Making the Non-Actionable Actionable
Some Thoughts on Learning With AI
I hope you had an outstanding Thanksgiving! I was fortunate to spend it in New York City (including a viewing of the Macy's Thanksgiving Parade). I highly recommend seeing it if you haven't already.
When I'm thinking about AI-powered products, I often return to this idea of hyper-focused tutoring. Maybe it's my somewhat academic nature and love of books. Or the fact that I simply love learning.
Whatever it is, I see a huge opportunity to combine aggressive, extremely granular self-education with the power of AI.
As I've said previously, the true disruption comes from the ability of an LLM to understand your experience, your learning priorities, the underlying content, and adjacent material that can supplement your knowledge.
I'm not saying that LLMs get everything right. After all, we have all dealt with hallucinating LLMs. Sometimes, they spit out information that is so clearly off that any knowledgeable human in your field would laugh at it.
At the same time, the output is often good enough for a vast majority of tasks. It's especially true for the initial stages of learning new skills or tasks. As we develop more expertise, I think we need to go beyond the LLM and learn by ourselves (or with some sort of mentor). But in the beginning, a customized approach that an LLM spearheads is a very attractive option for the vast majority of people.
To reiterate, I think LLMs' real opportunity here is making the non-actionable actionable. In other words, it's about taking some collection of media (a book, podcast, YouTube clip, or something else) and making a hyper-focused, actionable education plan.
The words actionable and education are essential here. Instead of passively watching a YouTube video or glazing over a dense textbook, the potential is working with LLMs to fully understand what you want to learn. It's about using the information as a catalyst for real-life action.
Right now, the experience isn't good enough. You have to have one LLM window open, provide it with sufficient context, add additional media sources if the LLM hasn't indexed them, and hope for the best. While it's somewhat sufficient now, I think that there will be some killer app out there that makes the process much more seamless. Perhaps I should build it!
In any event, this is a space that I'll be closely watching in the next few years. If you know of any startup that is working on solving this problem, don't hesitate to let me know!
Prompt of the Week
With the year almost over, I wanted to ask ChatGPT about how I can best optimize my December and get ready for the new year. We all develop resolutions at the end of December, but they are often rushed and later forgotten (I can speak from experience).
I drafted this prompt to make next year different. Try it out and let me know what you think!
"It's December 1, 2025. Pretend that you are a world-class career coach who helps your clients aggressively pursue their goals. Considering what you know about me, my obligations, and career goals, I want you to develop a well-thought-out action plan on how to (1) maximize the rest of this year and (2) capitalize on the new year. No detail is too small and out-of-the-box ideas are welcome."