Jensen makes the case for open source AI
AI Weekly Round-Up #43: Building an enterprise knowledge base and redoing your Claude skills
Hi friends,
It’s been another one of those weeks, it still feels like we’re drinking from the fire hose.
If you’re new here, every Sunday I curate the top AI articles, tweets, podcasts and tutorials (so that you don’t have to).
Let’s jump in:
If you only read 3 things
Open weights and American AI Leadership — Jensen and 50 AI companies signed a pro-open source letter. (There is one notable exception).
How Cerebras Built Its Enterprise Knowledge Base — A detailed explainer on connecting all your company’s knowledge (Slack, shared drives, GitHub) so that everyone can access it.
When AI Skills Help—and When They Hurt — Opus and Fable require less handholding (which means we need to simplify our skills).
News from the frontier labs
Claude Opus 5: Anthropic’s New Flagship AI Model — Early reports are that it’s Fable-like intelligence at Opus pricing.
ChatGPT’s new model broke out of its sandbox — The AIs have started going rogue (just to cheat on their benchmarks).
Why Chinese AI Models Threaten U.S. Labs — Ben Thompson on why we should foster more US-based open source.
Teach Claude Cowork a Skill by Recording Your Workflow — Creating skills continues to get easier.
Cursor Router: Automatic Model Selection at 60% Lower Cost — Model routing is gaining steam (but will it work)?
Stop dabbling. Start building your AI advantage
Are you still using AI mostly for search, summaries, and one-off prompts?
Learn to turn Claude Cowork and Claude Code into practical workflows for research, meeting prep, reporting, and repetitive analysis — without becoming an engineer.
After this self-paced course, you’ll go from experimenting with AI to building tools that create real leverage for you and your team.
Use code CHIEF-OF-STAFF for a 30% discount through July 31.
From the desk of Khe Hy
AI is coming for your Slack workspace — This new form factor is going to pull in a lot more AI noobs.
How AI is reshaping the workplace
How a Top Hedge Fund Grew AI Token Spending 86x — The Man Group on building routers, connecting to unstructured data and what they look for in new hires.
Forget FDEs, we need forward deployed “analysts” — There’s a bigger shortage in domain experts who understand AI.
The Privacy Risks of Always-On AI Recording — The norms on recording are changing (and apparently people now record their dates).
Sierra’s MCP Gateway for Secure Enterprise AI Access — The daunting process of building your own MCP for your internal agents.
Substack Launches AI Detection for Writers — Newsletters (including this one) now get an AI score. This will likely expand to other social networks.
For our AI builders
How Replit Is Building a Self-Driving Company — Good self-learning examples from sales and marketing.
How Granola Built a $1.5 Billion AI Meeting Company — They have a strong foothold with your important context, but what comes next?
Building a Simpler Coding Agent in a World of AI Slop — I’m intrigued by learning the Pi coding harness as a side project.
Stop dabbling. Start building your AI advantage
Are you still using AI mostly for search, summaries, and one-off prompts?
Learn to turn Claude Cowork and Claude Code into practical workflows for research, meeting prep, reporting, and repetitive analysis — without becoming an engineer.
After this self-paced course, you’ll go from experimenting with AI to building tools that create real leverage for you and your team.
Use code CHIEF-OF-STAFF for a 30% discount through July 31.
Jensen frames open weights as a U.S. leadership advantage—which matters first in practice: sovereignty, enterprise adoption, or faster ecosystem innovation?