Shantanu | Product x AI

9.4K posts

Shantanu | Product x AI banner
Shantanu | Product x AI

Shantanu | Product x AI

@Shan0o

Building things that matters.

San Francisco, CA Katılım Mayıs 2009
318 Takip Edilen387 Takipçiler
Shantanu | Product x AI
@trq212 This is great, I saw it in your CWC workshop. I happened to ask, how do you share these while collaborating with larger team? Drop’em on slack?
English
0
0
0
876
Thariq
Thariq@trq212·
HTML is the new markdown. I've stopped writing markdown files for almost everything and switched to using Claude Code to generate HTML for me. This is why.
Thariq@trq212

x.com/i/article/2052…

English
898
1K
12.1K
4.4M
Shantanu | Product x AI
Vibe coding discourse feels novel? We had this exact argument when compilers took over from hand-written assembly. Compilers had bugs. Zero-days. Optimization gaps. We didn't abandon them. We built better ones, added tooling, and evolved our trust model. The question isn't "should we trust AI-generated code?" It's "how do we build the compiler layer on top of it and abstract ourselves from the code itself as we did with assembly?"
English
0
0
0
26
Shantanu | Product x AI
The smarter the model, the worse it knows its own limits. ICLR 2026 showed it — more reasoning training increases tool hallucination in lockstep with task performance. Capability is easy to benchmark. Restraint isn’t. What does “knows when to stop” even look like in agent evals?
English
1
0
0
20
martin_casado
martin_casado@martin_casado·
LLMs are more like humans than software: unpredictable, error prone, malicious ... it's best to use the same interfaces / controls we use for humans: headless? no, a full browser state in a vectordb? no, a filesystem some sandbox proxy thing? no, a VM / mac mini etc ...
English
29
7
130
15.7K
Shantanu | Product x AI
This is amazing, this week I was teaching my dad (anesthesiologist) ai and finding a great starter, I wouldn't have gotten anything better than this. Thanks @yourgirlhils
Lenny Rachitsky@lennysan

"I keep meaning to really dig into AI, but I can't seem to find the time. I'm just so busy." If that sounds like you, @yourgirlhils and I made something just for you. Introducing the Couch-to-5K for AI—a free, fun, 30-day program that finally turns "I should get better at AI" into a daily habit. Fewer than 10 minutes a day. Builds your AI practice progressively. This will accelerate your AI learning more than anything else you've tried. Learn more: lennysnewsletter.com/p/your-couch-t… Get started now: couchto5k.ai

English
0
1
25
26.4K
Shantanu | Product x AI
The stack is: Model → Chip → Electricity Everyone invested in the top layer. Same pattern as cloud. AWS was not the best cloud app in 2010.
English
0
0
0
17
Shantanu | Product x AI
Chips need electricity. One LLM query uses 10x the electricity of a Google search. We scaled that to a billion users.
English
1
0
0
22
Shantanu | Product x AI
The AI race isn’t being won in a lab. It’s being won on the power grid. Let me explain the stack very few are talking about. 🧵
English
1
0
0
65