The most socially accepted form of self-sabotage:
Overthinking.
It’s why you can’t make decisions, finish anything, or sleep without scrolling.
The 6 hard truths from Naval Ravikant that shattered my obsession with "Mental effort":
Can AI have fun while coding? Can it find quick, clean, and correct fixes to the most subtle issues?
The depths of crazy fixes that exist in code are so awesome! Today, I added a "space" somewhere in the output streamed by a service to fix an edge case in UI 🫠 😎
Of course, soft skills matter, but which ones? In my opinion, every engineer should be better at these two: active listening and clear communication.
Listening isn’t just about hearing; it's about being present processing information, and asking the right follow-up questions. Too often, we listen passively, just waiting for their turn to speak and sometimes even hoping that we need not speak at all :) Instead of listening passively, listen with intent.
Clear communication is all about conveying ideas, decisions, and information easily clearly, and unambiguously. Be concise, structured, and complete; and say just enough to get the point across without unnecessary fluff. Apply this for your meetings, slack messages, or even documentation.
One thing I proactively do is to observe and measure the number of clarifications asked in my communication. If there are too many such questions, then my message probably wasn't as clear as I thought it was. I go back and iterate on it and ask myself
- did I really listen?
- did I really understand what I was talking about?
- did I communicate it well?
This happened to me recently, so, coming from my personal experience :) I am a work in progress, and will always be :)
What if you could edit images using just your words? Meet Vibe Photoshop—where AI-powered prompts transform your creative vision into reality, requiring no complex tools.
The Gemini system leverages
- JAX/XLA framework for computation
- 6th Gen TPUs
- A unified embedding space for different modalities
- Cross-modal attention mechanisms enabling integration across data types
@prakashd88 I was curious How does the system perform on top tweets ? What's the basis of the model ? Can we get a comparative analysis?
Like let's take past data. 20 viral tweets and 20 non-viral tweets, what's the result of the model ? What is the prediction accuracy?
Claude Sonnet 3.7 + @cursor_ai is wild: 1 prompt = a full landing page!
Meet Agently: Agentic platform with the agency.
(inspired by Andrej Karpathy's post)
Link and video in comments.
I dont know what the future is like!
As more and more of our lives are decisions driven by algorithms, are we not slowly becoming enslaved and addicted to these very algorithms 🤯
I guess we need to go beyond the copy paste paradigm of using AI chatbots or AI Code Editors. We officially have term for it now, "vibe coding" (by Karpathy) 💻
As modern agriculture, and food distribution removed the need for us to run and hunt wild animals, AI will fully remove the need for us to think on a lot of matters... 🧠