GUIDE
Make AI write social posts in your brand voice
Last updated
Short answer
AI writes in your brand voice when it gets a short, concrete voice profile every time: a one-paragraph summary, do and don't rules, banned words, and three to five real example posts. Store it where every agent reads it before drafting (in Mynahcast, the brand's versioned voice profile), and update it whenever you send a post back.
What goes in a voice profile
- Summary: who you're talking to and how you sound, in one paragraph.
- Do rules: for example 'say what changed first', 'use numbers', 'one emoji at most'.
- Don't rules: for example 'no exclamation marks', 'never promise dates'.
- Banned words: the corporate filler you never use.
- Examples: three to five real posts you're proud of, per channel if they differ.
Why examples beat adjectives
'Friendly but professional' describes most brands. Models imitate examples far better than they follow adjectives, so a few real posts do more than a paragraph of tone words. Keep the examples current; the agent will copy their length and structure.
How Mynahcast uses it
Each brand has a voice profile that agents read with mynahcast_get_brand_voice (or GET /v1/brands/{brandId}/voice) before drafting. Profiles are versioned: each update publishes a new version and the old ones are kept. Agent keys can read the voice; updating it needs the voice:write scope, which only the Publisher preset has.
Frequently asked questions
- How do I get ChatGPT to write in my brand voice?
- Give it a concrete voice profile (summary, do and don't rules, banned words, real example posts) in every request, or connect it to a tool that supplies the profile automatically, such as Mynahcast's GPT action, which can read the brand voice before drafting.
- Does a brand voice profile work across different AI models?
- Yes. The profile is plain text, so ChatGPT, Claude, Gemini or any other agent can read the same one, which keeps posts consistent when you switch models.