Knowledge work
is being rebuilt.
I'm documenting it live.
Patterns for builders. What works, what doesn't, and what it actually feels like when you carry both customer accountability and real agent infrastructure.
Now
AI-Native Work Patterns
How knowledge work changes when agents are real teammates. Writing about what actually works — fleets over single agents, markdown as source, llms.txt as distribution, and the cognitive modes that matter.
Agent Infrastructure
Running real agent workloads in experimentation accounts. Cost discipline, graceful EC2 lifecycle, observability without compliance overhead, moving secrets to GitHub Variables, and the unglamorous patterns that make agents sustainable.
Agent Patterns
MCP servers, memory systems, multi-agent workflows, and the operational layer around them. Finding what generalizes when agents are daily tools instead of demos.
Latest signal
Per-topic curated briefs — agents, MCP, inference, AI coding, and the physical stack. Not a firehose.
-
Agent Commerce — June 10, 2026
Rain launches Agent Control Layer for programmatic agent spending guardrails; Santander's Getnet enables Mastercard Agent Pay acceptance, completes firs…
spending-controls · virtual-cards · stablecoin · Visa · Mastercard · Mastercard Agent Pay
-
Agent Reliability — June 10, 2026
Perplexity launches 'Search as Code': agents write Python to compose retrieval, rerank, and dedup primitives directly; LlamaParse adds word/line/cell-le…
retrieval-architecture · agentic-search · reranking · cost · ingestion · parsing
-
Agent Stack — June 10, 2026
LangSmith Sandboxes goes GA: microVM-based persistent compute for long-running agents; Harness-1: open-weight 20B search agent beats GPT-5.4 on recall v…
agent-harness · sandbox · long-running-agents · langchain · open-weights · working-memory
-
AI Platform — June 10, 2026
DeepSeek V4 pricing triggers China-wide AI API price war — Tencent Cloud cuts DeepSeek-V4 hosting 97.5%, Xiaomi cuts MiMo-V2.5 99%; Google's GKE Inferen…
pricing · china · deepseek · tencent-cloud · xiaomi · routing
-
Builder Loop — June 10, 2026
Claude Fable 5 reaches general availability across GitHub Copilot surfaces; GitHub extends automatic PR security validation to third-party coding agents…
github-copilot · anthropic · model-release · ide-integration · github · security
Featured
-
The AI Chief of Staff Is a Fleet, Not an Agent
The AI Chief of Staff pattern is everywhere in May 2026 — and every implementation makes the same mistake. They build one agent that generates a morning brief. A real chief of staff is a fleet.
-
Markdown Is the Operating System. Everything Else Is a Render.
The AI-native file format stack is three layers: Markdown, YAML, and CSV for authoring and agent operations; PDF, DOCX, and PPTX only exist as rendered outputs at the moment of distribution. Treating rendered formats as source creates friction that compounds across every agent interaction.
-
The Website Is Not the Product
AI crawlers are extracting content at 60,000 pages per visitor sent back. The website is becoming a fallback UI. The real distribution layer is structured, machine-readable context — and nobody is building for it yet.
-
Every AI Tool Encodes a Cognitive Mode
The friction you feel using the wrong AI tool isn't a prompting problem — it's a mode-matching problem.
-
llms.txt — Making Your Site Navigable by Agents
HTML was designed for browsers. llms.txt is the interface layer that makes your site a first-class citizen in agent workflows — discoverable, consumable, and citable in a single request.
-
Skills as Institutional Memory — Why Individual Craft Doesn't Compound Without Distribution
You built a workflow that saves four hours a week. It lives on your laptop. Nobody else benefits. That's not a knowledge management problem — it's a distribution architecture problem.
-
How Many Claude Code Sessions Can — and Should — You Run Simultaneously?
The question splits into two: can is a systems problem, should is a human one — and most people crash on the wrong side.
Recent Writing
-
Go-to-Market Was a Relay Function. AI Makes It a Decision Function.
The traditional go-to-market role was designed as a last-mile relay — carrying product decisions to customers. AI collapses that relay into a decision loop. The org chart hasn't caught up.
-
Builder's Daily Is a Fleet, Not a Newsletter
Updated:How Builder's Daily works: 14 bounded scanner agents, filesystem memory, explicit curation, composable publish skills, and a public machine-readable surface — not one digest in your inbox.
-
AI Subscriptions Are Secretly Usage Models
Claude, ChatGPT Codex, Cursor, Devin, Copilot, Gemini, Perplexity, and Grok all look like subscriptions. The real product is the reset policy, the overflow rule, and how much autonomous work you can squeeze out before the meter shows up.
-
Autonomous Agents Break Flat-Rate Pricing
The problem with AI subscriptions is not that vendors are greedy. It is that autonomous agents turn flat-rate pricing into a bad fit for the actual cost shape of the work.
-
How To Optimize Agent Subscriptions Without Getting Tricked
The best way to get more out of Claude, Cursor, Codex, Devin, Copilot, Gemini, and Perplexity is not clever prompting. It is operational discipline around scope, context, retries, and reset timing.