How to build a reply persona from your writing examples in Voxily

To build a reply persona in Voxily, open the Personas tab in the Voxily side panel, paste two to five short posts you wrote yourself into the VOICE CLONING box, separated by ---, and press Clone My Voice. Voxily suggests Tone, Writing style, Bio and Expertise from those examples. Correct anything that isn't true, add your Name and Role, save, then test the persona on a real post.

Disclosure: I build Voxily, the Chrome extension this tutorial covers. It drafts replies inside the reply box on LinkedIn, X and Reddit, and you decide what to post. Voxily is an independent product, not affiliated with or endorsed by LinkedIn, X or Reddit. Labels and limits were checked in the code on October 3, 2026.

Dana Okafor, the persona in this guide, is illustrative, as are her posts, her knowledge entry and every draft quoted.

Which writing examples to use, and how many

The VOICE CLONING section asks you to "Paste 2-5 of your tweets/posts (separated by ---)". The analysis reads at most ten, and only the first 500 characters of each.

Pick replies and short posts you wrote yourself, on a mix of topics. Voxily saves up to two of them, preferring ones between 8 and 60 words. Leave out ghostwritten posts, unedited AI drafts, quotes of other people and private customer details. Small habits count: if one of three or more examples contains an emoji, the profile records occasional emoji; with two examples, one emoji counts as frequent.

Dana runs a small scheduling app for dental clinics. She pastes three of her posts:

Spent the morning at a clinic front desk. The phone rang the whole time and nobody touched the booking screen. --- Practice managers, do you confirm appointments by text, by phone or both? I hear both and never for the same reason. --- Shipped recurring appointments today. Small feature, three weeks, mostly daylight saving time.

How to create the persona in the Personas tab

On LinkedIn, X or Reddit, click the Voxily icon in Chrome's toolbar to open the side panel and choose the Personas tab. With no persona yet, the tab shows five starter personas and an "Or create a custom persona" button. If you already have one, from a starter or the setup screen, press Edit on its card. Free includes one persona, so editing is how you rebuild it.

The form opens in Write mode. Paste your examples into the VOICE CLONING box at the bottom and press Clone My Voice. It fills Tone, Writing style, Bio and Expertise, leaves Name, Role and the reply length number alone, and replaces those four fields each time you run it.

Fill in Name and Role, then press Create Persona, or Update Persona if you were editing. Your first persona becomes the active one.

What to put in Role, Tone and Expertise

Name and Role tell the model who is writing, and drafts are told not to claim a different name, role, employer or background. Use the name you post under and a role that is true today. Dana's Role reads: Founder, scheduling app for small dental clinics.

Tone and Writing style take a short phrase each, such as "direct, curious, plain words". Once Clone My Voice has run, reply drafts take their style mainly from the measured profile and your saved examples, so fix a wrong style by changing the examples and running it again.

Expertise is a comma-separated list, used in a reply only when the post shares words with it. If Clone My Voice suggested Healthcare, Technology, Business, Dana would narrow it to dental clinic operations, appointment scheduling, patient reminders.

Bio goes with every draft. Keep it under 280 characters, since that's all a draft uses. The number box below Expertise caps reply length. It shows 280 by default, and a draft never exceeds the platform's own limit.

What Clone My Voice does with your examples

Clone My Voice sends your examples to an AI model that describes your voice and suggests the four fields. Voxily also counts habits in the text: emoji, hashtags, lowercase sentence starts, exclamation marks, questions, contractions, missing end periods, hedges like "I think" and openers you repeat. A suggested signature phrase is kept only if it appears word for word in your examples.

The saved examples, at most 280 characters each, stay with the persona until you replace them. Each reply draft shows them to the model as samples of your writing, with an instruction not to reuse their topics. Voxily uses your writing examples this way; it doesn't train or fine-tune a model on them. The privacy and platform rules page explains what Voxily processes.

How knowledge entries differ from the persona

The persona describes how you write. Knowledge entries hold what you know. Open the KB tab, press + Add Entry, choose Fact, Opinion, Example or Resource, fill in Title, Content and Tags (comma-separated), then press Create entry.

For each reply, Voxily picks up to three entries that are switched on and share words with the post. Matching tags count most, then title words. Only the first 200 characters of each chosen entry's content are used, so put the point first.

Each entry card has an On/Off button. Entries switched off aren't sent with drafts.

Dana adds an Opinion titled "Phone booking": "Clinics that book by phone aren't behind. Plenty of good front desks skip online booking." Her tags are phone, booking, front desk.

How to tell from a first draft that the persona needs changing

Test on a post you'd answer anyway: click into the reply box, press the Voxily button, pick a mood and read the draft in the reply box before you post.

Dana tries a LinkedIn post from a practice manager: "We moved to online booking last year and no-shows went up. Anyone else seen this?" Suppose the first draft reads: "As a dentist, I've seen online booking transform practices. Patient engagement is key to reducing no-shows."

"As a dentist" is the wrong identity, so Role and Bio need to say plainly what she does. "I've seen online booking transform practices" is a claim she can't back; she deletes it, and her Phone booking entry gives later drafts a view she holds. "Patient engagement is key" sounds nothing like her posts, so she checks the mood first, then her examples.

Other signs: your expertise turning up on unrelated posts (narrow the Expertise list), drafts that run long (lower the number below Expertise) and drafts that echo your examples' subjects (swap in examples on other topics).

Dana's edited reply: "Were the no-shows mostly new patients who booked online? I'd check whether online bookings get the same reminder text as phone ones." The guides on editing AI drafts for X and LinkedIn comments for founders cover the editing step.

What a reply persona can't do

A persona knows only its fields, your examples and the knowledge entries that match the post. It can't check whether a sentence in a draft is true or whether a reply suits a community's rules. It doesn't post, reply or comment for you: it puts the draft in the reply box, and you edit it and press the platform's own button.

What Free includes and what needs Pro

Free includes one persona, five knowledge entries, 10 reply generations a day, eight standard moods and three voice analyses a day. Clone My Voice uses one analysis, and Extract with AI on an upload draws from the same allowance. Entries switched off still count toward the five.

Pro adds unlimited personas and knowledge entries and removes the daily caps. It also adds Custom Mood, Brand Mode, DM drafting and three reply options when enabled in Settings. Prices are in the pricing section.

Questions about Voxily reply personas

How many writing examples should I paste?

Two to five, as the VOICE CLONING section asks. Short replies and posts you wrote yourself work best.

Can I use different personas for LinkedIn and X?

On Pro, yes. Create one persona per voice and tick its sites under DEFAULT PLATFORMS. A linked persona is used on that site even when another is set active. On Free, your one persona drafts on all three sites.

Does Voxily train a model on my posts?

No. It uses your writing examples to describe your style and sends up to two of them with each draft request. You can edit or delete the persona at any time.

The AI reply Chrome extension page shows the full drafting flow on LinkedIn, X and Reddit.