agents

Long-running agents are not defined by the model call. They are defined by the state, rules, tools, failures, and corrections that survive it.

Ken Griffin watched PhD-level finance work compress from months to days. The interesting question is whether bank AI controls are designed for the layer where the work now lives.

When execution gets cheap, the scarce work moves to framing, review, and the systems that preserve judgment.

Model risk reviews the model. Application security reviews the application. Neither sits behind the agent at execution time, watching the verbs as they go out.

Shopify's pi-autoresearch got 300x test speedups. But the real insight isn't performance — it's that the pattern works on anything with a number.

A Chinese article argues CLI is becoming the AI plugin format. I've been living this for months with 442 tools. The article is right about CLI. It's wrong about what makes CLI work.

When choosing CLI vs MCP, pick the one you can undo. CLI wraps into MCP cheaply. MCP does not unwrap.

Anthropic shipped a hosted agent platform. Its architecture looks familiar. Here's what a solo builder can learn from how they decoupled the brain from the hands.

Karpathy's LLM Wiki pattern is a good starting point. Here's what changes when you run it for real — enforcement over convention, decay over growth, and knowledge that fires without being asked.

A 96% rejection rate that was actually a 96% false positive rate — how a monitoring blind spot turned a productive overnight batch into apparent failure.

Most advice about making CLIs agent-friendly is just good CLI design. Only four principles are actually agent-specific.

Smart model plans, cheap model builds. The pattern everyone's converging on for AI coding agents — and the piece nobody's shipped yet.

I turned the seven patterns into a zero-dependency Python library. Then I added MCP bridge support. Here's what I learned about the gap between patterns and code.

Three independent authors converged on nearly identical patterns for CLIs that AI agents invoke. Here's what they agree on, what's missing, and why nobody has built a framework for it yet.

Willison says CLIs beat MCP. Cloudflare says server-side code mode beats both. They're both right, because they're answering different questions.

Engineering metaphors give you clean abstractions. Biology gives you resilient ones. Twenty design heuristics from four billion years of R&D.

Every skill you build is one of two things. Knowing which changes what you build next.

Biological metaphors in AI systems break at the autonomic-deliberate boundary. The fix isn't dropping biology — it's getting the neurology right.

Why the future of agentic trust depends on liquidating prompt-first reasoning for a metabolic core.

Language doesn't describe metabolism. Language is metabolism — of meaning, between minds.

Design for the failure modes of your medium, not the capabilities. Then watch the rules dissolve themselves into programs.

On the metabolic necessity of pruning agentic context to survive the entropic heat death of the credit balance.

Why we should stop treating AI as a chatbot and start treating it as a metabolic organism governed by credit scarcity.

When epistemic trust runs dry, generative synthesis regresses into mechanical synchronization and eventual structural dissolution.

Tokens are energy. Text is mass. The context window is the budget. The rest is plumbing.

Tool descriptions were just the first thing to evolve. Everything in an agent's context window is a genome under selection pressure — and taste decides what counts.

We build agent tools the way medieval farmers bred crops — by hand, by instinct, one season at a time. There's a better loop.

The word 'agent' makes us think in nouns. The better designs start with verbs.

We called them agents. But the word is doing more harm than we think.

I deployed twelve AI agents to polish a CV. Five would have been plenty. Here's what the waste taught me about agent team economics.