AI Can Generate the Video. It Can't Decide What People Should Feel.

Inside Future Caribbean's Cinematic AI Video & Visual Storytelling Workshop with Dallas and Sharon of Mercury Tech.

AI has made video production dramatically more accessible. A founder can generate a scene, create a voiceover, produce music and assemble an advertisement without hiring a traditional production team.

But there is a catch.

Generating a video is not the same as telling a story.

That was the central message of Future Caribbean's Cinematic AI Video & Visual Storytelling Workshop, hosted by Lily Dash and led by entrepreneurs and builders Dallas and Sharon of Mercury Tech.

With 19 people attending, the workshop went far beyond a tour of AI video tools. Instead, it focused on the creative thinking, psychology, prompting and production discipline required to turn an idea into something an audience actually cares about.

The AI isn't the creative director

One of the first lessons from the workshop was also one of the most important: don't hand the creative thinking over to AI.

Dallas and Sharon demonstrated how easy it is to ask an AI system to "make a video" and receive something technically impressive but creatively meaningless.

Their approach begins before opening an AI video generator.

First, determine what the audience is supposed to understand, feel or want. Then develop the creative direction, story, characters, environments and individual scenes.

Only after those decisions have been made does AI enter the production process.

As Sharon explained during the session, one of the biggest mistakes in AI video production is leaving the creativity entirely to the model.

The distinction is subtle but fundamental:

AI can execute an instruction. A human still needs to decide what the instruction should accomplish.

From product features to human emotion

The workshop demonstrated why traditional advertising psychology remains relevant even as the production technology changes.

Dallas and Sharon contrasted straightforward informational advertising with storytelling.

A conventional product video might explain what a product does.

A story asks something different: what does this product mean to the person using it?

They used familiar advertising examples to illustrate the point. A mortgage advertisement rarely focuses on the mechanics of financing. Instead, it might show parents, children, a home and a car — visual cues that communicate security, family and peace of mind.

The product is the mortgage.

The story is the life the customer imagines having because of it.

The water bottle that was never really about water

The team's demonstration brought the concept to life.

They showcased a cinematic advertisement built around a water bottle and a character whose life unfolds through a series of difficult moments.

The bottle is present throughout the story, but the advertisement isn't simply trying to convince someone that the bottle is a good product.

Instead, the story uses the bottle as a recurring symbol — something that remains present as the character experiences failure, hardship and eventually growth.

The goal is emotional association.

The audience isn't being told: "This is a high-quality water bottle."

They're being invited to feel something about what the bottle represents.

That is what transforms a generated sequence of clips into a piece of storytelling.

AI doesn't have lived experience

One of the workshop's most fascinating discussions centred on how AI interprets human emotion.

Humans understand subtle signals almost automatically. A person's posture, facial expression, tone of voice, movements and surroundings can collectively tell us that someone is nervous, excited, disappointed or relieved.

AI doesn't experience those emotions.

Dallas and Sharon described this as the difference between "lived experience" and "observed experience."

If an AI is simply told that a character is excited, it may produce an exaggerated smile, dramatically widened eyes or other obvious signals because those are easily recognisable representations of excitement.

But human emotion is often much more subtle.

The creative director therefore has to communicate not just what a character feels, but what is happening around them and how that emotion should manifest physically.

The seven-tool problem

Before building their own platform, Dallas and Sharon described a workflow that required moving between approximately seven different tools.

They used one system for prompting, another for image references, another for video generation, another for voiceover, another for music, another for editing and their computer for managing the resulting files.

That process wasn't simply inconvenient. It created continuity problems.

A voiceover could describe one object before another even though the visual sequence showed the opposite. A generated clip could require another generation. References could become inconsistent. Files could end up scattered across different platforms.

Mercury Tech eventually built its own platform, Studio, to bring more of that workflow together and simplify the production process. The platform connects different AI models and allows the team to generate images and videos without constantly moving between separate applications.

The bigger lesson is that AI production is becoming a workflow problem, not simply a generation problem.

Why more expensive doesn't always mean better

Not every video requires the most powerful model available.

For simpler formats such as talking-head videos, Dallas and Sharon explained that less expensive models can be perfectly suitable. More sophisticated models become valuable when a production requires complex storylines, multiple locations, consistent characters, voices, music and sound effects.

But the real leverage isn't necessarily finding the cheapest possible generation. It's what AI makes financially possible.

Productions that might previously have required enormous budgets and large crews can now be created for a fraction of the traditional cost.

That changes who gets to make high-quality visual content.

A startup with limited capital can experiment with cinematic advertising. A solo founder can test creative concepts. A small team can produce multiple variations of a campaign. And builders who previously couldn't afford professional production can begin competing on the quality of their storytelling.

Stop romanticizing your product

Perhaps the strongest message of the entire workshop came toward the end.

"Stop romanticizing your product."

Founders naturally become attached to what they build. After spending months developing a feature, architecture or technical innovation, it is easy to assume that explaining how impressive it is will make people care.

But customers aren't necessarily interested in the complexity behind the product. They care about what it does for them.

Dallas and Sharon encouraged builders to put themselves in the audience's position and ask:

If I'm seeing this for the first time, what do I want? How do I want to feel? What problem am I trying to solve?

The strongest marketing isn't necessarily about explaining every feature. It's about helping someone recognise themselves in the problem — and see the product as the path toward a better outcome.

Don't wait for the perfect campaign

The workshop also challenged another common founder instinct: trying to get the marketing perfect before putting it in front of anyone.

Dallas and Sharon advocated experimentation instead. Create something. Put it out. See what happens. Then learn from the response.

They described testing multiple advertisements rather than assuming the team knows which creative will perform best. In their experience, the advertisement they expect to win isn't always the one audiences actually respond to.

That makes AI particularly powerful. If production becomes faster and cheaper, experimentation becomes more accessible.

Instead of betting everything on one "perfect" video, founders can test different stories, hooks, emotional angles and audiences.

The goal isn't to predict what people will love. It's to create enough opportunities to discover what they actually love.

Your story doesn't have to be dramatic

The final discussion returned to something many founders struggle with: how to tell their story authentically without manufacturing a dramatic origin story.

People don't necessarily need your story to be extraordinary. They need to be able to see themselves in it.

A founder can start with the genuine problem they experienced and explain how they built something to solve it. The story becomes compelling when the audience recognises their own frustration, ambition or desired outcome within it.

That is ultimately what separates visual content from visual storytelling.

AI can generate the characters. It can generate the locations. It can generate the camera movements. It can generate the voice. It can even generate the music.

But the meaning still has to come from somewhere.