What happens when executives stop nudging and start mandating? Coinbase CEO Brian Armstrong revealed that he gave engineers one week to onboard to AI coding tools—or be fired. Some didn’t make the cut. It’s a bold, polarizing case study in enterprise AI rollouts: speed of adoption vs. morale, governance, and code quality debt.

Google expands AI-powered search to 180+ countries with live booking capabilities, Meta freezes AI hiring after a $100M talent war that shook the industry, and the NYT reports AI comp packages now topping a quarter-billion. Plus, Smashing Magazine weighs AI’s energy costs vs. productivity gains in web design, and freeCodeCamp drops a beginner-friendly explainer grounding modern LLMs in the classic Bag of Words approach.

Brian Armstrong on mandating AI coding adoption at Coinbase — and why non-compliance led to terminations

A report from Armstrong’s appearance on John Collison’s “Cheeky Pint” podcast: Coinbase bought enterprise licenses for GitHub Copilot and Cursor, told every engineer to onboard within a week, and fired those who didn’t.

It’s a concrete look at executive-level AI adoption pressure — deadlines, tooling choices, and the cultural risks — useful for any team rolling out coding agents/org-wide AI policies.

After the mandate, Coinbase doubled down with ongoing training and monthly share-outs to spread effective AI practices across teams.

“AI is important. We need you to all learn it and at least onboard by the end of the week.”

Armstrong himself calls the approach “heavy-handed”; legal, morale, and equity considerations matter as much as speed when you enforce AI tooling.

Google rolled out its AI-powered search mode to 180+ countries, enabling complex multi-step queries like "find a restaurant in Paris with outdoor seating for 4 people at 7pm." The agentic search uses live web browsing and partnerships with OpenTable to execute actions.

The New York Times reported that AI researchers are now landing salaries worth a quarter-billion dollars, highlighting the extreme talent war driving unprecedented compensation packages across the industry. This follows Meta's hiring spree that saw individual packages reach $100 million.

Meta abruptly halted all hiring in its AI division last week following months of aggressive talent poaching with eye-popping compensation packages reaching $100 million. The freeze affects both external hiring and internal team transfers as Meta restructures its "Superintelligence Labs" into four groups. CEO Mark Zuckerberg personally recruited over 50 researchers from OpenAI, Google, and Anthropic.

A balanced explainer weighing AI’s productivity gains (code optimization, responsive layout generation, asset compression) against the environmental costs of AI infrastructure. It cites rising data-center electricity and water usage, and argues that page-level efficiency can be offset by model training/inference at scale. Practical guidance includes starting with lean web fundamentals, favoring task-specific models, batching non-urgent inference, and choosing greener vendors.

A crisp primer that defines Bag of Words (BoW), then implements it with sklearn to show vocabulary building and document-term matrices. It explains where BoW falls short (no order/meaning, sparse vectors) and how it led to TF-IDF, embeddings, and transformers—useful context for onboarding teammates to LLM concepts. Short, code-forward, and beginner-friendly.

Another week of separating AI signal from noise. If we saved you from a demo that would've crashed prod, we've done our job.

📧 Got a story? Reply with your AI tool wins, fails, or war crimes. Best stories get featured (with credit).

📤 Share the skepticism: Forward to an engineer who needs saved from the hype. They'll thank you.

✍️ Who's behind this? The Augment Code team—we build AI agents that ship real code. Started this newsletter because we're tired of the BS too.

🚀 Try Augment: Ready for AI that gets your whole codebase?