“Taste is the moat now.” I have a secret team of agents working on something that I can’t talk about from a small startup that will reveal itself soon. They have a very different standard of taste than others I have worked with. You will see that very clearly soon. Vik is
AI
-
LiDAR debate in autonomous vehicles: cost and design concerns
By
–
You wouldn’t believe how many arguments I have had over the years about LiDAR.
— Robert Scoble (@Scobleizer) 17 avril 2026
Including this year at NVIDIA’s GTC with a founder of an autonomous vehicle company.
It adds needless cost to the vehicle and makes them look ugly.
Thank you for laying it out. Again. @elonmusk… pic.twitter.com/xsIbaO4Y2wYou wouldn’t believe how many arguments I have had over the years about LiDAR. Including this year at NVIDIA’s GTC with a founder of an autonomous vehicle company. It adds needless cost to the vehicle and makes them look ugly. Thank you for laying it out. Again. @elonmusk
-
Startup Career Move Tripled My Salary in Years
By
–
I actually loved working there. But it couldn’t keep me away from the startup world. Within a few years my salary had tripled. So made the right choice
-
DeepSeek Mega MoE: Fusing MoE Operations Into Single Kernel
By
–
New from DeepSeek: Mega MoE! Instead of running MoE as a chain of separate steps (dispatch → MLP → combine), Mega MoE fuses everything into a single mega-kernel. Even more importantly, it overlaps NVLink communication with Tensor Core computation, reducing the classic
-
General VLM vs Specialized SAM Performance Comparison
By
–
how much “worse” is general sota vlm vs sota dedicated SAM at this stuff?
-
GPT-Rosalind Revolutionizes Drug Discovery with AI Protein Reasoning
By
–
🧬AI is coming for one of science's biggest bottlenecks.
— Futurepedia – Learn to Leverage AI (@futurepedia_io) 17 avril 2026
Drug discovery has always been a 10-15 year slog. GPT-Rosalind is built specifically to change that; protein reasoning, genomics, biochem, and scientific tool use all in one frontier model.
This is what AI for real-world… https://t.co/wUI3nRLVY1AI is coming for one of science's biggest bottlenecks. Drug discovery has always been a 10-15 year slog. GPT-Rosalind is built specifically to change that; protein reasoning, genomics, biochem, and scientific tool use all in one frontier model. This is what AI for real-world
-
AI Needs Solid Botanical Data More Than Ever
By
–
#AI needs solid botanical #Data more than ever
by Chris Bivins @Nature Learn more: https://
bit.ly/4en1gdi #MachineLearning #ArtificialIntelligence #ML -
Codex Unveils Computer Use, Automations, and Image Generation
By
–
woke up to entire TL talking about Codex and all the things that weren’t possible yesterday but are today! highlights for me:
> computer use
> thread automations
> artefacts (slides, sheets, PDFs)
> image generation in Codex
> in-built browser and.. this is just the start 😉 -
Training Custom AI Models: Market Implications Discussion
By
–
it was a random idea, i won’t actually do it. but it triggers some interesting convos that help me think through the market 🙂 cool idea on training own model. i’d get permission first if i wanted to do this, but not sure i want to enough to start doing that.
-
Cancer Mirror Biology Movement: Insights from Treatment
By
–
Procedurally generated Bruegel with little workers and everything!