About Perch

What if newsletters and blogs worked more like podcasts?

You don’t get an email every time there’s a new podcast episode. That would be incredibly annoying. Instead, you open Spotify and all of your episodes are right there. If you want to find, say, a new health podcast, that’s easy too. Just check the Health & Fitness charts. Contrast this with the current experience of trying to find the best health & fitness newsletter or blog — a Google search returns mostly SEO-optimized nonsense. Reading is also expensive (e.g. $10/mo per Substack or New York Times subscription), and great article recommendations are difficult to come by if you’re not an avid user of sites like Twitter, Reddit, or Hacker News.

Top Health & Fitness Podcasts on Spotify

Top Health & Fitness Podcasts on Spotify

Google search results for the query “Best health and fitness newsletters”

Google search results for the query “Best health and fitness newsletters”

These are problems that hundreds of millions of people face every day with the explosion of writing platforms like Substack, Medium, Beehiiv, and Wordpress. And while lots of venture capital dollars have been raised by these companies to build better writing platforms, nobody is focused on the reader. That’s why we’re building Perch.

The mission of Perch is to make great writing accessible. We believe that removing friction around cost, discovery, and usability in today’s reading experience is how we get to billions of people reading every day. Consider how much more music people listen to today because Spotify made it so easy versus buying CDs or $0.99 songs on iTunes. Perch will create a similar explosion in demand by making it easier to read great writing.

Removing friction and latency from discovery unlocks new demand to an extent that people continually underestimate. As Steven Levy writes in his history of Google, In The Plex, Larry Page and Google were obsessed with speed because they believed “speed is chronically underestimated as a factor in successful products.” Google found that removing just 100 milliseconds of latency from search queries led to a meaningful increase in the number of searches. We’d expect the same to be true for reading. But instead of saving people 100 milliseconds, we’re reducing the search cost for the perfect article from minutes to a fraction of a second. Not to mention all the time people waste reading bad articles.

We can tell you more about our plan during your interview, but the gist of it is solving lots of hard data engineering and machine learning problems while introducing an ad-supported freemium model that shares revenue with writers, similar to YouTube.

In the meantime, please try out the app for yourself here! It’s still early (our team is just Mike & Matt right now), but the product will improve rapidly over the next two months as we build out our founding team in New York City.

A few dozen people are already using the app almost every day even though we’ve done very little to promote it, and some users are even beginning to share it with their friends. Here are a few testimonials from real users we didn’t know personally before they downloaded Perch:

Screenshot 2024-10-22 at 5.58.06 PM.png

Screenshot 2024-10-28 at 8.06.57 PM.png

Screenshot 2024-10-22 at 6.01.29 PM.png

Screenshot 2024-12-03 at 8.25.31 AM.png

In terms of initial funding, we’ve raised $6 million from some of the best venture capitalists in the world, including Reddit cofounder Alexis Ohanian’s 776 Ventures, Naval Ravikant, Ashton Kutcher’s Sound Ventures, and Balaji Srinivasan.

For more on why we think building Perch is important, please read our mission statement.

The Role

The promise of Perch is to analyze all of the writing on the Internet and show you the best thing you can possibly read every time you open the app. We believe that if you can do this reliably and guarantee users that they will never waste 10 minutes on a bad article again, billions of people will open that app every single day.

Of course, there’s a tremendous amount of work that lies between the current state of Perch and this future we envision. Here are just a few of the engineering problems we need to solve:

  1. Index every blog and newsletter on the Internet. This involves building a web crawler, perfectly parsing all of the terrible HTML out there, and monitoring millions of websites for new posts with near-zero latency.
  2. Clean and organize our reading data. We are collecting tremendously useful data to gauge article quality and relevance that will help us better serve our users. For example, we know for any given article how many people opened it, finished it, bookmarked it, shared it, etc.
  3. Gather other useful data. The most significant AI breakthroughs typically come from identifying or creating new sources of data, not inventing new algorithms. Google’s PageRank is a good example — the key insight wasn’t algorithmic work, it was identifying backlinks as an incredibly valuable data source. Backlinks could be quite useful for Perch (e.g. if your favorite writer is Paul Graham and he links to a blog post, you’ll probably be interested in that post too). What else? Article shares on X, upvotes on Reddit, comments on Substack are just a few interesting data sources that come to mind.