How I Use AI, and Where I Draw the Line

Field notes · AI, day to day · about 5 minutes

A lot of people get a little uneasy when a marketer says they use AI. Here's exactly how I use it, the rules I keep, and how I'd help you use it in your own business.

My five rules

1. Client information stays protected.
2. Every number gets checked against its source.
3. A person approves anything that goes out.
4. Nothing gets invented.
5. Decisions stay with people.

How I see it

How I see it

I understand the unease. The word AI now covers everything from a helpful first draft to a robot writing fake reviews, and from the outside it's hard to tell which one you're getting. If you work with me, this page is what you're agreeing to.

AI is a fast, tireless assistant that's confident about everything, including the things it gets wrong. That combination makes it useful and risky at the same time.

It's very good at reading a lot, sorting a lot, and getting a first version on the page. It's bad at knowing which number matters, what your customers actually care about, and when it's wrong. So I use it for the first pass and keep the judgment, the final read and every decision for myself.

Day to day

What I actually use it for

These are real, weekly uses, not a wish list.

• Reading the pile. When a client hands me a messy export, AI helps me see the shape of it fast. On one hiring account, it helped me reconcile 820 applications across two systems so I could see that hiring was quietly running twice. Spotting that took the tool a few minutes. Deciding what to do about it took me.
• First drafts. Reports, proposals, emails, social posts. I get a rough version in minutes, then rewrite it until it sounds like me and says only what I can stand behind. This quarter I'm running my own social media this way: an AI assistant drafts the posts into my scheduling tool, and nothing goes out until I've read and approved it. I'm tracking the time it takes and what it earns, and I'll publish the results in December.
• Research and testing. Before I sold search and AI visibility as a service, I tested it on my own business. Google's AI Overview thought "Bellmark Projects" was a typo for Belmark, a packaging manufacturer here in Kentucky. I fixed what I could control and then refused to call one good search a result, because AI summaries don't give the same answer twice.
• Turning notes into a plan. I paste in messy meeting notes and get back a task list. Then I fix the owners, the dates, and anything it guessed.
• Checking my own work. I'll ask it where my writing is unclear or where a report skips a step. It's a second set of eyes, never the last one.

Guardrails

My guardrails

1. Client information stays protected

I don't paste private client data into AI tools that can use it for training. When I need to work with real numbers, I use tools set up to keep that data private, and I strip out names and anything personal first.

2. Every number gets checked against its source

AI will hand you a wrong number with total confidence. Nothing goes in a report until I've checked it against the actual system. When I once reported two automated messages and the real count turned out to be seven, I corrected my own record rather than let the number stand.

3. A person approves anything that goes out

Nothing AI drafts is sent to a client, posted publicly or put in front of a customer until I've read it and said yes.

4. Nothing gets invented

I don't use AI to make up reviews, testimonials, results, customers or quotes. If I can't prove it, it doesn't get published.

5. Decisions stay with people

AI can recommend. Your budget, your hiring, your customers and your brand get decided by you and me.

How I can help

How I can help you use it

You don't need to overhaul your business to get value out of AI. Most of the businesses I work with are better off starting small.

• Find one task you already dread. Weekly reporting, answering the same customer question, sorting applications, writing job posts. We pick one, set it up with guardrails, and check what comes back.
• Set it up so it's safe. The right tool, the right account settings, clear rules for what goes in and what doesn't, and a person who approves the output.
• Train your team on it. A short, plain-language walkthrough of what the tool is for, what it isn't for, and how to check its work.
• Make sure AI describes you correctly. People now ask ChatGPT and Google's AI for recommendations. I check what those tools say about your business and fix the sources they read: your Google Business Profile, your listings and your website.

And if you'd rather not use AI at all, that's fine too. Everything I do works without it. It just helps me get to the useful part faster.

Sound familiar?

Not sure where AI fits in your business?

If you have a feeling you should be doing something with AI and no idea where it would fit, or you've tried it and don't trust what it gave you, that's a normal place to start. It's usually a short conversation.

Start a conversation