How to Use AI to Become a Learning Machine

TL;DR: Today we’re releasing a new episode of our podcast AI & I. I go in depth with Simon Eskildsen, the cofounder and CEO of AI startup turbopuffer. We get into Simon’s approach to learning; how the contextual intelligence of LLMs has accelerated his process; and his take on how AI rewires the way we learn, and what that means for the next generation. Watch on X or YouTube, or listen on Spotify or Apple Podcasts.

Simon Eskildsen is a learning machine. 

I first interviewed him in 2020 about how he leveled up from being an intern at Shopify to becoming the company’s director of production engineering by reading and applying insights from hundreds of books.

A lot has changed over the last four years. Language models have made it possible to access and contextualize information faster, easier, and more cheaply than ever before—and in this episode, Simon and I talk about how this changes the way he learns. 

Simon is now the cofounder and CEO of AI startup turbopuffer, which is building a search engine that makes vector search—an approach to information retrieval that uses machine learning to gather context—easy and affordable to run at scale.

We spent an hour talking about how he leverages LLMs’ contextual intelligence to supercharge his learning, such as helping him pick up new words as a non-native English speaker, do odd jobs to maintain his rural cabin in Quebec, and articulate technical concepts in legalese. As we talk, we screenshare through his Anki setup, including the flashcard template he finds most useful, and the custom AI commands he’s created in productivity software Raycast. Simon tells me about the clutch of AI tools he experiments with for journaling, writing, and coding, as well as his thoughts on how language models will fundamentally reshape the way we learn. Here’s a link to the transcript of this episode.  

This is a must-watch for note-taking aficionados and anyone who wants to supercharge their learning with AI.

Watch on X or YouTube, or listen on Spotify or Apple Podcasts. 

If you want a quick summary, here’s a taste for paying subscribers:

Simon’s method to be a lifelong learner 

As the cofounder of a growing company and a new parent, Simon’s rituals around learning have evolved since the last time I interviewed him. Here are two ways he approaches learning today, one old and one new: 

• Spark joy to remember more. Simon is, and has been for over a decade, a voracious user of flash cards to remember all kinds of facts, from the waiter’s name at a restaurant he frequents to ways to store JSON data. He advocates creating cards that not only serve practical purposes, but also “bring you a little bit of joy” and “nostalgia,” deeming this important “if you're serious about making this a habit.”
• Launch to accelerate learning. If you want to make yourself into a “learning machine,” Simon advises that you should try getting a startup off the ground. Simon cofounded turbopuffer in 2023, and, he says, “There's nothing that challenges you more on your breadth and your skills than running a startup and building it from zero.” 

How Simon uses language models to supercharge his learning

LLMs have changed Simon’s learning process by giving him context that makes information useful. His typical prompt to an LLM is along the lines of, “Hey, I think it could be done like this, I don't know a ton about this domain, can you riff on this with me?” allowing the model to “place [the information] somewhere in latent space, find association around it, and pump that back to you.”

Most often, Simon accesses LLMs through Raycast, because he was already using the tool and it saved him from creating a new workflow with ChatGPT or Claude. Raycast allows users to select the language model they want to use and access it with keyboard shortcuts—in Simon’s case, “Command, space command space, type your question, tab done.” 

A Raycast feature that Simon often finds himself using is “AI commands,” which allows him to define a custom prompt, complete with detailed instructions and examples, for repetitive tasks. These are a few examples of the AI commands that Simon has created for his custom prompts:

• “Recipes” for crisp, well-formatted recipes which cater to his family’s dietary requirements. He uses this command to find new flavor pairings, “boost[ing] the [LLM’s] creativity” by instructing it, “I cook enough that I only really need the list of ingredients, not the instructions, give me a bunch of options that might be interesting.”
• “Define” to help him learn new words as a non-native English speaker by contextualizing the meaning with example sentences that are educational in themselves, synonyms, and an image associated with the word. “[W]hen I see a word now, I get all jittery to run this prompt because it just works so well,” he says. 

I have a list of words that I like, and we run this command to search for the meanings of two of these, “lambent” and “eigengrau.” Impressed with the results, Simon remarks that the prompt exemplifies the strength of LLMs, “where you take the average of human knowledge and cause it to go nuts on associations, but draw it in a particular direction in the latent space around things that are educational and connected.”

• “Emoji suggestion” and “Friendlier” to give Simon’s writing a warmer tone and add a few emojis. “As a northern European…the writing is sometimes a little bit too direct,” he explains. 

When it comes to using LLMs outside of Raycast, Simon says that he’s subscribed to all major tools—like ChatGPT, Claude, and Perplexity—because part of “being in AI” is “spending a hundred dollars a month on these various subscriptions, jumping around them, and getting inspired.” 

