Can your SOPs answer questions?
Your operating knowledge lives in SOPs, rate sheets, and exception playbooks written for people. Why AI tools fail on them, and what to fix first.
Notes from the work
Insights, sourced from what we actually found.
Six short pieces, each tracing back to a real repo file — a live-store audit, our own CRM log, or the method behind a service. No invented statistics.
Read next
Six posts.
You lose RFPs on turnaround, not capability.
Every RFP and security questionnaire starts from a blank page and a document hunt. A response desk built on approved, sourced answers fixes that.
What an AI-agent readiness score actually measures.
A readiness score is a proxy built from a small set of checks, not a guarantee an AI agent will find or quote your store correctly. What it stands in for, and what to do with it.
Why we publish what we couldn't verify.
Every audit hits a limit: a page we couldn't reach, a claim we couldn't confirm. We publish those gaps instead of hiding them — and that is the point, not a weakness.
Can an AI agent answer your store's shipping question?
What we found reviewing ten independent D2C stores and four more with a full crawl: product pages are fine. Shipping, payments, and FAQ answers are not.
What "agent-ready" actually means.
Agent-ready is not a certification or a badge. It is a checkable property: can an AI system retrieve and quote a correct, current answer from your own source material.
The 5-point AI-readiness spot check, explained.
What the spot-check on this site actually looks at, in plain terms, and why five checks instead of a longer scored audit.
Why we label our demos DEMO.
A DEMO label is not a disclaimer we bury in fine print. It is a promise about which parts of a shown result are real and which are illustrative.
What we learned sending cold emails from our own CRM.
Aggregate, anonymized lessons from auditing our own outreach CRM — bounce patterns, drafted-versus-sent gaps, and what we changed as a result. No recipient names or domains.
How to read an AI spend bill.
A plain-English guide to the line items that actually drive an AI vendor bill up: repeated context, model choice, and retries — and where to look first.
A clear first move