I Stopped Using Google and Amazon to Shop: ChatGPT Replaced Both
Agentic commerce shifts power from checkout pages to AI agents, making authorization, trust, and liability the real battleground.
nate
15 posts
While Executives Panic, This Skill Is Becoming Rare
The rare 2026 career skill: connecting AI know-how, business workflows, and a concrete implementation plan.
Lindy, JP Morgan, and OpenAI Built the Same Layer. Most Teams Haven’t Yet.
AI agents now need a separate judgment layer to validate risky actions before they execute in real systems.
I watched $5.5 billion move in one week. Your AI budget is wrong.
The Lily incident shows why AI budgets must price agent permissions, auditability, revocation, and architecture before procurement.
Human oversight is not slowing AI down — it is protecting it
AI capabilities are moving fast, but economic adoption depends on trust, audit trails, governance, and institutional inertia.
How Fast Can We Really Replace Workers with AI?
AI labor displacement depends less on raw capability than on deployment, adoption, integration, and social inertia.
Everyone is getting better at prompting. Almost nobody is packaging the work.
The next productivity gain from agents is not longer prompts, but reusable scaffolding around repeatable workflows.
Frontier or Comfortable: Where Do You Actually Stand?
Why the real value of AI now comes from practical fluency, domain evaluation, and the compounding lead of people already working at the frontier.
When Code Meaning Breaks: The Gap Undermining Security
Mozilla’s Mythos experiment points to a future where software trust comes from verifiable agentic review, not merely human authorship.
While Markets Panic, This Is What Happens
The AI advantage is shifting toward speed: testing, evaluating, and adopting models before slower organizations react.
I tested OpenClaw against model churn: what survived
OpenClaw is becoming a more durable runtime layer: the real strategy is to build workflows that survive model churn.
16 Million Fake Accounts Stealing AI Capabilities
Anthropic frames capability extraction from Claude as a national-security issue, linking reasoning data, model access controls, and censorship use cases.
The real primitive behind AI agents
Nate B. Jones argues that AI agents need more than access: they need software that exposes the meaning of work.
Why AI Is Harder to Contain Than Nuclear Weapons
Nuclear weapons depend on physical bottlenecks. Advanced AI can end up as files, outputs, and capabilities that move far faster.
Consumer AI’s unresolved assistant problem
AI agents can act, but they still make users manage them. The next leap is useful anticipation without extra overhead.