A dedicated publishing area for weekly AI signal: commentary, public opinions, and practical analysis for professionals making decisions.
February 26, 2026
Perplexity's CEO argues multimodel orchestration is where the value lives. It's a sharp thesis — and exactly the argument you'd make if you don't build the models.
February 26, 2026
Karpathy says the era of typing code is over. His own example proves the era of knowing what to type never ended — the skill just moved up the stack.
February 26, 2026
A hedge fund co-authored a report that preceded an 821-point Dow drop. Then the attribution was quietly changed. The regulatory gap that made it possible is by design.
February 26, 2026
Starting a project with AI is easier than ever. Shipping a real product still isn't. The gap between demo and done is where the flattery falls apart.
February 26, 2026
A clean Coasean framework for AI and firms — until it declares the instruction layer solved and skips the part where physical work, accountability, and trust resist compression.
February 26, 2026
AI doesn't threaten engineering seniority because seniority was never about typing code. The question is whether organisations still produce the friction that builds judgment.
February 26, 2026
Block cuts nearly half its workforce and Jack Dorsey frames it as AI-driven reinvention. The severance is generous. The binary framing deserves more scrutiny.
February 26, 2026
A viral thread about escaping AI slop, delivered in the exact format it claims to transcend. The one useful idea is buried under nineteen paragraphs of atmosphere.
February 24, 2026
Harness engineering is a real practice with real results. But trust built on commit volume isn't the same as trust built on understanding — and most codebases aren't greenfield.
February 24, 2026
Cognition's CEO admits the original Devin was too early. The update is real progress — but the headline metric is PRs per week, not PRs that ship without a rewrite.
February 24, 2026
Code is cheap. Software isn't. But the line between the two keeps shifting, and most commentary treats a moving target as a fixed truth.
February 24, 2026
Will Manidis builds a beautiful metaphor about French and English gardens to critique how technology reshapes systems. The metaphor is a lens, not a law.
Subscribe to the AI Primer Briefing for weekly commentary on what is happening in AI and what it means for your work.