Look, I publish AI-avatar videos on my page.
To be honest, it’s because I had ACL surgery and physically couldn’t set up my camera, lights, and microphone to record videos.
Luckily, I had filmed AI training footage before my surgery, which allowed me to create an avatar and produce videos while recovering on my couch.
During that time, I experimented with AI-generated content and the various tools released this year. I started creating commercials, motion graphics, and production workflows with just a few prompts.
It’s astounding how powerful these tools are, and I believe agencies that ignore them will eventually fall behind.
But I would never use an AI avatar as a fake customer testimonial.
That distinction matters.
The question for agencies is no longer whether to use AI. The question is how you will use it and what promise the content is making to the audience.
If the promise is “this really happened,” we need proof.
If the promise is “imagine what’s possible,” AI can expand the canvas.
If the promise is “here’s something useful,” an avatar may be appropriate.
My working framework is simple:
Real footage for proof. AI for imagination. Avatars for information.
Real footage for proof
Some content asks the audience to believe something about a person, product, or company.
That includes:
Customer testimonials
Founder stories
Case studies
Product demonstrations
Employee experiences
Behind-the-scenes footage
Documentary-style brand stories
When credibility is the point, use real people and real situations.
A customer testimonial matters because a customer actually said it.
A product demonstration works because the audience can see the product performing and needs to believe that what arrives in the mail is what they saw in the video.
Behind-the-scenes footage builds confidence because it shows what happened outside the polished final edit.
Using AI to manufacture these moments may save money, but it also changes the nature of the claim. At that point, it is effectively lying.
The audience is no longer watching evidence. It is watching a simulation of evidence.
That is a dangerous trade for an agency to make on a client’s behalf.
AI for imagination
AI becomes much more interesting when nobody is being asked to mistake it for reality, but is instead taken on a visual journey that requires some suspension of disbelief.
This is where I think agencies should experiment.
Use AI to:
Develop campaign concepts and previsualizations
Build motion graphics and visual effects
Create impossible environments
Explore surreal and fantastical product films
Animate scenes in 3D, stop motion, anime, and other styles
Produce funny visual gags that are trending on social media
Test creative directions before a full production
Participate in AI formats where the AI slop is part of the joke
Make the bottle walk through a dreamscape.
Animate your team in stop-motion to tell your story.
Put the product in a world that would cost more than the entire campaign budget to construct.
Turn an ordinary creative brief into something the client can watch before locations, crews, and talent are booked.
In these situations, AI is not manufacturing trust. It is making an idea visible.
That can help an agency sell a concept, test a direction, or give a smaller client access to an ambitious visual style that would have cost a fortune a year ago.
The standard should still be high. AI does not excuse weak creative direction, inconsistent products, or bad storytelling.
It simply gives the creative team a larger canvas.
Avatars for information
AI avatars occupy a useful middle ground that I feel has been evolving over the past year.
They can help agencies produce:
Internal training
Product explainers, without creating fake AI product features
Software walkthroughs
Frequently asked questions
Educational videos that would otherwise remain unwritten
Online courses and digital products
Industry news recaps
An avatar can be effective when it is clearly a presentation tool.
The problem begins when the avatar is presented as a real customer, employee, expert, or founder who never gave the message being delivered.
That crosses the line from simplifying production to manufacturing credibility.
And I get it. You can point at my content and ask why I choose to use an avatar.
I use avatars because they allow me to share information and experiment with new formats. I don't need the audience to believe that a fake presenter is a real customer with a real experience.
The use case is transparent. The value comes from the information.
Now, if I create a company overview video with my avatar interacting with fake clients without explicitly telling you it’s a joke, you have permission to call me out, if you can find the real me in person :)
Why agencies need a policy before the client asks
Consumers are not rejecting every use of AI, but the research shows that trust depends heavily on context.
A May 2026 YouGov study found that 45% of Australians would trust a brand less after learning its advertising was created mainly with AI. The same study found that 86% believed brands should clearly disclose advertising created primarily with AI.
A global Canva and Harris Poll study found that 78% of consumers would rather see ads made by people, while 87% believed the best advertising still requires a human touch.
At the same time, 68% said they did not mind AI in advertising when it made the content more helpful or relevant.
That is the nuance agencies need to understand.
Consumers are not necessarily asking us to stop using AI.
They are asking us not to use it carelessly.
With growing concerns about misinformation, data-center energy use, and job displacement, AI has become a sensitive subject. Use it wisely by understanding your clients, their customers, and the audiences they want to reach.
A three-question test for client work
Before using AI in a client-facing asset, ask:
Is this content presenting evidence? If yes, use real footage wherever possible.
Could a reasonable viewer mistake this for a real person or event? If yes, disclose the use of AI or change the execution.
Does AI make the idea better, or does it only make production cheaper? If cost is the only benefit, make sure the savings are not coming at the expense of credibility or quality.
AI belongs in agency work.
But it should expand what the creative team can imagine, not manufacture trust the client has not earned.
Real footage for proof. AI for imagination. Avatars for information.
And if you’re still hesitant about using AI in your content, remember that there are still humans willing to capture real footage for you, supported by a worldwide talent pool that can add VFX or animation to increase the production value.
Sometimes, you can save more by hiring a creative professional instead of burning through Higgsfield credits only to end up with something that still looks like slop.
Where is your agency drawing that line?
