top of page
Search

How to Make ChatGPT Sound Like You Instead of Everyone Else

After the first wave of AI adoption, something strange happened.

Every LinkedIn post began to sound like it had been written by the same polite consultant who drinks oat milk, uses the word “insightful” too often, and ends every message with three bullet points about leadership.


It was efficient. It was readable. It was completely forgettable.


Teams realised that while ChatGPT could produce endless content, none of it sounded like them. Law firms sounded like marketing agencies. Startups sounded like universities. Government reports sounded like motivational posters.


The problem was not the model. The problem was that nobody taught it their voice.

In Enigmatica workshops, especially when training comms teams or executives, the breakthrough comes when people realise voice is data. If you feed the model strong examples of your writing, it can analyse tone, pacing, vocabulary, humour, and structure. Then you can turn that analysis into a reusable instruction that keeps your style consistent across emails, reports, proposals, and LinkedIn posts.


This matters more than people think. In 2022 and 2023, the internet was flooded with AI content. By 2024, regulators were asking about disclosure. By 2025, audiences were simply bored. They could tell when something sounded generic.


Your voice became your advantage.


The technique is simple.


Take real writing samples that represent your best work. Ask the model to analyse them and describe the style. Then use that description as a prompt in a new conversation when you generate content.


Suddenly, your AI output sounds like your team, not like a template.


This approach is also safer. Instead of inventing personas or copying someone else’s tone, you are formalising your own. It keeps brand identity consistent and avoids the awkward moment when your CEO sounds like a teenager who learned marketing from TikTok.


Voice is not decoration. It is strategy.


And in a world where everyone has access to the same tools, the only thing that remains unique is how you say things.


Practical Tips for Training Your AI Voice

  1. Use Real Writing Samples Choose emails, posts, or reports that represent your best work.

  2. Provide Enough Text One paragraph is not enough. Give multiple examples to capture patterns.

  3. Separate Voice From Content Analyse tone and structure first. Generate new content in a fresh thread.

  4. Document Your Voice Instructions Store them in your Copilot or Enigmatica prompt library.

  5. Update Periodically As your brand evolves, refresh the voice samples.

  6. Test Across Different Tasks Try the voice prompt on emails, proposals, and reports.

  7. Keep It Honest Do not try to mimic someone else’s personality. Consistency builds trust.


Prompts

# VOICE ANALYSIS PROMPT

## ROLE
You are a writing style analyst.

## INPUT
Text samples from the same author:
**[paste 3 to 5 writing samples]**

## OUTPUT
1. Tone description
2. Vocabulary patterns
3. Sentence structure tendencies
4. Use of humour or emphasis
5. Formality level
6. Audience assumptions
7. A short reusable "Voice Instruction" paragraph that another AI can follow

# VOICE GENERATION PROMPT

## ROLE
You are writing new content using a defined voice.

## INPUT
- Topic: **[subject]**
- Audience: **[who it is for]**
- Goal: **[inform, persuade, entertain]**
- Voice Instruction: **[paste paragraph from Voice Analysis]**

## OUTPUT
Write content that follows the voice instruction while staying accurate and useful.
Match tone and pacing without copying any original sentences.

# VOICE CONSISTENCY CHECK

## ROLE
You are a brand editor checking tone consistency.

## INPUT
- Draft text
- Voice Instruction

## OUTPUT
1. Does the draft match the voice
2. Where it drifts
3. Suggested edits
4. Final revised version

 
 
 

Comments

bottom of page