I Was Saving Highlights to Readwise for Two Years. I Don't Think I Read a Single One Twice.
I was saving highlights to Readwise for two years. I don’t think I read a single one twice.
I also have Readwise Reader. They send me a weekly digest email. I have never opened one.
That’s not a Readwise problem. That’s a “I built a pipeline that ends at a destination I don’t visit” problem. My highlights went in. Nothing came out. I had perfect capture with zero absorption.
The pipeline I had
Three main things I read: web articles people share, LinkedIn posts, Kindle books.
Web articles went to Readwise Reader. Nice interface. Highlights synced. Sent to Craft Docs. I opened Craft Docs maybe twice.
LinkedIn posts - this is the underrated pain point. Every LinkedIn “save” feature just saves a URL. Not the content. So when the post gets deleted or the algorithm buries it, it’s gone. I had hundreds of saved posts that were, effectively, nothing.
Kindle books worked fine, actually. The highlight sync is genuinely good. But good capture into a place I don’t revisit is just… a tidy graveyard.
The chain was long. The reflection never happened.
This isn’t really about Readwise
Readwise has a resurface digest. An AI mode to interact with your highlights. A reader. Integrations with almost everything. It’s a well-built product.
The problem isn’t the features. It’s the architecture.
It’s a destination. Another app. Another tab. Another mental context I have to remember to switch into - on top of the 20 other things competing for that context. We’re already drowning in bookmarks we never return to, newsletters we meant to read, tabs we’ll “get to later.”
Bookmarking feels like progress. Returning is the actual work. And returning requires the tool to be embedded in where you already spend time, not somewhere you have to navigate to.
I was also, if I’m honest, overachieving with this setup. More tools, more integrations, more sync paths - I thought I was building a better system. I was just moving the overhead around. The less software I have to use, the less mental overhead I carry. The overhead doesn’t disappear when you replace Readwise - it moves. But if it’s out of sight, it stops interrupting.
What I actually want from a second brain
Not capture. That part is easy. Capture is solved.
What I want: to actually think about what I’m reading. To draw connections between a book highlight and something I read last week. To sit with an idea long enough that it changes how I work.
A second brain that doesn’t make you think is just a well-organised archive.
The missing ingredient wasn’t another plugin. It was a process: something that forces the reflection step, inside the place I’m already working, without requiring me to remember to do it.
The replacement: three plugins and a Claude skill
The plugins handle capture.
Obsidian Web Clipper - clips articles from the browser straight into my vault with title, source URL, author, and tags. One click. Lands in 4 References/Clippings/ with a consistent structure.
Social Archiver - saves LinkedIn posts with the actual content, not just the URL. This one mattered - it’s the only tool I found that actually clips the text so the post survives deletion or algorithm burial.
Kindle Highlights - syncs directly from my Kindle library into Obsidian. No Readwise in the middle.
Each note lands with the same structure: frontmatter, ## Summary, ## Reflection, ## Content. Same format regardless of source.
The part that actually changed how I read
I built /digest - a Claude Code skill that runs on any captured note.
Point it at a Kindle book, a web clipping, or a LinkedIn post. It reads the content, writes a 2-3 paragraph summary, then asks 3-4 reflective questions - tailored to what you just read, not generic prompts. Each question has 4 suggested choices so you’re not staring at a blank box on a low-energy day. You pick what resonates, write what doesn’t fit any option, and it records your response with a date.
Then it searches your vault for related notes and surfaces them as wikilinks. Connections you wouldn’t have made manually, because you weren’t going to manually search.
Everything writes back to the source note automatically. You never leave Obsidian.
That’s the piece Readwise can’t replicate yet - not because they haven’t built the features, but because the thinking happens somewhere else. The reflection has to happen inside where the thinking lives.
The honest bit
Setup took me a couple of hours across the plugins. The Web Clipper template configuration is fiddly.
And yes, the overhead moved rather than disappeared. But out of sight, out of mind - if a plugin works quietly in the background and my workflow gets faster, I’m not thinking about it. I am thinking about Readwise, because I had to choose to open it.
The multiple-choice reflection questions also felt mechanical in theory. In practice they’re not - because they’re generated from the specific content, not a template. Having options to react to is easier than a blank page. That’s by design.
If your highlights aren’t making you think, they’re just clutter with good metadata
The goal was never to capture more. It was to actually absorb what I was reading.
Building a second brain means building your own structure, the one that makes sense for how you think - and then using it to focus on what actually matters: writing, reflecting, drawing connections. Not searching, not linking, not maintaining pipelines.
With Claude inside Obsidian, most of that infrastructure runs itself. I can say what I’m thinking in plain words and get back relevant content from my own knowledge base. The capture is automatic. The reflection is prompted. The connections surface.
That’s what I wanted two years ago. It just took this long to build it.
← Back to blog