A running list of the sources I use to keep up with AI, plus the ones I share with people who want to get up to speed. It is opinionated on purpose. If something made the list, I think it earns your time.
If you are starting from zero, skip ahead to Sources To Get Started and come back to the top once the vocabulary clicks.
Sources I Follow
The core of what I track. Sub-sections are by audience so you can pick the one that fits. Delivery format is noted up front, since some of the best signal moves across YouTube, podcasts, newsletters, and X.
For everyone
Knowledge workers using AI in daily work. If you skip everything else, review these.
- Newsletter. Ethan Mollick at oneusefulthing.org. The best single signal for how AI is changing knowledge work. Practical, grounded, no hype.
- YouTube. Claude. A top pick. Short, practical, fastest way to see what is possible.
- YouTube. How I AI, with Claire Vo. Real people showing real workflows.
- YouTube. Grace Leung. Sharp, well-produced walkthroughs.
- YouTube. Tina Huang. Ex-Meta data scientist. Practical tools and workflows for analytical work.
- YouTube. Anthropic. Deeper sessions and research-flavored content.
- YouTube. Google Cloud Tech. Good for the Google and Gemini side of the house.
- Podcast. Everyday AI. Daily cadence. More technical than the name suggests; treat as a buffet.
- Newsletter. Anthropic News. Product news and research, straight from the source.
- YouTube. Matthew Berman. Fast on new releases. Delivery grates; coverage is worth it.
- YouTube. Greg Isenberg. Hit or miss. Cherry-pick the building-with-AI episodes.
For leaders and strategists
Strategy, adoption, market direction.
- X. Andrew Ng. Clear-eyed on the technical and business reality of AI. An antidote to both doomers and boosters.
- X. Dario Amodei. Anthropic's CEO. Long-view thinking on where the frontier is heading.
- X. Marc Andreessen. a16z co-founder. The strategic bull case, distilled. Grain of salt on boostier posts; market direction usually holds up.
- X. Greg Brockman. OpenAI president. Sparse but useful signal on where OpenAI is going.
- Podcast. AI + a16z. Top pick for strategic, longer-horizon thinking.
- Podcast. The a16z Show. Same strengths, broader tech aperture.
- Podcast. Me, Myself, and AI (BCG × MIT). Enterprise case studies told by the people who ran them. Strong on real-org adoption.
- Podcast. AI in Business (Emerj). Applied, vertical by vertical.
- Podcast. No Priors. Sarah Guo and Elad Gil interview frontier researchers and founders. Strong companion to a16z. Also on Apple Podcasts — pick one feed.
- Blog. a16z. Strategy and market point of view.
- Research. BCG, Deloitte, and Accenture. Enterprise adoption research and frameworks. Heavier and more formal; useful when you need defensible material for a leadership conversation.
For builders and operators
Developers, PMs, founders, and the people shipping products.
- Blog. Simon Willison. The best hands-on writer on how to actually use LLMs. Practical, well-linked, updated constantly.
- Podcast. Latent Space. Swyx and Alessio. Pairs with the AI Engineer conference. The builder-flavored counterpart to Cognitive Revolution.
- Podcast. Lenny's Podcast. PM-centric, increasingly AI-flavored. Good on how product teams actually adopt these tools.
- YouTube. AI Engineer. Conference talks and workshops for engineers building with LLMs. Heavy signal-to-noise.
- YouTube. Nate Herk. Hands-on agent and workflow builds, mostly in n8n. Fastest way to see end-to-end automation.
- YouTube. Microsoft Developer. Azure AI, Copilot APIs, agent frameworks. The place to go if you work in the Microsoft stack.
- YouTube. Microsoft. Corporate channel. Skim for product announcements and keynotes; skip the marketing-heavy posts.
- YouTube. GitHub. Copilot deep dives and the developer side of AI tooling.
- YouTube. Databricks. Data and AI on the Databricks platform. Strong on RAG, MLOps, and governance for data and platform teams.
- YouTube. Langchain. Agent and LLM application patterns. Skewed to the LangChain stack, but the patterns generalize.
- YouTube. Matt Pocock. TypeScript and AI engineering. Sharp on craft and on how to use the SDKs well.
- YouTube. Product School. Practical PM content. Lighter than Lenny's; useful on-ramp.
- YouTube. Y Combinator. Founder framing on AI, plus Startup School talks. Useful even if you are not building a company.
- YouTube. YC Root Access. YC's deeper technical channel. More builder-flavored than the main feed.
Frontier and research
Where AI is heading, and the technical foundations underneath them.
- X. Andrej Karpathy. Ex-Tesla and OpenAI, now teaching-first. The clearest technical explanations of how LLMs work anywhere.
- X. Ilya Sutskever. Co-founder of OpenAI, now Safe Superintelligence. Rare posts, but each one is worth reading. One of the sharpest minds on where models are heading.
- Video course. Neural Networks: Zero to Hero, by Andrej Karpathy. Build a neural net, and then a GPT, from scratch. Finish this series and you actually understand the field.
- Podcast. Dwarkesh Patel. Long-form interviews with frontier researchers and operators. The closest thing to reading their papers in conversation form.
- Podcast. The Cognitive Revolution. Deeper conversations with builders and researchers.
- Video. Stanford CS230, Lecture 8: Agents, Prompts, and RAG. Graduate deep-learning lecture. Worth it if you want agents, prompting, and retrieval from first principles.
Sources To Get Started
If you are looking for a place to start, this is it. These build the foundation: how to think about the tools and how to prompt them well.
Courses and guides
- Anthropic Academy. The hub. Two tracks are worth your time: Claude for work for team and organizational use, and Claude for personal for individual projects and daily tasks.
- Google AI Essentials. A legitimate on-ramp if you are truly starting from zero. If you already use these tools daily, it will feel basic.
- Gemini for Google Workspace Prompting Guide 101. The clearest short guide to writing good prompts I have found. Persona, task, context, format — the framework applies to any model, not just Gemini.
- The 5 Step Playbook for 10x Your AI Productivity, Jeremy Utley. Thirteen minutes, very accessible. The "walk me through your thought process step by step" tip alone is worth the watch.
- LinkedIn Learning and Coursera both have solid catalogs beyond this. Most of the good material sits behind a subscription.
Tools worth trying first hand
Reading only gets you so far. Open these and use them.
- Agent Skills and Claude Agent Skills Explained. Learn about skills!
- Claude Projects. Bundle documents, context, and instructions in one place so Claude works from your material. Fastest upgrade to how most people use a chatbot.
- Google NotebookLM. Point it at your own sources and it becomes a research assistant grounded in them. The audio overview is an unusually good way to absorb dense material.
- Gemini image generation, Nano Banana. Google's image model. Strong at legible text in images and at editing from reference photos. Start in the Gemini app; move to AI Studio when you want more control.