Will's Journal.
AI Agents State Machines for Agent Loops: Determinism Where It Matters
Agent loops need freedom inside a cage. Explicit states, legal transitions, revision ceilings, and escalate paths are how non-deterministic models ship deterministic ops.
AI Agents MCP vs Native Function Calling: Portability Tax vs Shortest Loop
Native function calling wins for one app's short tool loop; MCP earns the tax when tools must be shared and governed across hosts—not a LangChain swap.
Websites The Trades and SMB Website Playbook: Speed, Trust, and the Phone Call
HVAC, plumbing, and local service sites that win: fast loads, proof up front, and a phone call path that works on a job-site phone.
Automation Cron vs Webhook: Pick the Trigger That Matches the Failure Mode
Choose cron vs webhook by failure mode: webhooks miss when subscriptions die; schedules miss when the job never runs — hybrid reconciliation covers both.
AI Visibility Schema Markup That Answer Engines Actually Use
Skip schema soup. Ship accurate Organization, FAQ, Article, and LocalBusiness JSON-LD that helps answer engines trust and cite your pages.
AI Visibility Pick an Engine by Where Your Buyers Ask — Then Keep a Baseline Everywhere
Optimize first for the engine your buyers already use — ChatGPT, Perplexity, or Google AI Overviews — then keep a thin baseline on the other two surfaces.
Automation Schema Contracts Between Tools: The API Discipline Most Automations Skip
Why automations need schema contracts between tools, how to validate JSON in n8n, and how to stop silent nulls from poisoning your CRM.
Websites Google Business Profile Gets the Call — Your Website Closes the Ones Who Need Proof
On a limited budget, fix Google Business Profile first for map calls — then add an owned site for proof, services, and a fallback if Google suspends you.
AI Agents RAG That Does Not Lie: Retrieval Contracts for Business Knowledge
Retrieval is search, not truth. Production RAG needs corpora rules, citation requirements, refuse-on-empty behavior, and contradiction handling — or your agent will invent policy with confidence.
AI Agents LangGraph vs CrewAI vs a Custom Loop: Choose Control, Not Fashion
Pick LangGraph, CrewAI, or a custom loop by how much control you need—then compare them on the same golden set and cost band, not on Hacker News fashion.
Websites Lighthouse 90+ Without Killing the Design
How to keep cinema-grade brand design and still hit strong Lighthouse and Core Web Vitals on real phones — not only on a desktop lab run.
AI Visibility From Knowledge Panel to Model Memory: Owning Your Brand Facts
Knowledge panels and AI model memory both depend on consistent, corroborated brand facts. Here is how smaller brands own the narrative without fake Wikipedia.