Nishith Prabhakar

tech, edtech, products, running, reading, history, space

[Andrej Karpathy @karpathy]

This works really well btw, at the end of your query ask your LLM to "structure your response as HTML", then view the generated file in your browser. I've also had some success asking the LLM to present its output as slideshows, etc.

More generally, imo audio is the human-preferred input to AIs but vision (images/animations/video) is the preferred output from them. Around a ~third of our brains are a massively parallel processor dedicated to vision, it is the 10-lane superhighway of information into brain. As AI improves, I think we'll see a progression that takes advantage:

1) raw text (hard/effortful to read)
2) markdown (bold, italic, headings, tables, a bit easier on the eyes) <-- current default
3) HTML (still procedural with underlying code, but a lot more flexibility on the graphics, layout, even interactivity) <-- early but forming new good default
...4,5,6,...
n) interactive neural videos/simulations

Imo the extrapolation (though the technology doesn't exist just yet) ends in some kind of interactive videos generated directly by a diffusion neural net. Many open questions as to how exact/procedural "Software 1.0" artifacts (e.g. interactive simulations) may be woven together with neural artifacts (diffusion grids), but generally something in the direction of the recently viral https://t.co/z21CP5iQfu

There are also improvements necessary and pending at the input. Audio nor text nor video alone are not enough, e.g. I feel a need to point/gesture to things on the screen, similar to all the things you would do with a person physically next to you and your computer screen.

TLDR The input/output mind meld between humans and AIs is ongoing and there is a lot of work to do and significant progress to be made, way before jumping all the way into neuralink-esque BCIs and all that. For what's worth exploring at the current stage, hot tip try ask for HTML.

[sunil pai @threepointone]

idea guys are in shambles: for years they treated engineers like the tax you pay to manifest “the vision.” hand them a magic build machine and suddenly you find out the vision was mostly vibes and powerpoint.

programmer guys are in shambles: for years they watched suits take victory laps on their commits. now they’ve got infinite idea machines and still mostly build toys for themselves; because “what users want” is the one prompt they can’t autocomplete.

[Nishith Prabhakar @nprabhakar]

@championswimmer Is that AI effect or just a covid/zirp correction?

[@levelsio @levelsio]

If I look at the last 15 years of knowing people in startups, and then seeing who became successful and who didn't, I'm starting to see some general patterns

The people I know who became successful (and very rich) regularly asked for help and feedback, and then applied that feedback very quickly (think minutes) and shipped fast while maintaining their own vision

The people who didn't become successful are the ones who worked on stuff for months/years without asking for help or feedback, or when they did took weeks/months to apply it

So I think the feedback -> implement loop and speed of it is possibly very important

[Andrej Karpathy @karpathy]

I've never felt this much behind as a programmer. The profession is being dramatically refactored as the bits contributed by the programmer are increasingly sparse and between. I have a sense that I could be 10X more powerful if I just properly string together what has become available over the last ~year and a failure to claim the boost feels decidedly like skill issue. There's a new programmable layer of abstraction to master (in addition to the usual layers below) involving agents, subagents, their prompts, contexts, memory, modes, permissions, tools, plugins, skills, hooks, MCP, LSP, slash commands, workflows, IDE integrations, and a need to build an all-encompassing mental model for strengths and pitfalls of fundamentally stochastic, fallible, unintelligible and changing entities suddenly intermingled with what used to be good old fashioned engineering. Clearly some powerful alien tool was handed around except it comes with no manual and everyone has to figure out how to hold it and operate it, while the resulting magnitude 9 earthquake is rocking the profession. Roll up your sleeves to not fall behind.

[Nishith Prabhakar @nprabhakar]

@collision Best podcast covering business and tech! Super insightful!

[Paul Graham @paulg]

I met a founder today who said he writes 10,000 lines of code a day now thanks to AI. This is probably the limit case. He's a hotshot programmer, he knows AI tools very well, and he's talking about a 12 hour day. But he's not naive. This is not 10,000 lines of bug-filled crap.

