Most owners I talk to have already tried AI for their marketing. They ask Claude or ChatGPT for a Facebook post, get one back, and the post is fine. It's also a little too slick: it promises a bit more than they'd promise, says "limited time only" when there's no time limit, and sounds like every other business online.
The problem usually isn't the AI. Your values live in your head, and the AI only hears them when you remember to type them. On the days you're tired or rushed, it doesn't hear them at all.
This 3-part series is about fixing that. Part 1 covers the simple fix that works for everyone, and then a new tool from Anthropic called mods that takes it further for owners who want to go further. Part 2 gives you 7 example mods for social media, email and your blog. Part 3 walks you through setting one up and checking that it's safe.
If you only read this first part, you'll still leave with the most useful idea in the series.
The fix that works for everyone: write your values down
Before any tool, there's a page. Write down, in plain words, how your business talks and what it never does. Keep it under a page. Something like this:
How we talk: Like a neighbor over the fence. Plain words, no jargon.
What we never do: Invent urgency. Shame or scare anyone into buying. Quote a customer without permission. Promise results we can't guarantee. Use our faith as a sales pitch.
What we always do: Tell the truth about price and timing. Say when we're not the right fit.
That page does more than any setting or plugin, because it makes you decide what you believe about your own marketing before the AI writes a word. Part 3 has a fuller example you can copy and change.
Then give it to the AI every time, not just when you remember. If you use the regular Claude chat, the easy way is a Project: a workspace inside Claude where you can save instructions, so every conversation in that Project starts with them. Paste your values page into the Project's instructions and do your marketing writing there. There's no code and no new software, and it works today.
For a lot of owners, that's the whole answer, and it's a good one.
The next step: Claude mods
On October 1, 2026, Anthropic added mods to Claude Code. To understand them, you need to know what Claude Code is.
Claude is the AI assistant made by Anthropic. Most people use it in a chat window. Claude Code is a different way of working with the same AI: instead of only answering, it can take action on your computer. It saves files, runs programs, and works through a job one step at a time while you watch. You can use it in a terminal (the plain text window programmers type commands into), in the VS Code editor, or in the Code tab of the Claude desktop app. If you've never opened a terminal, the desktop app is the one to use.
Anthropic describes a mod as a small piece of code that changes how Claude Code looks and behaves. In plain English, every time Claude Code does something, it announces it. You sending a prompt is an event, and so is Claude getting ready to save a file or draw part of the screen. A mod listens for one of those events and gets a say in what happens next. It can watch the event and let it carry on, change it before it goes through, or stop it and handle it itself.
That's what makes a mod different from a values page in a Project. A Project gives Claude your instructions. A mod can enforce them. It can add your values to every request automatically, hold a draft that contains a phrase you've banned until you look at it, or pause a scheduled post and ask whether you meant it.
According to Anthropic's announcement, a mod can rewrite a prompt before Claude reads it, block or change an action Claude is about to take, approve or deny a permission request, add its own panel beside the conversation or a band above the box where you type, and add new commands. Some of Claude Code's own features are now built as mods.
You don't have to write the code
Mods are written in JavaScript or TypeScript, two programming languages, and you don't need to know either one. Claude Code includes a built-in skill for writing mods, so you can describe what you want in plain words and Claude writes the mod, installs it and reloads it while you watch.
The real work isn't programming. It's deciding what you want the AI to do, and not do, on your behalf, which is the values page again. A mod is only as good as the page behind it.
What "centered on Christ" means here, and what it doesn't
I want to be careful with this, because it's easy to oversell.
No tool puts Christ at the center of a business. That happens in the owner's walk with God, in prayer, in how he treats his people and his customers when nobody's watching. A values page and a mod can't do any of that for you.
What they can do is narrower and still worth doing. They can keep the words your business puts out from contradicting the faith you actually hold. "Whatever you do in word or deed, do all in the name of the Lord Jesus" (Colossians 3:17, NASB, the New American Standard Bible). Word or deed covers the Instagram caption you write at 9 at night, and it covers the email Claude drafts while you're on a job site. A tool that catches a dishonest line before it goes out under your name is a small way of taking that verse seriously.
And a tool doesn't keep watch for you. It's a reminder you set up while you were thinking clearly, so it can catch you when you're tired. You still have to read what goes out.
What a mod can't do
It only works in Claude Code. If you write in the regular Claude chat, mods don't apply, and the Project with your values page is your tool.
It can't judge for you. A mod can add your values or stop and ask, but it can't decide whether a post honors God.
It isn't sandboxed. Sandboxed means walled off from the rest of your computer. Mods aren't. A mod runs with the same access to your computer that Claude Code has, which means it can read your files and see your prompts. Anthropic says plainly to install mods only from sources you trust. Part 3 shows you how to check one in plain English.
Who should bother with mods
If you never plan to touch Claude Code, skip mods, and that's a fine answer. A values page in a Claude Project gets you most of the way there.
If you already use Claude Code, or you're curious enough to try the desktop app, mods are worth an afternoon. They turn "I hope the AI remembers who we are" into "the AI is set up to remember who we are."
In Part 2, I'll show you 7 mods built around the places your business talks to the public: social media, email and your blog.
Read the series: Part 1 (you are here) · Part 2: 7 mods that keep your marketing honest · Part 3: How to set up your first mod