I accidentally discovered how to compress a semester of learning into 48 hours.
A grad student at MIT showed me his NotebookLM setup. I thought he was just organized. Then I watched him pass a qualifying exam on a subject he'd never studied before.
Here's exactly what he did:
First: he didn't upload a textbook.
He uploaded 6 textbooks, 15 research papers, and every lecture transcript he could find on the subject.
Then he asked NotebookLM one question:
"What are the 5 core mental models that every expert in this field shares?"
Not "summarize this." Not "explain this topic."
Mental models. The stuff that takes professors years to develop.
But the next part is what broke my brain.
He followed up with:
"Now show me the 3 places where experts in this field fundamentally disagree, and what each side's strongest argument is."
In 20 minutes he had a map of the entire intellectual landscape of the field:
the debates, the consensus, the open questions.
Most students spend a full semester just figuring out what those debates even are.
Then he did something I've never seen before.
He asked:
"Generate 10 questions that would expose whether someone deeply understands this subject versus someone who just memorized facts."
He spent the next 6 hours answering those questions using the source material. Every wrong answer triggered a follow-up:
"Explain why this is wrong and what I'm missing."
By hour 48, he could hold a conversation with his thesis advisor without getting destroyed.
The tool didn't change. The questions did.
Most people treat NotebookLM like a fancy highlighter.
These students are using it like a private tutor who has read everything ever written on the subject.
The difference between a semester and 48 hours isn't the amount of content.
It's knowing which questions to ask.
It's possible to make web sign-in more frictionless. Help simplify web authentication for everyone by joining the WebAuthn immediate mediation origin trial → https://t.co/RcuQoNVgjO
#PasskeysWeek
If AI can code 100x faster, why aren't you shipping 100x faster?
Because AI code is not production-ready code, and definitely not code where you understand and can vouch for every line
Introducing the Command Center alpha. Support our Product Hunt launch!
wrangler v4 just landed!
The big changes are under the hood, with a big update to our esbuild version, improved type generation, and we now default to local mode across all sub-commands.
Excited to introduce Icon, The First AI Admaker.
We’re backed by Peter Thiel’s Founders Fund & execs of frontier AI labs like OpenAI, Pika, & Cognition.
Icon (https://t.co/RKWomrKrAu) is like ChatGPT + CapCut, but for making winning ads with AI in minutes.
How it works:
1. Icon looks at your video library & tags scenes (e.g. "close-up," "unboxing"). These scenes become reusable clips used as lego blocks for making ads.
2. Prompt Icon’s AdGPT to generate scripts focused on specific angles & audiences.
3. Icon finds perfectly matching clips for every script scene & generates an ad that is 80-99% complete.
4. Make edits with our CapCut-like video editor until you're happy.
3-person creative teams make 30 ads per month. With Icon, they make 300.
We co-built Icon with $100M+ revenue brands like Ridge, Jones Road, Immi, Backbone, & MUD\WTR to solve big pain-points:
1. Making lots of ads is extremely painful. Icon helps you automate the tedious parts of scriptwriting, scene matching, video editing, audience research, UGC creation, & more.
2. AI-generated ads look like trash. Icon remixes your existing footage into new ads, matching your production quality & maintaining brand aesthetics.
3. Existing solutions charge $2K-$30K/month for:
🔍 Competitor ad spying & cloning
📈 Creative analytics
📹 Custom & stock avatars for AI UGC
👥 Audience research
📁 Video storage & tagging
Icon does scriptwriting & video editing on top of everything above for just $999/year (their margin is our opportunity 😉).
If you’ve made it this far, we have a surprise for you 👇🏻
Check this out:
- Familiar API. Working with files is as simple as working with URLSearchParams, FormData, and other web APIs: get/set/has
- Swap out this `storage` object for anything! In dev write to disk, in production write to S3 or D1 or Blink-182, app code doesn't care!
Just realized Cloudflare .com domains are now cheaper than Namecheap! 😮
Using Cloudflare directly is much more convenient since you need to redirect the name servers there anyway
What do you all use for domains?
We were super stoked to see @AnthropicAI launch the Model Context Protocol (MCP) this week. 🫡
Today, we’re rolling out the first version of our Cloudflare MCP server with initial support for R2, D1, KV, and Workers.
I just joined Category Newsletter Creator by @dickiebush and @Nicolascole77.
If you're looking to start (or grow) your newsletter in 2025, check it out.
It has everything you need to build a 6-figure Category Newsletter:
https://t.co/ho1CtXjvqf
NEW: React Router v7
React Router is the best bridge for React 18 👉 19. Use it as a full framework or library with your own architecture
The same router used by millions of React apps, now ready for the next decade 💪
Read our blog post for more
https://t.co/baMiKPgwcm
React Router v7 prerelease is now available
This pre-release is the first glimpse at the new Vite plugin coming to v7. This plugin enables React Router to be a full-stack framework, giving you features such as:
- code splitting
- data loading
- actions
- server rendering
- static pre-rendering
- pending states
- optimistic UI
- and much more
I've long said that programs which are easier for computers to reason about are also easier for humans to reason about
Now @kellogg_martin has tested that
Enter "On the Relationship between Code Verifiability and Understandability"
@kentcdodds@kettanaito Just purchased the Testing Fundamentals pro workshop from @kettanaito . It’s really great–not fluff. I like that all contents in EpicWeb are straight to the point. Time is so valuable nowadays.