Here are a few ways Simon has integrated AI in his daily life:

• ChatGPT as your DIY assistant. Simon turned an out-of-use freezer in his rural cabin in Quebec into a fridge by following ChatGPT’s simple instructions to use an inexpensive gadget typically used for brewing beer at home. “I was about to buy a fridge for $1,000, and now I have this…$20 device from Amazon converting—I would never have thought to do that,” he says.
• Communicate better with AI. While maintaining this cabin, Simon and his family often find themselves in need of a translator, a problem LLMs are adept at addressing. “[W]riting Québécois French is an art itself…my wife uses [language models] all the time to convert something into Québécois French,” he explains.
• ChatGPT for wellness hacks. Simon followed ChatGPT’s advice to remedy the aches that come with being a knowledge worker hunched over a screen, and found it to be helpful. “For exercise stuff you need to point it in a pretty tight direction and it's not always amazing at reasoning about how it got there…but as an average of the internet…it’s pretty good,” he notes.

This is how Simon uses AI for work:

• Become an 80 percent expert with AI. Simon uses LLMs to work efficiently with his company’s lawyers by converting technical or business understanding into legal language. He cites an example: “I need to explain the exact algorithm with which we measure the uptime of turbopuffer in a way that makes sense because I don't think what the lawyer came up with made sense, let me just send it back in legalese and…minimize round trips like that.” 
• Let AI prompt your creativity. While iterating on copy, especially when it needs to be brief and punchy, Simon turns to LLMs for inspiration. “[I]t's rarely the thing that it spits out that I end up going with,” he says, explaining that the value lies in the ideas that the response prompts.
• Leverage AI’s contextual intelligence. Simon uses Notion AI, appreciating the tool’s ability to integrate context when he’s making a note or thinking a problem through. He cites an example: “‘Hey I was having his discussion with someone and I feel maybe I didn't represent myself well in this,’ and it gives you feedback…that kind of conversation has been really valuable to have.”

Beyond the uses described above, Simon also experiments with more specialized AI applications. Here’s what in his toolbox:

• AI voice-to-text tool superwhisper to journal. He’s been experimenting with talking to the tool and reflecting on the summaries it generates.
• Email application Superhuman in his inbox. He’s been finding himself increasingly using AI-generated prompts in his responses.
• Reading productivity tool Readwise Reader while reading, to look up the meanings of words and ask questions about the content. 
• AI copilot for developers Supermaven, to complete the code he is writing. 
• He wants to use Cursor—calling it the best AI code editor—but doesn’t use it as much as he’d like because it’s based in VS Code, while he’s attached to another text editor, Vim.

How AI will reshape the future of learning

Simon thinks the future of AI will  be “an evolving of us [as humans] as it is of the tools.” Here are a few applications at the intersection of learning and technology that he’s particularly excited about: 

• Broadening the horizons of curiosity. Simon envisions an exciting future for his daughter, where LLMs make any question that she might have answerable. “I feel like she's going to grow up with these tools in a way where it's going to feel incredibly natural [like] she's just talking to [a toy]…that’s really interesting to me,” he explains.
• AI as the ultimate multilingual playmate. According to Simon, AI can be used to teach children a range of languages that they may not be exposed to otherwise. Explaining that he wants to teach his daughter Danish in a non-Danish speaking community, he says, “We'll set all the UI interfaces to Danish, but we can also set ‘Wally the speaking walrus’...to only speak with her in Danish.”
• Immersive tech to accelerate learning. The two use cases of VR and AR that Simon finds intriguing are: when the technology will reliably replace a monitor setup, and how it could generate visual imagery that would make it easier to remember things. He thinks these applications are “starting to become, in the adjacent, possible.”

You can check out the episode on X, Spotify, Apple Podcasts, or YouTube. Links and timestamps are below:

• Watch on X
• Watch on YouTube
• Listen on Spotify (make sure to follow to help us rank!)
• Listen on Apple Podcasts

Timestamps:

1. Introduction: 00:01:06
2. How entrepreneurship and parenthood changed Simon’s learning rituals: 00:02:51
3. How Simon accelerates his learning by using LLMs to find associations: 00:12:59
4. Simon’s Anki setup and the flashcard template he swears by: 00:18:24
5. The custom AI commands that Simon uses most often: 00:26:02
6. How Simon uses LLMs for DIY home projects: 00:37:45
7. Leveraging LLMs as intuitive translators: 00:40:48
8. Simon’s take on how AI is reshaping the future of learning: 00:51:38
9. How to use Notion AI to write: 00:59:10
10. The AI tools that Simon uses to write, read, and code: 01:08:53

What do you use AI for? Have you found any interesting or surprising use cases? We want to hear from you—and we might even interview you. Reply here to talk to me!

Miss an episode? Catch up on my recent conversations with star podcaster Dwarkesh Patel, LinkedIn cofounder Reid Hoffman, a16z Podcast host Steph Smith, economist Tyler Cowen, writer and entrepreneur David Perell, founder and newsletter operator Ben Tossell, and others, and learn how they use AI to think, create, and relate.

If you’re enjoying my work, here are a few things I recommend:

• Subscribe to Every
• Follow me on X
• Subscribe to Every’s YouTube channel

Dan Shipper is the cofounder and CEO of Every, where he writes the Chain of Thought column and hosts the podcast AI & I. You can follow him on X at @danshipper and on LinkedIn, and Every on X at @every and on LinkedIn.