Agents

Writing Content That Doesn't Sound Like AI: A Brand-Voice Primer

[ 8 min read ] · May 15, 2026 · Veqiro

How to define your brand voice so precisely that AI-generated content is indistinguishable from your best human writing — plus the exact framework we use.

There's a specific kind of content horror that founders who use AI tools know intimately. You ask for a blog post. You get back something that is technically correct, occasionally even informative, and completely indistinguishable from every other piece of content on the internet.

"Unlock your potential." "In today's rapidly evolving landscape." "Leveraging cutting-edge AI solutions."

This isn't a failure of the AI. It's a failure of the brief. Generic inputs produce generic outputs, and most founders give their AI tools approximately zero context about their actual voice.

Here's how to fix that — and how to get content out of Maya that your audience will actually recognise as coming from you.

What Brand Voice Actually Is (and Isn't)

Brand voice is not "professional tone." It's not "friendly." It's not "clear and concise."

Those are adjectives that apply to basically every brand that's ever tried to describe itself. They're not distinctive. They don't help an AI write like you.

Real brand voice has three elements:

1. A specific perspective. A point of view that's actually yours. Veqiro's perspective: "lean teams deserve the same leverage as big ones, and AI employees are how you get it." That's a real position, not a platitude.

2. A recognisable sentence style. Some brands write in long, flowing sentences that build to a conclusion. Others punch in short bursts. Some use em-dashes like punctuation. Some start sentences with "And." Know your rhythm.

3. A vocabulary set. Words you use (and words you'd never use). Veqiro uses "crew," "leverage," "ship," "lean." We don't use "solutions," "ecosystem," or "next-generation."

Without all three, you're not describing a voice. You're describing vague intentions.

The Brand Voice Framework

Here's the framework for building a voice document that actually works with AI tools:

Component 1: The Three-Word Brand Personality

Describe your brand with three specific adjectives. Not "professional" or "approachable" — those mean nothing. Pick adjectives with contrast:

  • Direct, witty, irreverent (Veqiro)
  • Earnest, rigorous, dry (Stripe's early writing)
  • Warm, practical, no-nonsense (37signals)

Each adjective should include one or two example sentences that demonstrate what it looks like in practice. "Direct: 'We charge $39/month. That's it.' Not: 'We offer flexible, transparent pricing designed to meet your needs.'"

Component 2: The Vocabulary Map

Create two lists:

Words and phrases we use: List 15–20 terms that show up in your best content. Include functional terms (how you describe your product) and personality terms (your shorthand, your jokes, your references).

Words and phrases we never use: List 10–15 terms that immediately read as "not us." Often these are corporate buzzwords, overused industry terms, or words associated with competitors' positioning.

This list is underrated. Telling an AI what not to write is sometimes more useful than telling it what to write.

Component 3: Sentence Rhythm Calibration

Show, don't just tell. Include 2–3 examples of your preferred rhythm:

Preferred (punchy, parallel):
"Maya writes the post. Scout found the angle. Sage checks the keywords. 
You approve or edit. The content ships."

Not preferred (flowing and over-explained):
"By leveraging Maya's advanced content generation capabilities alongside 
Scout's research synthesis and Sage's SEO analysis, founders can create 
content that is both engaging and optimised for search."

Component 4: Topic Ownership Map

What topics do you write about with authority? What topics do you avoid?

This shapes what kind of content you publish, which in turn shapes what your audience expects from you. Veqiro writes about founder productivity, AI strategy, startup operations. We don't write about enterprise IT procurement or Python development.

Component 5: Platform Calibration

Your voice should adjust by platform while remaining recognisably you:

| Platform | Calibration | |---------|------------| | LinkedIn | More formal, longer, thought-leadership framing | | Twitter/X | Short, direct, often a single provocative observation | | Blog | Room to develop an argument; more nuance, more structure | | Email newsletter | Conversational, like a letter to a smart friend | | Instagram | Visual-first; caption is secondary to the image concept |

Give Maya specific calibration notes for each platform you publish on.

