How to Make an AI Influencer (Classic or Quirky)

Start with a character, not a face

Most AI influencers fail for the same reason: they look like every other AI influencer. Symmetrical features, flawless skin and a vague lifestyle give viewers nothing to remember. Before you open the studio, write one sentence about who this person is and why anyone would follow them. For example: a retired ferry mechanic who reviews cheap tools, or a botany student who rates houseplants like restaurant critics rate food.

That sentence drives every later decision: the face, the voice, the content pillars and the hashtags. Biggsfluencer, found at /studio/influencers, is built around this flow. You design the look, Biggsfield generates reference angles so the face stays the same, and you attach a persona and voice that carry into every photo, video and talking clip.

Classic or quirky: choosing the look

Biggsfluencer lets you design a creator from scratch, from polished to proudly quirky. You choose hair, teeth, nose and outfit directly, or hit Surprise me if you want a starting point to react against. Mullets, gap teeth and unibrows are explicitly welcome, which matters more than it sounds.

A classic look suits niches where trust and aspiration lead: skincare, fitness, finance explainers, fashion. Keep it believable rather than perfect. Pick one or two small, specific details, such as a faint scar through an eyebrow, slightly uneven front teeth or a crooked nose bridge, so the face reads as a person instead of a template.

A quirky look suits comedy, reviews, gaming, food and anything where personality is the product. Quirky works best when the features match the character's story. A ferry mechanic with a grey mullet and a gap tooth makes sense instantly, and that coherence is what makes a face memorable at thumbnail size.

Whichever direction you choose, decide on a signature outfit element: a specific jacket, a pair of glasses, a colored beanie. Repeated wardrobe is how real creators become recognizable, and it gives your character a visual anchor beyond the face.

The automatic reference pack

Once you pick a face, Biggsfield shoots five reference angles: both sides, a profile, a laughing close-up and a full body shot. This pack is attached to everything you make with the character, so the same person shows up whether you generate a still, a video or a talking clip.

Review the pack carefully before moving on. If the laughing close-up shows teeth that don't match your choices, or the full body proportions feel off, fix it now. Every later generation inherits from these references, so small errors compound.

For maximum consistency, Biggsfluencer can train a private LoRA in one click. Do this once you are happy with the look and plan to post regularly. A private model is the difference between a character who looks similar from post to post and one who looks like the same person.

Persona: bio, tone, pillars and hashtags

Biggsfluencer stores a persona alongside the face: a bio, a tone, content pillars and hashtags. Treat this like a show bible.

Write the bio in the character's own voice, in two or three lines. Define tone with concrete contrasts rather than adjectives alone, such as dry but never mean, or enthusiastic about tools and suspicious of gadgets. Pick three to five content pillars, recurring formats your audience can expect. Keep hashtags specific to the niche; broad tags rarely help a new account.

Bio: Spent 31 years fixing ferry engines. Now I test the cheapest tools online so you don't waste your money. Opinions are free, refunds are not.

Pillars: Under-20 tool tests; Things that broke on day one; Workshop myths debunked; Viewer question Fridays.

Giving your character a voice in 29 languages

Each character can carry an ElevenLabs voice in 29 languages. Choose a voice that fits the face and the story, and keep it fixed. Changing voices confuses an audience faster than changing hairstyles.

For scripted voiceovers, ElevenLabs v4 supports inline English audio tags such as [whispering], [laughs], [excited] and [sighs] to direct delivery. It costs 17 credits per 1,000 characters (about $0.16 on Creator). Write scripts the way people talk: short sentences, contractions, a pause before the punchline.

Because the voice works across 29 languages, you can test new markets without rebuilding the character. The Translate and Lip-sync app dubs a clip into another language in the character's voice, with lips matched to the new words.

[sighs] Okay. Nineteen ninety-nine for a full socket set. [laughs] Let's see how long this lasts.

Making photos and videos with the character

Stills are the cheapest way to build a feed. Nano Banana 2.1 is strong for realism and character consistency and reads natural sentences, not keyword lists; a 1K image costs 8 credits and a 4K image 30. Krea v3, at 9 credits per image, is tuned for fashion, lifestyle and editorial photorealism. For carousels and lookbooks, the Consistent Image Set app produces 4 to 8 frames with the same people, outfits and style from different angles and moments.

For video, Seedance 2.5 generates native audio with lip-synced speech. Put spoken lines in double quotes with who says them and how, and refer to reference images as @Image1. A 5-second 480p clip with sound costs 184 credits. Kling O3 Pro keeps characters and products consistent across up to four references, from 117 credits for 5 seconds with sound; write any dialogue in English.

For straight-to-camera talking content, OmniHuman 2 turns one photo plus an audio track into a realistic talking-human video of up to 5 minutes, at 400 credits per 5 seconds. Pair it with an ElevenLabs voiceover for scripted monologues.

@Image1 stands at a cluttered workbench under a single hanging bulb, holding up a cheap socket wrench and squinting at it. Medium close-up, slow push-in. He says, flatly: "Nineteen ninety-nine. Let's find out why." Ambient workshop hum, a radio playing faintly.

A warm, natural photo of the woman from image 1 repotting a fern on a sunny apartment balcony, soil on her hands, laughing at something off camera. Shot on a 35mm lens, soft afternoon light, real skin texture.

Content ideas that suit AI characters

Formats with a fixed structure work well because the character becomes the constant. Try rating series (every post scores something out of ten), reaction formats, myth versus fact explainers, day-in-the-life photo carousels and recurring character bits with a catchphrase.

Quirky characters can lean into Biggsfield's viral effects apps, such as Levitate, Turn to Stone or Behind the Scenes, for occasional pattern-breaking posts. Use them sparingly; the persona should carry the account, not the effects.

If you want a YouTube channel that runs on a schedule, Biggs Autopilot can use your AI influencer as the host. It finds a topic, writes the script, records the voice, generates scenes, edits and uploads. Videos wait for your approval and upload as private by default.

Disclose that the character is AI

Be upfront. Say in the bio that the character is AI-generated, use each platform's AI labeling tools and check its current rules before you post. Biggs Autopilot sets AI disclosure the way YouTube requires, but on other platforms labeling is your responsibility.

Disclosure protects the audience and the account. Followers who discover an undisclosed AI character tend to feel misled, while an openly synthetic character can build a loyal following on personality alone. Never present the character as a real person giving genuine testimonials, and only clone a real voice with that person's consent.

Questions

More guides