Here's a timeline, with examples, of how marketing evolved over the years and how some of those changes reflect what I've seen in tech.

Google just announced Stitch, a new tool for building UIs with both code and visual editing. Features: composable design, fully customisable themes, live components, export to Figma and editable HTML/CSS code. Seems like this tool is part of the broader vibe coding trend.

"The AI race isn't just about building the smartest models, it's about building the next platform that will power the next generation of AI apps." . H&M, virtual assistants and the future of AI interaction @hm @sama @OpenAI.

In case you missed YC’s first AI Startup School, here's a short recap:.

RT @ashtom: It already can! 😎 Assign an issue to Copilot to experience the magic + check out our docs to see what the Copilot coding agent….

"People think AI is all about the model. But it's really about how it fits into someone's life" - Peter Deng @pxd. What really matters:. 1. Proprietary data that makes the AI smarter over time. 2. Seamless workflow integration so it feels like part of the way people already work.

It's probably just a matter of time before GitHub Copilot starts fixing tech debt: spotting bad patterns, suggesting cleaner code, refactoring functions, generating missing tests, updating docs, even recommending better architecture. Can’t wait for those PRs @ashtom.

“High Agency” in 5 minutes. Plus a quote from a former colleague who's always mentoring and sharing his knowledge.

There's a new generation of entrepreneurs using all these AI tools and plotting the next revolution from their college dorms. I've seen some amazing apps that are about to come out, and I'm excited about the future.

I honestly think this is a great time to be an engineer, scientist, researcher, and above all, an entrepreneur. AI is giving rise to super startups, companies that will innovate at lightning speed and take on the big players. If someone had said 3 years ago that a startup would.

New AI technique makes LLMs write code more like real programmers (but we're not there yet) . Tags: parallel coder agents, neural nets, reinforcement learning, memory graphs, internal feedback loop.

Main takeaways from Huang's Nvidia keynote in Paris:. - Quantum computing is at an inflection point. CUDA-Q launched to bridge quantum and classical systems. - Humanoid robots expected to become widespread by 2027. - Introduced token-based Al and "agentic" models that can think.

Blockchain armageddon: if Bitcoin relies on cryptographic algorithms, could quantum computing break them? Wonder how long Satoshi Nakamoto has to develop post-quantum cryptography and upgrade the network!.

The Al race is not just about building the smartest models. It's about controlling influence. If your agents dominate ads, reviews, social and search, you control perception, demand and the market. This is why I think @sama needs @io. OpenAI has the brains, now it wants the body.

Tracking real human engagement will get harder. If AI agents start browsing, clicking, and even buying, then ad metrics could become meaningless and ad budgets could end up being completely wasted. #AIagents #AdTech.

I was reading the book "AI Valley" and came across the story of Inflection AI. Building foundation models isn't cheap. Training a model alone can cost hundreds of millions. Add infra, talent, data, and scaling, and $1 billion seems not enough.

The top 10 use cases in 2025 indicate a shift from technical to emotional applications, including growth in areas such as therapy, personal productivity, and personal development. More and more people are using LLMs to find purpose and improve themselves.