The Next Chapter of Machine Learnings

tl;dr: I’m sunsetting the Machine Learnings newsletter and starting The DeBrief, a newsletter that turns AI hype into actionable playbooks for startup operators and their teams. Sign up here.

In 2017, I started the Machine Learnings newsletter as a side quest.

A few months earlier, I co-founded a startup called Journal. We hadn’t built a product yet, but my co-founder was a machine learning engineer and we knew the technology would play a central role in it. We also knew we’d need to build an audience for our eventual product.

I had grown newsletters for other companies, so I figured I’d take a crack at starting one about machine learning. The thing was, I knew next to nothing about machine learning. So before sending the first edition of the newsletter, I did a bit of googling, starting with “What is machine learning?”

If you can believe it, it was hard to find anything of substance to read about artificial intelligence back then. Most coverage of the topic was limited to two extreme ends of a spectrum. On one end, we had technical research papers by Big Tech companies meant to attract engineers. On the other end, there were sensational “robots are going to take your job and kill your family” articles.

After a few hours of googling, I could barely define machine learning. I figured other people must be struggling too. I was convinced that if the newsletter crossed the low bar of making it easier for people to find good articles about AI, it would be well received.

And hopefully, the people who enjoyed the newsletter would be curious to try the not-yet-built product if I mentioned it from time to time.

Then lightning struck.

I created a newsletter in MailChimp (dating myself), added a few AI articles I read that week, and sent the first ~10 editions to 30 of my friends.

Then one afternoon, I hit publish on a Medium post called “The Non-Technical Guide to Machine Learning.” (Note: I would link to it here, but I’m banned from posting on Medium. It’s a story for another time.) The post was a collection of interesting companies, news sources, and people to know in the AI space. At the bottom, I asked people to sign up for my newsletter.

A few minutes after publishing, my phone started to vibrate. I got a MailChimp notification every time a new subscriber signed up for the newsletter. The single weekly notification I was used to became 5 per minute. Then 10. Then 50. Then I turned my notifications off.

Medium featured "The Non-Technical Guide to Machine Learning" on its homepage. The Machine Learnings subscriber list grew by hundreds every time I refreshed the page. Eventually, more than 1 million people saw the post, and 50,000 signed up for the newsletter.

Stuck between a rock and a hard place.

In 2017/18, there were only a few 50,000-subscriber newsletters about niche topics like artificial intelligence. With engineers, product managers, investors, and founders among the readers, we’d built the audience we hoped to almost overnight. And then came the hard part.

It took us years to build a product that people loved and that group of users was quite different from the folks who read Machine Learnings. As time passed and the goals of the startup and the newsletter began to diverge, I lost the motivation to write it every week.

It was hard to justify spending time on the newsletter given my competing priorities as a founder, but it felt foolish to stop sending the newsletter given the quality of the audience we’d built. What started as a religious dedication to sending a newsletter edition every Sunday night turned into major guilt over missing one every now and then. It escalated to taking off for long stretches during the holidays, then having other people (thank you Ryan Gilbert and Brendan Langen) write it for me, and eventually not sending it at all.

I’m excited to start writing about AI again.

The world and my world have changed a lot since the last time I sent out an edition of Machine Learnings.

Since the last edition I sent, I’ve gotten married, sold my startup, and become a father. AI has also officially gone from niche to mainstream. 

We no longer need to convince people that AI matters—now, I want to focus on helping people use AI, not just read about it.

I’ve also missed having an excuse to stay on the cutting edge of AI, the act of writing, and the experience of helping a group of people tap into insights that’re hard to come by.

So what’s next?
I’m starting a new newsletter.

Machine Learnings was about understanding the importance of AI. The DeBrief is about applying it.

Every time I talk to founders I hear some version of “AI is moving so fast. I can’t keep up with everything that’s happening, let alone put it to use within my startup.”

I’m excited to move past curating articles and to helping people and teams use AI to make themselves 100x more effective. 

Unlike generic AI roundups, The DeBrief will provide first-hand insights from my experience co-founding and scaling an AI-native company from zero to acquisition. It will share lessons from covering the space for nearly 10 years. It will also offer actionable posts based on the lessons I learned from implementing AI workflows within companies, helping you leverage AI both personally and professionally.

My writing will be equal parts frameworks for thinking about AI and tactics for implementing AI workflows.

We’ll explore topics like:

• How high-performance individuals & teams actually use AI

• When and what not to automate at work

• Increasing context windows for LLMs

• Reinforcing hiring and cultural practices with AI

• AI’s emotional dynamics in the context of knowledge work (ie. why we feel guilty when we use ChatGPT)

While OpenAI deserves all of the credit for popularizing AI, I can’t help but feel some pride knowing that Machine Learnings played a small role in helping people wrap their heads around AI – and hopefully feel less caught off guard by all the major developments that have happened recently.

Thank you for reading and for sharing your thoughts and feedback along the way. I hope you’ll join me for this next chapter.

If Machine Learnings helped you understand AI, I promise The DeBrief will help you use it.

Click here to sign up for the next edition:

Thank you to Jackson and Ryan Gilbert for reading drafts of this.