
Building AI x learning products. Harvard’17. Side project: longcut.ai YouTube: youtube.com/@ZaraZhangg Substack: zarazhang.substack.com
Page 1 • Showing 15 tweets
Many people underestimate how fast it is to build with AI and how difficult it is to get other people to care about what you built
In my experience, first-time founders focus on product, second-time founders focus on distribution AI has made building easier, but made distribution harder Build it and they won’t come Attention must be earned
AI can make anything look good on the surface. Beautiful words, nice formatting, professional polish. But it can't fix a hollow core. People think because AI wrapped their idea nicely, their content is good. They're confusing packaging with substance.
New workflow: 1. Open a voice memo app 2. Brain dump your thoughts on a topic; record it 3. Transcribe it 4. Import transcript into NotebookLM 5. Get it to generate a slide deck using Nano Banana Pro 6. See your rambling thoughts visualized into structured & beautiful slides and feel smart
If you're non-technical, AI coding tools are for pitching/communicating to engineers, not replacing them. I use Claude to make interactive prototypes that show exactly what I want to build. Then I hand it to a real developer. The prototype isn't the product. Conflating the two is dangerous.
We should stop categorizing people as “technical” and “non-technical”
A friend in China recently asked me how to improve her English. She’d struggled with it for years as a student, labeled herself “bad at languages,” and now wanted to try again for her career. But she couldn’t sustain her efforts. I asked how she’d been studying. She said she bought grammar books, pronunciation guides, vocabulary lists, and tried to memorize them. Why? “Because I need to start from the basics and learn systematically.” There it was. Even years after leaving school, she still had the student mindset. The education system had brainwashed her into thinking learning must be systematic, bottom-up, textbook-driven. I told her: This is no way to learn as an adult. Instead, find content in English you’re already interested in. Decipher it with dictionaries and AI tools. Learn vocabulary from that content. Define the “job to be done” for English—a tool for communication and understanding—and work backwards. What are you trying to understand? Read that. Make that your textbook. As adults, we need to do a 180-degree reversal of how we learned as students. In school, you start with foundations and work upward. You won’t apply what you learn until years later. But the basics are also the most boring. If you try to learn this way as an adult, you’ll quit. Adults should do the opposite: Start with a real-world problem. Figure out what knowledge you need. Then go backwards to learn it. AI accelerates this dramatically. It’s a 24/7 world-class tutor that answers any question, as long as you know what to ask. I experienced this myself when trying to become more technical. A few years ago, I wanted to learn coding, so I watched intro CS lectures. I got bored fast. I couldn’t connect the content to anything I needed in real life. This year, I started using AI coding tools like Google AI Studio, Replit, and Cursor. I described my product ideas in plain language, and they turned into working products in minutes. Every time an idea became real, I felt elated, empowered. Before, screens of code gave me headaches. They reminded me of my inadequacy. Now, I saw code as just a tool to achieve my goals. I asked AI to walk me through the codebase structure, the languages and technologies, what each file did, how everything connected. I had it add detailed comments everywhere. After a few projects, I’d learned more about coding than a year of college CS lectures. It wasn’t “systematic.” But I’m not trying to become an engineer. I’m trying to become technically literate so I can use code as a tool. Why this reversal? • Students’ full-time job is learning. Adults have other jobs and responsibilities. After a long day at work, who has the energy for boring lectures? • Students must learn. The system forces accountability: exams, homework, class attendance. Adults learn purely by choice. Nobody forces you. This means sustainable learning requires intrinsic motivation. If it’s boring, you quit. • Students have defined goals. Do well on exams. Adults must define their own goals. Otherwise, you lose motivation fast. The common thread: Sustaining motivation is the key to learning anything as an adult. Because you’re not going to learn overnight. So if you’re trying to learn something, start with a problem or project. Figure out the job to be done. Work backwards. Learning becomes fun. Learning becomes sustainable. Forget “learn first, do later.” Do first, learn later. You don’t get good and then produce output. You produce output and then get good.
I turned Twitter into my inspiration database Vibe coding ideas → folder Good design → folder New products to try → folder Prompts that work → folder Then I batch experiment with everything Way more useful than just scrolling
Starting to believe that the best place to use Nano Banana Pro might actually be in NotebookLM
The future of job search is "show your work" not "tell me what you did"
Compiled my favorite resources for learning AI into a free website:
Zara's AI learning library | Faces
My daily AI tools now: - Claude for writing & therapist (better personality) - Gemini for multi media & long context window tasks & vibe coding Haven't opened ChatGPT in a couple weeks; can't think of a compelling use case
This year I transitioned from marketing to product. The best way to learn product is to build and ship products. Not listening to podcasts about building products. Not watching tutorials. Not attending bootcamps or business school classes. What I love about AI coding tools is that they let me learn this way. In the past, to ship a real product, you needed to hire a team, raise money, find engineers. Now it takes an idea, time with AI coding tools, and putting it in front of people who will actually use it. In a few months, I've made dozens of prototypes, shipped a side project in actual production, and learned what kinds of ideas resonate with real users. Every project is a data point that trains your intuition: "if I ship this, people will behave that way." That's product sense. You can't get it from a book or a class. AI has been the best tutor for me not because it teaches me information, but because it lowered the barrier to hands-on execution. The best way to learn anything is to actually do it.
Upload a meeting transcript to NotebookLM and get it to turn it into a slide deck using Nano Banana Pro. Absolutely insane.
I spent 8 years away from X because it felt like noise. When I came back 6 months ago for AI, my feed was full of clickbait and topics I no longer cared about. So I did something most people don't do: I intentionally shaped my algorithm. Every time X pushed clickbait at me, I clicked "not interested." I unfollowed accounts from 8 years ago. I followed builders, PMs, and people actually creating things. I did this religiously for a month. Now my Twitter feed is basically an AI learning platform. Product launches, technical discussions, startup updates. The video feed is just product demos. Here's what I learned: most people complain about social media algorithms controlling them. But you can control the algorithm if you're willing to put in the work. The platforms aren't inherently bad. They're just optimized for engagement, not learning. But if you're intentional about what you engage with, you can bend them to serve your goals. X can be TikTok for learning. You just have to train it.