Loading Your Voice Into Maya

Once your voice document exists, loading it into Maya is a one-time setup:

  1. Upload the full voice document to Maya's Brand Brain
  2. Add 10–15 example posts across your main channels (pick your absolute best work)
  3. Run a calibration batch — ask Maya to write 3 posts on topics you know well, then mark each line with "on voice / off voice" annotations
  4. Update the voice document based on patterns you notice in the corrections

After this initial pass, Maya's Brand Brain has a rich model of your voice. Most founders see consistent on-voice output within the first 2 weeks.

The Editorial Review Process

AI content isn't finished content. It's a fast first draft. Your editorial process determines the quality gap between what comes out of Maya and what goes out to your audience.

For blog posts: Read through once for structure and argument (is the logic right?), once for voice (does this sound like me?), once for facts (is everything accurate?). Three reads, 15–20 minutes per post.

For social posts: One read, focus on the opening line. If the hook is right, the rest is usually fixable in 60 seconds. If the hook is wrong, rewrite it.

For email newsletters: Read aloud. Anything that feels wrong to say aloud will feel wrong to read. The ear catches what the eye misses.

Before and After: What the Difference Looks Like

Without brand voice context, Maya might write:

"In today's competitive business landscape, content marketing has emerged as a critical strategy for startups looking to establish thought leadership and drive organic growth through SEO-optimised content."

With full brand voice context (Veqiro style), Maya writes:

"Your content is either building an audience or wasting everyone's time. There's not much middle ground. Here's how to make it do the first thing."

Same AI. Different brief. Completely different result.

The difference isn't a better model. It's a better brief.


Brand voice is the most leveraged investment you'll make in your content operation. Write it once. Load it once. Then let Maya run — at scale, consistently, in your voice.

See how Maya handles your content →

questions people keep asking.

Why does AI-generated content often sound generic?

Because generic inputs produce generic outputs. If you give an AI agent no context about your voice, it defaults to the average of everything it's been trained on — which is competent but forgettable. Brand voice definition is the difference between 'AI content' and 'your content, produced by AI.'

What should a brand voice document include?

At minimum: 3–5 tone adjectives with examples, vocabulary you use and avoid, sentence rhythm preferences, topics you own and avoid, and 3 examples of writing you consider 'on voice.' Longer is not better — clarity is better.

How long does it take for an AI to learn a brand voice?

With a well-structured brand voice document and 10–15 example outputs, a capable AI agent produces consistently on-voice content within 2–3 weeks of iteration and feedback cycles.

Can AI write genuinely creative content, or just competent content?

Competent content at scale is the reliable default. Genuinely creative content — with unexpected turns, earned metaphors, and ideas readers didn't know they were waiting for — still requires human creative direction. Use AI for execution volume; keep a human in the creative direction seat.

How do I brief an AI content agent for social media posts?

Include: the platform (LinkedIn vs. Twitter vs. Instagram have very different norms), the content goal (educate, promote, engage), the topic with 2–3 specific angles to explore, the tone calibration (more formal or more casual than your baseline), and any examples of posts you loved. The brief quality determines the output quality.

your crew is waiting.

Six AI employees. One subscription. $39/mo.

Start free — 7 days on us →

more from the crew

Founders

How a 3-Person Startup Ships Like a 10-Person Team

The exact operating model that lets lean startups punch above their weight — using AI employees to cover execution while humans focus on what actually moves the company.

[ 8 min ] · Jun 9, 2026
Use Cases

The Best AI Tools for Agencies in 2026

A curated breakdown of the AI tools that actually move the needle for agencies — covering content, research, SEO, client reporting, and operations.

[ 9 min ] · Jun 5, 2026
Comparisons

Lindy vs Sintra vs Veqiro: Which AI Employee Platform Is Right for You?

A direct comparison of three leading AI employee platforms — Lindy, Sintra, and Veqiro. Features, pricing, ideal user, and what each one actually gets right.

[ 11 min ] · Jun 2, 2026