Noah Brier | February 27, 2024
The WITI Recommends Edition
On products, projects, and fifth birthdays.
Noah here. We are coming up on our fifth birthday (!!!), and for about four of those years, we’ve been thinking about building out a recommendations site off the back of our nearly 1500 editions. Well, we are finally making it happen. I’m happy to announce Why is this interesting? Recommends, which pulls together over 1,000 products, books, software, and other kinds of recommendations from across our many emails from the past five years.
Why is this interesting?
WITI is a long-term experiment. We are never quite sure where things are headed, and we continue to do it because we really enjoy it. This year, we brought on an amazing editor in Louis Cheslaw and have been thinking about some other new ideas to extend the concept and community. This is one of them. We have a bunch of thoughts on where we’d like to take this, but in the spirit of launching early, we wanted to get it out into the world. As you might expect, there are affiliate links where possible, and we will make a buck or two if you buy something, so click and purchase away. (Making WITI more financially sustainable is another goal for the year.)
The core of it is pretty simple: we sucked the product recommendations out of all the emails, scraped the product pages for descriptions and images, and categorized the whole thing. Most of the manual work was made possible thanks to AI, which, while taking lots of heat these days in the world of media and beyond, is amazing at solving stupidly annoying problems like how to pull the recommendations out of some 1,000,000+ words.
If you’re interested in the nitty-gritty, I wrote a bit about it in a recent BrXnd newsletter (that’s where I write about marketing and AI stuff). The gist is that AI makes a fantastic fuzzy interface: a simple way to transform unstructured data (like many essays in HTML) into structured product recommendations. This process isn’t overly complex (minus the scaffolding to set up the pipelines). Mainly, it’s asking the AI to extract product recommendation links from a given post. I posted the full prompt over on BrXnd Dispatch, as well as a bunch of other execution details if you’re interested.
Back to the site, I particularly like the pages that pull together many tips, like this one from our most popular post, The Go-Bag Edition. This points to a direction I generally would like to head with things, which is to pull together multiple recommendations into a list of sorts. I didn’t quite get it done for this release, but the idea is that if you’re looking for a great pen or a novel, you could find it easily. (For now, search does work pretty well.)
It’s also fun to see all the various things I’ve linked to over the years.
There’s much more work to do, and we’re open to feedback. So click around, buy some stuff, and let us know what you’d like to see.
—
Thanks for reading,
Noah (NRB) & Colin (CJN)
—
Why is this interesting? is a daily email from Noah Brier & Colin Nagy (and friends!) about interesting things. If you’ve enjoyed this edition, please consider forwarding it to a friend. If you’re reading it for the first time, consider subscribing.