AI avatar videos for the weeks you can't film

An AI avatar is a digital version of a presenter that reads new scripts without you filming again ^[1]. You type what it should say, the tool renders a video of a face saying it, and you post that video like any other.

There are four kinds. They differ in what you hand the tool at the start.

For a coach or a consultant, the digital twin is usually the one worth setting up, because your face and the way you talk are part of what clients buy. HeyGen's newest twin model learns your movement from a 15-second video ^[6]. For a twin, both HeyGen and Synthesia make you record a consent clip on camera first ^[5][2], so a twin is only made with the person's knowledge ^[5].

Making the avatar takes a morning, once. These are the steps for a digital twin in HeyGen, the route that keeps your own face and movement; a photo avatar skips the recording ^[3].

1. Open Avatars, choose New Avatar, then Clone a Real Person, and record live or import footage ^[4].
2. Record one uninterrupted take of at least 2 minutes, talking about your work the way you would to a new client ^[7].
3. Record the consent clip: read HeyGen's text exactly as written, in under 30 seconds ^[5].
4. Make a voice clone during setup, which HeyGen strongly recommends ^[6].
5. Type a script of 30 to 60 seconds, under 100 words, render it and check the face, voice and lip sync before you make more ^[8].

The HeyGen avatar guide has the filming rules and the fixes when a test looks off.

You probably know video would bring you clients. Making it is the hard part. One short needs a script, a hook, footage, B-roll, captions, graphics and a version for each platform ^[9]. Then YouTube, Instagram, TikTok and LinkedIn each want their own cut, and you're doing all of it between client sessions, with no editor.

An avatar takes one item off that list: the filming. Here's who does what for one weekly tip video once you have one.

So an avatar won't rescue a missing plan or a weak script. Work those out first, and the weekly run below covers everything after the filming.

An avatar earns its place on the posts you'd make every week if filming took no time. These five fit a business built on your own expertise.

1. Your weekly point of view: the thing you explain to clients on calls becomes a short video every week. You're usually the most credible voice in your business and the busiest person in it ^[1], so this is where an avatar saves the most hours.
2. A recurring update, such as a property agent's local market update, read by the same presenter every week with one clear next step ^[1]. A lawyer could do one rule change a week, and a consultant one trend from their clients' industry.
3. Explainers for what you sell: what happens in a first session, how your fees work, what to bring to a viewing. One short video per question, each pointing to where people book.
4. Answers to the questions people search, with one question per video, made the same way each time so viewers know what they'll get. Anything health, legal or financial still needs your real expertise behind the script ^[1].
5. Onboarding and follow-up: a welcome for each new client, a walkthrough of how you work, a note after a call. Recording these one by one would take you hours ^[1].

Some posts get worse with an avatar. Check yours against this list before you build one.

• You don't have a point of view yet. An avatar makes a generic script faster to produce, and the video is still generic ^[1]. Write down what you believe first; there's a prompt for it below.
• The moment needs trust. A hard conversation with a client, an update in a crisis, a personal thank-you. Show up in person ^[1].
• It's live. Q&As, livestreams and podcasts still need you there ^[1].
• The script makes a claim you wouldn't make on camera yourself. If you'd hesitate to say it face to face, don't have your twin say it.

If your work is regulated, such as law, finance or health, the avatar's scripts need the same review your own posts get, because it's speaking as you.

The avatar is the easy half. AI avatar videos usually fall flat because they sound like everyone else's ^[1], and you fix that before you open a video tool:

1. Write your point of view: what you believe about your work and what you argue against ^[1]. Every script should serve it. The personal brand template walks you through it if you've never written it down.
2. Capture how you talk: your rhythm, the words you reach for, how you open a point and how you land it ^[1]. This keeps the scripts sounding like you and not like a chatbot.

Below are transcripts of me talking about my work: a client call, a voice note and a talk. Read them and write a voice guide I can paste above every script request.

Include:
- How I open a point (quote three real openings)
- How I land a point (quote three real endings)
- Words and phrases I repeat
- Words I never use
- My usual sentence length, and how often I use a very short sentence
- What I believe about my work, in my words, as five lines
- What I argue against, as three lines

Quote me exactly. Don't tidy my grammar and don't add anything I didn't say.

[PASTE TRANSCRIPTS]

Record the transcripts the easy way. Answer a client question into your phone's voice memo app, or export the transcript of a recorded call. Three of them are enough to start.

Once the avatar exists, each week runs the same way. One short needs a script, a hook, footage, B-roll, captions, graphics and a version for each platform ^[9], so it helps to take them in order.

1. Make the avatar once: record one session or build a photo avatar, and write a reference sheet of its face, style, voice and personality. Every later step pulls from that sheet, so the character doesn't drift ^[9]. The HeyGen avatar guide covers the recording.
2. Plan the week, on Monday if that's your quiet morning: pick topics from the questions clients asked you last week, write the hooks and scripts, and list the shots each one needs ^[9]. Score the hooks before you make anything and drop the ones that don't score ^[9]. The hook generator gives you openings to start from, and the video script template gives each script its shape.
3. Make the stills: the pictures and B-roll starting frames, all matched to the reference sheet ^[9].
4. Render each shot with the tool that suits it: talking shots go to an avatar tool, B-roll to a cinematic video model, stylised moments to a third (the table below has the split) ^[9].
5. Edit with a person deciding. AI can handle the heavy lifting on captions, motion and text, and you decide when it's good ^[9]. The next section has the checks.
6. Cut a version per platform, each with its own hook or angle and the same face ^[9].
7. Post on the one platform your clients use most and go deep there before going wide, and keep a note of which people keep engaging ^[1]. Those names are the people to message, or to mention on your next call.
8. Learn from each post: score which hooks, topics and formats worked ^[9], and feed the winners into next week's scripts.

