Migrating from Bear to Obsidian

In all my years of writing for publication I’ve never really talked about the tools I was using. A person’s choices and setup are so individual to their needs, I would think. Who wants to read about someone else’s setup?

Then a few weeks ago, after reading a 1,000-word post about vibe-coding a home network dashboard, I realized there’s a weird satisfaction to reading about other people’s configurations.

So I thought I might talk through the evolution of my personal notetaking setup, culminating in a recent migration from Bear to Obsidian.

TL;DR

I’ve known about Obsidian for years and always suspected it would be my eventual tool of choice, but I was also intimidated by what it would take to set it up well. So I started with smaller, less customizable programs and only upgraded once I had a use case that required functionality my current software didn’t have.

By the time I got to installing Obsidian, I knew exactly how I wanted it set up, and I was able to create a system that worked for me almost immediately without much consternation. I could never have gotten to the more complex tool without first using the less complex tools and bumping up against their limitations.

Another lesson this process taught me: When migrating to a new system, don’t just blindly replicate the old system’s features. Figure out what jobs those features were doing for you, and find the best ways to use the capabilities of the new system to do those jobs.

Also: A great deal of my smooth transition is due to the assistance of AI, which makes boring tasks like data migration and cleanup infinitely easier.

Now, to the details.

Starting Simple: Apple Notes

I’ve never been too fussy about my notetaking app. Most of my notes were taken on a work computer, so access to notetaking software was always variable based on the company’s security and support preferences. For a while I used Evernote, then OneNote when it became available for Mac. More often than not I would just make a text file and store it in a project folder, trusting search to help me find anything I truly needed.

When I was laid off I realized I didn’t have a decent repository for notes on my home machine. Not wanting to spend too much time on tool selection, I just started using Mac’s default Notes app. I had a few folders for major areas (a coaching program, networking materials, notes from a nonprofit board I served on). Other than that, I made sure every note had a decent title, and that was it for organization.

Moving Up to Bear

Apple Notes worked OK, with two challenges: First, my notes were completely contained within the app, which, on my elderly MacBook, offered only PDF export. Plus, even if I brought all my software up to date, Notes still wouldn’t have Markdown live-editing. These both became a problem as I started experimenting more and more with AI tools and shifting my default writing mode to Markdown.

I went out in search of a Markdown editor to make things easier on myself. I still resisted Obsidian as being too configurable — I didn’t want to spend much time setting up the system. A writer I admire mentioned Bear as her editor of choice so I gave it a try. With its clean, subtly pretty interface, Bear was a lot more pleasant to use than Notes, and it had the benefit of live-editing and import/export for Markdown files.

Plus, Bear had a web clipper, so I could clip web pages to throw into my repository. This became crucial when I started building out automations for my job search: I could clip job postings that I came across into Markdown files, assign tags as I clipped, and then run automations based on the tags. The job posting .md files were also helpful for using AI to help me parse what skills to emphasize in my application materials.

Bear was not without hiccups, though. Foremost for me: it’s Apple-only. I could use it on my Mac but there was no capturing thoughts on my Android phone. Also, Bear didn’t have folders but instead was reliant on tags alone to organize notes. While I was able to create a tag structure that would mimic the folders I’d had in Apple Notes, it required constant maintenance. For example, any ’#’ in a clipped article (like the #null that showed up in every clip from LinkedIn) would be interpreted as a tag, so I was constantly monitoring for terms I didn’t want showing up in my tag hierarchy.

And finally, like Apple Notes, Bear was a self-contained universe. Though everything in it was a Markdown file, I had to export the file to manipulate it in other programs, then re-import it to resume editing in Bear, which would create a duplicate copy of the note. This started to get in my way as I began to explore more agentic versions of my automations, allowing my AI tools to rewrite, rename, and rearrange files according to instructions.

From Bear to Obsidian (Thanks ChatGPT)

It had been less than a year since I’d started using Bear, but it was clearly time to upgrade. I suspected Obsidian, which stores notes as Markdown files within your file system, was what I needed. But after using Bear I knew I also wanted something more attractive than Obsidian’s default UI. Luckily I found a blog post that walked through ways to make Obsidian look (more like) Bear, so I gave the migration the green light.

This is where AI really came into play: I wanted to stress-test Obsidian as my long-term solution without having to bash my head against the wall, so I called on ChatGPT as a programming partner.

When I ran a test import of my Bear notes into a throwaway Obsidian vault, it was still messy. Attachments with UUID-like filenames were mixed in with my notes in a way that made it difficult to scan folder contents. Also, Obsidian had interpreted HEX values and some URLs in the text of notes as tags, making the taxonomy wonky. So I asked ChatGPT to write some cleanup scripts, which I tested on this throwaway corpus until I was confident they could clean up the files without causing damage.

Another example: Without folders, Bear forced tags to do double duty, standing in for both “what is this” and “where does this belong.” Obsidian has folders, tags, and properties, so I could finally split those jobs apart: folders for where something lives, tags for the concepts I wanted to tie together. (And some properties to track status of actionable notes, but that’s for another post.) I worked with ChatGPT to help suss out my mental model around folders and tags, keeping the focus on my own jobs to be done instead of simply migrating over the tag taxonomy from Bear.

In fact, setting up folders and tags surfaced a behavior I didn’t even realize my system needed to support. If you’d asked me at the beginning how I wanted to move through my notes, I’d have said search. But as I started clicking through my Obsidian setup I realized I really wanted to browse — click once on a folder, then scan a recency-sorted list to find the note I wanted. I used ChatGPT as a natural-language search for a plug-in to meet my requirements and found Notebook Navigator — which I wouldn’t have even known to look for until I examined my own mental model.

Finally, ChatGPT helped me set up and troubleshoot the open-source Syncthing to sync my notes across Mac and Android. So I now have a locally-stored, accessible through the filesystem, interoperable repository of notes and web clippings, synced across my devices at no extra cost.

In Conclusion

Like many strategy-oriented product people, I tend to see smaller steps and interim states as falling short of the vision. This process has reminded me that sometimes those interim steps are necessary to get to the desired end state. Had I started by trying to create the perfect Obsidian setup, I’d probably have overcomplicated things and ended up with a system I didn’t use. Instead, hitting the edges of Apple Notes and Bear functionality showed me what I really needed so I could configure Obsidian around actual, and not abstract, needs. Then ChatGPT made the unglamorous parts (migration scripts, plug-in research, corpus cleanup) fast enough that the whole thing was worth doing at all.

Why did I go through any of this? Partly portability — I just didn’t like the idea of my notes being locked up in a piece of software, no matter how easy it was to import/export. But ultimately it was about making the corpus accessible to both me and the tools I use. What insights might be gleaned by looking at my thinking across saved articles, meeting notes, and writing drafts? I look forward to finding out.