[Unitree @UnitreeRobotics]

Unitree Introducing | Unitree R1 Intelligent Companion Price from $5900
Join us to develop/customize, ultra-lightweight at approximately 25kg, integrated with a Large Multimodal Model for voice and images, let's accelerate the advent of the agent era!🥰

[Arnav Gupta @championswimmer]

Was trying to build the front end of https://t.co/B1myuLycFY with 0 coding

Tried these

- @lovable
- @boltdotnew
- @anyappai
- @getcreatr
- @cursor_ai
- @windsurf_ai

Lovable produces the best UI design. Cursor and Bolt get closest to 100% functionality

The worse thing is, almost all of them got the app to 80-90% of what I need - with the exception of some small bugs, for example

- not reading users location
- not making group join correctly
- confusing between the group id and join secret
- not categorising the places properly

And apart from that almost all of them fail at the second order prompt where I ask to fix one of the bugs

This is a very frustrating situation usually because while it gives a ton of excitement to see soooooo much of the basic code base generated (ton of which is boilerplate), it is just not as much fun to fix bugs in someone else’s code as it is to write greenfield code even if a lot of it is repetitive boilerplate.

And at that point you wonder whether it is better to just start from scratch again (writing code yourself, of course with Cursor/Copilot) or trigger just one more blank-slate vibe coding attempt. Your mind says fixing it in the 80% project might be faster, your heart doesn’t want to touch AI slop and fix it.

Interesting times.

[Lenny Rachitsky @lennysan]

New episode with Michael Truell (@mntruell), CEO and co-founder of @Cursor_ai

Cursor is at the very forefront of changing how software gets built, and is one of the fastest-growing products of all time. They hit $100m ARR just 20 months after launching, and recently crossed $300m ARR just two years after launch.

In a rare interview, Michael shares:
🔸 His vision for “what comes after code” and how programming will evolve in the next few years
🔸 Their early pivot from automating CAD
🔸 Why Cursor built their own custom AI models despite not starting there
🔸 Why “taste” and logic design will become more valuable engineering skills than technical coding ability
🔸 Why the market for AI coding tools is much larger than people realize—and why there will likely be one dominant winner
🔸 Michael’s advice for engineers and product teams preparing for the AI future
🔸 Key lessons from Cursor’s unprecedented growth
🔸 Much more

Don't miss this one.

Listen now 👇
• YouTube: https://t.co/MDqgAj0MyH
• Spotify: https://t.co/i9AjHGCuqa
• Apple: https://t.co/d4MS2fcbDA

Thank you to our wonderful sponsors for supporting the podcast:
🏆 @Get_Eppo — Run reliable, impactful experiments: https://t.co/fMTIjddBaM
🏆 @TrustVanta—Automate compliance. Simplify security: https://t.co/JHcQhNsK42
🏆 @oneschema_co — Import CSV data 10x faster: https://t.co/awfcZaoXqj

[Hiten Shah @hnshah]

The AI that wins won’t feel like AI.

[Dr. Parik Patel, BA, CFA, ACCA Esq. @ParikPatelCFA]

When Trump said he was going to get rid of capital gains tax I thought he was going to get rid of the tax part not capital gains

[Corey Lynch @coreylynch]

My team has been working for a year trying to solve intelligence for the home

Meet Helix 🧬: the first Humanoid Vision-Language-Action model

This is one Helix neural network running on 2 Figure robots at once

They've never seen any of these items before

[Alex Lieberman @businessbarista]

Always reminds me of a convo I had with one of the founders of FanDuel.

He’s running 4 companies today and I asked him “how can you possibly be as successful or effective across many companies?”

He said “I spent 80% of my time at FanDuel on the wrong things because I didn’t know what to work on. Now I just spent 100% of my time on the right 20% across all 4.”

[Nishith Prabhakar @nprabhakar]

Thread.sleep(60000);

[OpenAI @OpenAI]

Today we are launching our next agent capable of doing work for you independently—deep research.

Give it a prompt and ChatGPT will find, analyze &amp; synthesize hundreds of online sources to create a comprehensive report in tens of minutes vs what would take a human many hours.

[Trung Phan @TrungTPhan]

meanwhile, European power sockets

[Stripe @stripe]

We built a machine to track Black Friday/Cyber Monday on Stripe.

Explore real-time data all weekend: https://t.co/67Jw9zaOpQ.

[Nishith Prabhakar @nprabhakar]

@onghu @santattech I still find “unless” so confusing

[Joy Bhattacharjya @joybhattacharj]

A veteran cricketer and coach who came so close thrice in the last 25 years finally got his hands on a trophy. Two iconic cricketers whose supporters often clash proved that they cared for the India flag more than anything else. A player who was booed by his home crowd for displacing an icon finally had redemption.
A fielder nervelessly pulled off a boundary catch that justified all the years of practice
The greatest bowler of our time did it when it mattered most. And his one year old son was there to watch him!
The end of an amazing story, perhaps the start of another.