[PASTE YOUR VOICE GUIDE]

Write three scripts for short videos my AI avatar will read this week. I'm a [what you do] and my clients are [who they are].

Topics:
1. [topic]
2. [topic]
3. [topic]

For each script:
- Give me five hook options first, each under 12 words, and mark the one a busy client would stop for
- Then write the script: 30 to 60 seconds spoken, under 100 words
- One point only, said the way my voice guide says I talk
- End with one next step: [book a call / read the guide / reply with a word]
- List the shots: which lines are me talking to camera and which need B-roll, and what the B-roll shows

Don't invent numbers, client stories or results. If a point needs a number, write [NUMBER] and I'll fill it in.

The render is a raw talking clip, and the edit is where it starts to look like your business. A full setup uses After Effects for the motion graphics, captions, text animations and templates for the hook, the middle and the call to action, with an AI assistant that suggests edits, writes captions and on-screen text and keeps the style consistent ^[9]. AI does the heavy lifting. A person decides when it's good ^[9].

You don't need After Effects for a weekly tip video. Any editor that captions and exports will do, as long as you check these on every render:

• Eye line, gestures and how the avatar says names: yours, your business's and your clients' ^[10].
• The hook on screen from the first frame, and captions all the way through for anyone watching on mute.
• B-roll over any line that's abstract, so the picture changes when the idea does.
• The same templates for the hook, the middle and the end card every week ^[9], so clients know it's you before they hear a word.

Then turn the master into each format: 9:16 for Shorts, Reels and TikTok, 16:9 for YouTube, and 1:1 or 9:16 for LinkedIn, each with its own hook or angle and the same face ^[9]. The short-form video editing guide covers the edit step by step.

Start with one tool and one route ^[1]. Most people looking for the best AI avatar maker are really choosing between a twin of themselves and a stock presenter, so settle that first. If clients book you for you, as a coach, consultant or agent, make the twin. If the video explains a process and the face doesn't matter, a stock presenter is fine.

• HeyGen makes photo avatars and digital twins. Its Avatar V model is for real people and starts from a 15-second video of you; Avatar IV animates photos and suits cartoon, 3D and other non-human characters ^[6]. Avatar V videos can run up to 3 minutes ^[6].
• Synthesia has stock avatars you can use straight away, personal avatars made from a photo or short video of you, and studio avatars for teams that want the most realism ^[2].

Avatar IV, if you've seen the name, is HeyGen's engine for photo looks. It reads the tone, rhythm and emotion in a voice and turns them into facial movement ^[11]. It's also the quickest way to make a drawn character talk, as long as its face is close to human proportions ^[3].

Postmosis can make a short video without filming: upload a clear photo of yourself and type up to 30 seconds of script, and your photo speaks it, or pick an AI presenter to read it, up to 2 a day. See videos without filming

Tell people it's an avatar, because your clients will assume the face on your posts is you. Each platform has its own switch for it.

• Instagram: Meta requires a label on photorealistic video or realistic audio made or changed with AI, and a realistic AI voiceover counts ^[12]. Turn on Add AI label when you post ^[12].
• TikTok: label AI-generated content that shows realistic-looking people or scenes ^[13]. TikTok can remove, restrict or label videos that should have had one ^[13].
• YouTube: set AI use to Yes in YouTube Studio when a video makes a real person appear to say something they didn't ^[14]. YouTube says disclosing on its own won't limit who sees the video ^[14].

Say it in the caption or the bio too, wherever it matters to the viewer ^[1]. A client who finds out later that the friendly weekly video wasn't really you will start wondering what else wasn't.

Three things decide it ^[1]:

• You'll treat video as something you keep doing for months, not a one-week test ^[1].
• You're happy to be the face, and you don't want to run the production.
• You'll keep going while month two looks like nothing is happening ^[1].

If the third sounds hard, write the first two months of scripts before you build anything. Then the posting doesn't depend on how you feel in week six.

Watch for two signs that it isn't working. If people watch but nobody books, look at the script's next step before you blame the avatar. If clients comment that it looks fake, re-record the twin with more expression than feels natural ^[15], or move that format back to real footage.

You don't have to choose between the avatar and real footage. Use the avatar for the weekly posts, and film yourself for launches, events and the personal moments ^[1].

Postmosis edits the video you filmed or rendered, in the browser: it cuts the pauses, adds captions in 6 styles and B-roll, and exports 9:16, 4:5, 1:1 and 16:9 in one run. See the video editor

The first month builds a library and every month after reuses it ^[9]. Keep these in one folder:

• Character reference: photos, angles, style, voice and personality ^[9].
• Prompt library: scripts, visual styles, shot templates and calls to action ^[9].
• Templates: caption styles, end cards and any graphics you reuse ^[9].
• Brand assets: logo, fonts, colours, a headshot and your intro ^[9].

With those in place, a normal week is new scripts and a check of each render, and that's about it.