archive into it was another.
My Notion workspace had accumulated years of material, and exporting it revealed just how much information I’d been collecting. There were thousands of pages and files, nested structures, databases, attachments, and plenty of organizational decisions that probably made sense to me at the time.
Probably.
I wanted to preserve as much of that history as possible, rather than simply abandon it and start fresh.
The migration took experimentation. Some things imported beautifully. Others required investigation, adjustments, and a better understanding of how the two applications represented information.
At one point, I found myself examining how AFFiNE stored documents internally.
Which is a perfectly normal thing to do when you’re trying to move some notes from one application to another.
Eventually, though, I got the archive moved over.
Not every organizational problem magically disappeared. There is still cleanup to do, and I’m still learning how best to use AFFiNE.
But the important part is that the material has a new home.
My home.
Well, technically, a computer in my home.
You get the idea.
The Librarian Who Doesn’t Exist Yet
Here’s where the project gets more interesting.
My ultimate goal isn’t simply to recreate Notion on my own server.
I want to build something that helps me manage the enormous amount of information I’ve accumulated.
Because there’s a dirty little secret about digital note-taking applications.
They make saving information incredibly easy.
They don’t necessarily make maintaining that information any easier.
Over time, you collect duplicates. Projects change names. Ideas get abandoned and resurrected. Articles become outdated. Notes that should be connected remain isolated because you filed them six months apart and forgot the relationship.
Eventually, your carefully constructed second brain starts to resemble your actual brain.
Which, in my case, is not necessarily a compliment.
So the long-term plan is to develop an AI Librarian.
I want an assistant that can examine the archive, recognize related information, suggest better organization, identify duplicate material, and help consolidate scattered notes into something more useful.
Imagine being able to ask:
“Didn’t I research this problem two years ago?”
Or:
“Show me every project idea I’ve had involving maps.”
Or:
“These seven notes seem to be about the same thing. Can we combine them?”
And instead of getting a list of vaguely relevant search results, the system could help maintain the knowledge base itself.
Of course, I don’t want an AI enthusiastically deleting fifteen years of personal history because it decided something looked redundant.
The Librarian will need rules, safeguards, and a healthy respect for the fact that sometimes two seemingly identical documents are different for a reason.
In other words, it needs to be a librarian, not a raccoon with administrative privileges.
This part of the project is still a work in progress.
But it’s the part I’m most excited about.
A Second Brain, Not a Second Junk Drawer
I don’t regret using Evernote. It introduced me to a way of preserving information that fundamentally changed how I work.
I don’t regret moving to Notion, either. It gave me a much more flexible way to organize that information, and I used it happily for years.
And I don’t think everyone should run out and buy a server just to escape a software subscription.
Self-hosting comes with responsibilities. You become the IT department. Backups, updates, security, maintenance, and troubleshooting are suddenly your problem.
If the server breaks, there’s no customer support representative to yell at.
Just me.
And possibly a very disappointed future version of me wondering why past me thought this was such a brilliant idea.
But there’s something deeply satisfying about having a system that I control, running on hardware that I own, preserving information that matters to me.
AFFiNE isn’t the finish line. It’s the foundation.
I’m still learning its strengths and limitations. There are things I’d like to improve, features I’d like to build around it, and an entire archive that needs some serious housekeeping.
But for the first time, I feel like I’m building a personal knowledge system that can evolve with me rather than depending entirely on the business decisions of whichever company happens to be hosting my notes.
Maybe someday I’ll have an intelligent, beautifully organized archive capable of connecting decades of projects, photographs, writing, research, and memories.
Or maybe I’ll just have the world’s most sophisticated digital junk drawer.
Either way, at least it’ll be my junk drawer.