AI guides
Using AI safely is six questions, answered in writing.
Not a binder. Not a certification. It is the short list of written answers that keeps your tools, spending, information, AI use, and decisions from quietly drifting.
Free resources
Use these today. No email, no sign-up.
Free
30-Day Starter
Governed in thirty days — half an hour at a time.
Free
AI & Tool Inventory
Every tool, one page — including what the AI can see.
Free
AI Use Policy Starter
A one-page AI policy your team will actually read.
Free
Decision Log
Decisions, exceptions, and open questions — in one written place.
Free
Official Governance Sources
Where the ideas come from — the official, free originals.
Free
Safe Input Guide
What’s safe to enter into AI tools — and what never is.
Tools
Every tool has one named owner, and new tools have one approver, so the stack stops growing by accident.
Spending
Subscriptions are decisions, reviewed on a date, not charges that renew because nobody was looking.
Information
Each kind of information has one home and one owner, so corrections reach every copy and departures don’t take knowledge with them.
AI
You can say what your AI tools touch, what never goes into them, and when a human checks the output.
Decisions
Operating decisions, and the exceptions to them, live in one written log rather than in memory or chat threads.
The six questions. Select one to read the answer.
01
Who owns each tool?
One name per tool, not a committee, not “everyone.” The owner knows what it costs, what it touches, and whether it’s still earning its keep. If nobody can be named, that itself is the finding.
02
Who approves new tools?
One approver and a thirty-second rule: before a new subscription starts, it gets a line in the inventory, what job it does, what it replaces, what it costs. Most tool sprawl is just sign-ups nobody wrote down.
03
What information is safe to enter into AI tools?
A written list everyone has seen. Client names, financials, credentials, and anything you’d not post publicly stay out unless a specific tool has been deliberately approved for it. Unwritten understandings don’t survive a busy Tuesday.
04
When does AI-generated work require human review?
By default: always, when it reaches a client, a contract, a price, or a public post. The rule is one sentence, “a person who understands the work reads it before it leaves,” and it’s written where everyone can point to it.
05
Where are exceptions and unresolved decisions recorded?
In the decision log, the same place as the decisions themselves. An exception that lives in a chat thread becomes a precedent nobody agreed to. An unresolved decision that isn’t written down gets made by default, usually by a renewal date.
06
When will the rules be reviewed again?
On a date, not “when things calm down.” A quarterly half-hour re-read is enough for a small business: what changed, what rule no longer fits, what exception should become the rule.