A voice profile tells the AI how you write: your tone, your audience, what you avoid. It's what makes generated content sound like a person, not a machine.
When you use ChatGPT to write a LinkedIn post, the result sounds like ChatGPT. Not like you.
The problem isn't the quality of the model. It's that it knows nothing about you: how you write, who you're addressing, what topics you avoid, what register you use. Without that information, the model generates the statistical average of all the LinkedIn posts it has seen. Which is exactly what it looks like.
A voice profile solves this problem. It's the layer of information that tells the AI who you are as an author.
In Pensend, the voice profile (called Mi Voz) is built with six fields:
How you'd describe your style. It doesn't have to be formal: it can be "direct and no-nonsense," "analytical but accessible," "conversational like talking to a friend in the industry," "technical for people with context."
The more specific, the better. "Professional" isn't enough information. "Professional but no corporate jargon, with concrete examples and without moralizing" is.
Who you write for. Not "industry professionals," but something more concrete: "strategy consultants with 5-15 years of experience," "CEOs of tech SMEs without a marketing team," "content creators starting on LinkedIn coming from an offline background."
The AI uses this field to calibrate the level of assumed prior knowledge and the type of examples it uses.
What you do, your value proposition, what methodologies you use, what type of clients you work with. This field is especially useful if you have your own approach or specific terminology from your practice.
Example: "Organizational transformation consultant. I work primarily with industrial companies in cultural change processes. I use the Kotter framework and agile methodologies adapted to non-tech environments."
The AI can incorporate references to your context without you having to repeat it in every generation.
Things you don't want appearing in your content. This can be thematic ("I don't want to talk about politics even if it relates to my sector"), tonal ("never phrases like 'in today's fast-paced world'"), or positioning-related ("don't mention direct competitors").
This field acts as a negative filter: the AI knows what not to do.
Two or three real texts of yours: LinkedIn posts, blog articles, client emails. These are the most direct reference for how you write in practice.
The AI analyzes them to extract patterns: sentence length, paragraph structure, habitual vocabulary, real vs. stated formality level, use of questions, punctuation rhythm.
Once you configure the five fields above, Pensend automatically generates a summary of your voice profile. It's a synthesis of 150 to 300 words that describes how you write, designed to be used as an instruction by the AI agents.
You can edit it if something doesn't accurately reflect your style. This is the document that's prepended to the system prompt of every generation.
Without an active voice profile, the Deep Writer generates a competent but generic article. With an active profile, the same prompt produces a result that uses your vocabulary, respects your restrictions, and speaks to your audience.
The difference is more noticeable the more specific your profile is. A voice profile with tone "professional," audience "professionals," and no writing samples doesn't contribute much. A profile with detailed tone, concrete audience, specific business context, and two or three real samples substantially changes the output.
Between 10 and 20 minutes the first time. The free-text fields have no minimum or maximum length: you can write three sentences or three paragraphs. The quality of the result scales with the quality of the information you provide.
Once configured, the profile is automatically applied to all Deep Writer generations. You don't have to repeat it or remember it: it's just there.
A voice profile has the greatest impact the more differentiated your voice is as an author. If you've been publishing content for a while and have a defined style, the profile allows the AI to replicate it more faithfully. If you're just starting out, the process of setting it up also helps you define how you want to write.
It's available on the Profesional and Creador plans. The Creador plan allows up to three different voice profiles, useful if you write for very different audiences or if you manage content for more than one author.
You can set up your voice profile from the Mi Voz section of the dashboard. If you don't have an account yet, the Explorador plan is free and you can start without a credit card.
Continue reading
Create LinkedIn content in your voice with AI
Trend Scanner, Deep Writer, Slide Writer and Broadcast — four agents, one workflow.
Start for free