Before you start
- Any member of your workspace can create a voice. There are no role restrictions.
- Prepare at least one example of your brand writing. The more representative your examples, the more accurate the extracted voice.
How to add an existing voice
- In your Brivvy workspace, navigate to the Voices tab.
- Select Upload existing. If you already have voices in your workspace, open the dropdown on the New brand voice button and select Upload existing.
- In the Type of voice dropdown, choose a category that fits your use case. This is optional, but it helps Brivvy name and contextualize the extracted voice.
- Under Examples, add your brand materials using one or more of the three source types: URL, Text or File. You can mix source types in a single voice.
- Select Continue.
- Wait for extraction to complete. This typically takes a couple of minutes. Brivvy analyzes your examples and maps patterns to tone dimensions and writing rules.
- Review the extracted voice configuration. Check that the tone sliders and rules reflect how your brand actually communicates. Adjust anything that does not look right.
- Select Publish to make the voice available across your workspace.
Source types
You can provide examples through three input methods. Each one feeds the same extraction engine, so choose whichever format is most convenient for the content you have.URL
Paste a link to a published page that represents your brand voice. Brivvy reads the content of that single page and uses it as an extraction input.URLs are scoped to a single page. Brivvy does not crawl an entire domain. This is intentional. The goal is quality over quantity, so choose pages that best represent how your brand communicates.
Text
Paste raw text directly into the wizard. This works well for content that is not published online, such as internal messaging guidelines, draft copy or email templates. Each text sample must be at least 100 characters.File
Upload a document that contains examples of your brand writing. This is useful for style guides, brand books or long-form content stored locally.Tips for better results
- Choose standout examples. The wizard works best when you provide content that clearly represents your intended voice. A polished blog post or a well-written landing page is more useful than a generic status update.
- Mix your sources. Combining a URL, a text snippet and a file gives Brivvy a broader picture of how your brand communicates across formats.
- Provide two to five examples. A single example can work, but multiple inputs help Brivvy identify consistent patterns rather than one-off stylistic choices.
- Review before publishing. Extraction is a starting point. Always check the generated tone sliders and rules against your expectations before publishing.
How to know it worked
After publishing, open the voice and confirm that the tone dimensions and rules match your brand. You can also use Preview brand voice to see how the configuration shapes AI output in practice.What happens if extraction fails
In rare cases, extraction may not complete successfully. If this happens, Brivvy notifies you and you can try again. Common fixes include providing longer or more varied examples and making sure URLs point to pages with substantial text content.What’s next
- Preview brand voice to test the extracted configuration before using it in production.
- Create a voice from scratch if you prefer to set tone and rules manually.
- Connect an AI tool to start generating content with your new voice.