Practical insights for business owners and teams
Straight answers on growing your business, improving operations, and putting technology to work. No fluff, just practical guidance.
Series
3 parts · updated September 16, 2026
Three measurements of how AI agents actually use a skill library: why a list failed, why subagents never browsed it, and what Claude Code's native plugin evals do and do not prove.
2 parts · updated August 26, 2026
How we taught our AI systems to keep the code they already wrote: the registry, the first paired benchmark, and what each measurement changed about the method.
We ran Claude Code's new plugin eval harness on a real plugin: 8 cases, 46 agent runs, about $12. Here are the three ways the results JSON misleads you, what the eval sandbox can and cannot reach, and the five-line hook that fixes the plugin-root variable Bash never sees.
The first paired benchmark of our procedural-memory registry measured a deterministic share under 0.5% on one workload and about 55% on another, caught its own row-append step failing silently, and ran under two conditions that rule out any arm-versus-arm verdict. Here is what held up, what did not, and the six variables the next cohort holds fixed.
We counted how often our AI subagents invoked skills from a several-hundred-skill library: zero of fifteen dispatches. Here is why cold-started agents never browse the library, the three injection levers that work, the spike methodology that proved the hook surface, and the log-first measurement instrument now watching every dispatch.
Most AI automations re-derive the same work every run. Here is how procedural memory, storing the working code rather than a description of it, makes AI agent workflows faster, cheaper, and more predictable.
Rhize Tasks turns Jira work into a realistic daily plan across Google Calendar and Apple Reminders, with local-first privacy and human approval controls.
Flat lists of AI agent skills fail silently as they grow. Every failure we hit was a relationship failure, so we built a generated graph, and the case for generating it came from our own code drifting within hours.
Most AI marketing advice is a tool list. This is the order of operations for a $1–5M service business — what to fix first, what to skip, and how to tell if it worked.
AI search now handles early buyer research, so blog traffic falls while branded visits rise. Here's how to redesign your website for pre-informed buyers.
Inside our build for an industrial supply distributor: a 155,000-product Neo4j knowledge graph, human-approved SKU matching, and search from a photo.
Inside our build for a major asphalt paving company: a 5:30 AM paving-decision text, an end-of-day KPI pipeline, and where AI actually earns its keep.
Every AI term you'll actually meet running a business — agent, MCP, AEO, PRD, LLM — defined in plain English. 32 terms, no jargon, no hype. (139 chars)
A Claude skill is an SOP that teaches Claude to do a job your way. A plugin is the binder of skills you install once. What they do for a business.
AI Overviews now answer 68% of local searches. Here's how we build location pages that rank in Google and get cited by AI — step by step, with screenshots.
SEO ranks you in Google's links. AEO gets you picked as the answer. GEO gets AI tools to name your business. The differences, and what to do first.
A PRD (Product Requirements Document) is a written plan for what you're building, who it's for, and what "done" looks like — and it's how you brief an AI builder.
An AI agent is software you give a goal, not just a question. It plans the steps, uses your tools, and finishes the job — unlike a chatbot, which waits. (148 chars)
Claude Code is an AI assistant that works inside your computer — it reads files, runs programs, and finishes multi-step work. A guide for owners.
An MCP server is a standard connector that lets an AI assistant safely use one of your business tools — calendar, CRM, files. Here's what that means, in plain English.
Obsidian is a free note app — not the volcanic glass — that stores notes as plain files you own and links them like a personal wiki. That's the "second brain."
Rhize comes from 'rhizome' — the root network that makes weeds impossible to kill. Here's why that's the blueprint for AI that runs your business without you.
The Obsidian Second Brain plugin brings Zettelkasten workflows, semantic search, and vault automation to Claude Code — full skill and command reference.
Project Launcher is a Claude Code plugin that turns an idea into a stress-tested PRD, reviewable visual plan, and a scaffolded GSD v2 project directory.
rhize-devflow is a Claude Code plugin enforcing session hygiene, error triage, cache-consistency checks, and Sentry instrumentation for production work.
rhize-meta is a Claude Code plugin for skill governance: vetting external agent skills safely with rhize-skill-forge and refining your own from usage.
rhize-ops is Rhize Media's Claude Code plugin for running a team — delegate work to teammates via Jira and Slack, then track skill usage with a live dashboard.
Rhize Plugins is a public Claude Code plugin marketplace with 6 plugins for SEO, knowledge management, and dev workflow. See what each does and how to install.
The SEO/AEO/GEO plugin turns Claude Code into a search optimization practitioner. See every skill, command, and DataForSEO credential setup step it ships with.
Skill Forge is the @rhize/skill-forge npm CLI that quarantines, scans, and vets Claude Code skills and MCP servers before they touch your working set.
A practical way to use AI in content operations: prepare better briefs, draft from approved sources, keep editorial review, and measure the whole workflow.
Most AI marketing guides are tool lists with unsourced statistics. This one cites its sources, says which claims don't hold up, and covers what implementation actually costs.
New research from MonitLabs across 27,000+ domains reveals backlinks still help in traditional SERPs but play a much weaker role in Google's AI Overviews. Here's what the data shows.
Where to start
Start with a free look at your online visibility. See what the scan finds before deciding on your next step.
The next step is a 30-minute call.
Book a 30-minute call and we'll map where the business depends on you — and what it looks like once a system you own runs the routine. You leave with a plan, whether you hire us or not.
No pitch — you leave with a plan. We'll tell you exactly what we'd do, whether you hire us or not.
