Brand voice

Brand voice AI that learns from your edits.

ADpilot drafts each brand's voice from its website, then compares every draft you edit with the version you saved and keeps the difference as rules for that brand: sentence length, tone, emoji, hashtags, the call to action and words to avoid. The next drafts follow them.

From €79/mo. 7-day free trial of Scale, no card.

One edit

What it learned from you.

Draft: Cold brew is SO easy, you are going to love it!! Grab your beans and let's go!

Your edit: Cold brew does not need a fancy setup.

Sentence length
Shorter
Tone
Calmer, less hype
Exclamation marks
Removed
Avoid
"SO easy", "let's go"

Use photos of the real roastery and team rather than illustrations

Imagery · asked 2 times · suggested as a rule

An owner or admin accepts it. Only the short rule reaches the writer.

Demo brand Demo brandAn example edit on a demo brand draft, and the client suggestion from the demo brand's review settings.

How it works

Three steps.

  1. 01

    It reads your website

    Setting up a brand from its address drafts the tone, what the brand covers, trust signals, approved calls to action and hashtags, for you to check.

  2. 02

    It learns from your edits

    Save an edit and the change is compared with the draft: shorter or longer sentences, more casual or formal, emoji and hashtags added or removed, a softer or stronger call to action.

  3. 03

    It writes by the rules

    Every caption, title and article is written against the voice and what it learned. Rejection reasons count too, recent ones more than old ones.

On the real screens

See it in the app.

Setup

Your website is the first brief.

Give it the brand's name, website and one sentence about the business. It reads the site and drafts the voice, look and market for you to review.

The brand setup screen: name, website and a short description of a coffee roastery

Your edits

The edit is the lesson.

Change the hook, the caption or the hashtags and save. The difference between the draft and your version is what it learns from.

The post editor with the hook, caption and hashtags of a carousel next to its live preview

Client feedback

Requests that repeat become rules.

When a client keeps asking for the same change, ADpilot suggests a short rule. An owner or admin accepts or dismisses it; the client's own words never reach the writer.

Suggested from client feedback: use photos of the real roastery, with Accept and Dismiss

Real ADpilot screens, demo brand

You decide

What you control.

  • Tone, and what the brand never sounds like
  • What the brand covers
  • Trust signals it may use (it never invents them)
  • Approved calls to action and hashtags
  • Whether posts carry a call to action
  • Which client suggestions become rules

Specs

The details.

Specs
How it works in ADpilot
Learns from Your saved edits, rejection reasons, rewritten image prompts and accepted client suggestions
What it notices Sentence length, tone, emoji, hashtags, call to action, vocabulary, patterns to keep and things to avoid
Memory The last 50 text lessons and the last 30 image lessons per brand
Scope Per brand: one client's style never reaches another brand
Languages 22, set per brand

Questions

Straight answers.

Where do I see and change the voice?

In each brand's settings under Voice: tagline, tone, what the brand covers, trust signals, approved calls to action and hashtags. You can regenerate it from the website at any time.

Does one brand's voice leak into another?

No. What it learns is stored per brand and only used for that brand's drafts.

Does it learn from images too?

Yes. When you rewrite an image prompt and approve the post, the change is kept as an image preference for that brand.

Will it make up awards or numbers about my business?

No. Trust signals come only from what you list under Voice; the writer is told never to invent them. Read claims about prices and results before you approve.