Overwhelmed, I had to declare ‘information bankruptcy’. How I made Substack usable again in <1 hour.
I love the idea of Substack, but it stops being fun when you’re subscribed to 169 publications, as I am at the time of this writing.
The slow drip becomes a fire hose
At first, when I was only subscribed to a few accounts, the occasional email notifications were manageable. But as that number climbed, my inbox quickly became overwhelmed by this fertile source of perpetual interruption and distraction.
Gmail filters didn’t work
In need of quick relief, I set up a Gmail filter to label the Substack notification emails and shift them out of my inbox to avoid the constant pings. I figured I could consciously choose to check in on these notifications on my own time, instead of being interrupted throughout the day.
But what ended up happening is I never got around to checking the notification emails, and so I eventually stopped reading any of my Substack subscriptions.
The website isn’t any better
The other approach was always to log in to the website and check in on my subscriptions there. Unfortunately, this planned ritual never caught on for me either.
And to make things worse, lately they’ve turned the website into an X.com clone with microblogging and a rich media feed — and the last thing I want is to be sucked into another shallow, hyperengineered algorithmic feed.
The irony of it all
What attracted me to Substack in the first place was its focus on long-form, meaningful content. Yet the web interface was peddling exactly what I came to get away from on the other platforms: infinite feeds, cheap social micro-interactions, and short-form content with annoying hooks.
Eventually I just stopped reading Substack altogether. 169 subscriptions, and I wasn’t reading a single one. Information bankruptcy.
The solution
Last night, in the span of an hour, I finally put this issue to rest with Claude Code.
Step 1: Grab my subscriptions
I pointed Claude Code with the --chrome flag to the Library page to grab the URLs of all the publications I’m subscribed to and save them into a feeds.json file.
By suffixing /feed to any Substack page URL, you get a clean RSS feed for that publication’s content. Using this, I can generate a daily report of all posts published across all of my feeds in the past 24 hours.
Step 3: Let an LLM do the skimming for you
With LLMs, we can do one better — we can also ask a model to generate a short outline of each article. This is the real unlock. Instead of scanning 30 posts to figure out which ones are worth my time, I get a condensed summary of each one and can decide in seconds whether to commit to a full read.
Step 4: Automate it
I had Claude Code generate a Python script for the daily digest, and set it up with a GitHub Action that runs every morning at 7am. The script emails the report to me via a new Gmail account that I’ll continue to use for notification emails and reports like these.
It’s best to create a dedicated “service account” for notification emails rather than using your main Gmail account. You want to limit your security surface area. See also: the principle of least privilege.
One day in, it’s already paying off
I’m only a day in and I’m really enjoying how effortless this new workflow is. In this morning’s email, I already found a great article from a Substack I subscribed to a long time ago and had completely forgotten about.
This is another great value of agentic engineering — to help you quickly build solutions to navigate information overload, separate the signal from the noise, and tune into knowledge that was otherwise difficult to access.
Open SourceThe code for my Substack daily digest is available at github.com/tikiverse/substack-digest