A scenario, told by a marketing lead in a twelve person team. The steps are the app’s own, as the app and the BearingBridge help center show them on November 3, 2026.
The job, and why it hurt
Four of us write for the brand: me, our sales lead, the person who answers support, and a freelancer who drafts our articles. Each of us used AI to go faster, and each of us briefed it our own way.
You could tell. Our LinkedIn posts were breezy and our newsletter was stiff. A support reply once promised more than our ads ever had. The rules lived in my head and in an old brand book PDF nobody opened.
I wanted the voice written down once, in a place every AI run would read it, whoever was writing.
What I did in BearingBridge
Write the voice down first
In Brand Content, then Text, the Brand foundation page opens for the brand picked in the header. It’s a copy studio with seven formats, from the manifesto to the Tone of voice guide. I picked the guide. Its one needed field is Personality; I added Contexts to cover and our Non-negotiables.
I kept The plan first, so the run stopped at an outline I could correct by hand, for free, before any copy was written. The guide came back with principles, do and do not, and rewritten examples, its tokens and price next to the version. Where my brief left gaps, the writer listed them in a To confirm note, apart from the copy.
Turn it into guidelines the AI follows
A guide written on that page is a document for people. It doesn’t instruct the AI by itself. That job belongs to the brand guidelines, in the brand editor: Brands in the administration menu, Edit on our brand, then Voice & tone.
The form is long, so I didn’t start from a blank one. I pasted our About page address into the reference box and clicked Draft with AI. The draft also read our brand profile, up to eight of our approved or published posts, five accepted articles and anything already in the form. Where the material said nothing, the field stayed empty rather than guessed.
Then I went through it with the guide open beside it:
| Field | What I put there |
|---|---|
| Voice traits | Three, each with what it must not turn into |
| Tone by context | One line each for social posts, articles, ads, email, Service Desk and reports |
| Prefer, Avoid, Never use | Our words, the ones we steer away from, and a short list of hard bans |
| Sounds like the brand | Real lines from our best posts |
I kept Never use short. It applies to every output, so a word banned in ads is banned in a support reply too.
Add the look from the brand book
At the top of Visual identity, Read visual identity took our brand book as a PDF (up to 20 MB) and filled only the palette, logo, typography and imagery fields. The voice I had just reviewed stayed as it was. Then I clicked Save changes. Nothing is kept before that, but a draft is charged as soon as it runs.
On cost: editing the guidelines is free. Each AI draft is paid from the wallet, and the message under the bar gives its price after “This draft:”.
The result
Every text run for the brand now appends the guidelines: copy, articles, social posts, ad copy, reports and the Service Desk assistant. Each one reads only its own Tone by context line: a support reply gets the Service Desk line, an ad gets the ad line. In the article workspace, Check guidelines scores a draft against them and lists each breach, and our freelancer now runs it before sending anything.
The guidelines change what’s written next, not the pieces already out. When we edit them, the next run picks up the change.
The video below walks through the same routine in the app.







