The impressive clip is not the finished film

Generative video can now produce images that would once have required complicated location work, specialist effects or substantial budgets. A single convincing shot, however, is not a campaign. A useful film needs a coherent idea, an audience, an editorial structure, credible details, sound, pace and a reason to exist. The ability to generate ten striking clips in an afternoon can actually make those decisions harder: there is more material, but not necessarily more meaning.

The new production question

Instead of asking whether AI can make the video, ask which parts of the story require authenticity and which benefit from imaginative freedom. A customer testimonial requires the real customer and their real words. A product demonstration needs accurate behaviour. An abstract brand sequence may be an ideal opportunity for generated visuals. This is not an ideological distinction; it is a question of evidence and trust.

Creative direction still matters

A director decides what the audience should feel, notice and remember. A producer balances that ambition against time, budget and practical risk. An editor creates meaning through selection, rhythm and juxtaposition. AI can assist all three, but it does not automatically understand the commercial context, brand history or legal implications of a shot. Without a clear brief, more generation simply produces more choices to review.

Consistency is the hidden cost

A generated scene may look superb until the product changes shape between cuts, a presenter’s hands become inconsistent or the lighting shifts for no narrative reason. Teams need visual references, continuity notes, approved key frames and a disciplined review process. When consistency is critical, it can be faster to film or animate a difficult detail properly than to regenerate it repeatedly.

Voice, sound and the human factor

AI voices can be useful for multilingual content, drafts, training and some product films. But delivery still needs direction: emphasis, pronunciation, pauses, emotional credibility and the relationship between voice and picture. A convincing voice is not automatically an appropriate one. Consider consent and usage rights where likenesses or cloned voices are involved, and avoid creating the impression that a real person said something they did not.

The value of a hybrid workflow

A practical workflow might begin with a real interview or studio product shoot, use AI to extend environments or create supporting illustrative sequences, and finish with professional editing, sound design and colour. This approach protects the evidence that matters while allowing creativity where it genuinely adds value. AI is strongest as a production capability within a well-managed process, not as a replacement for having one.

A simple five-question AI review

Before approving a generated shot, ask: Is the factual content correct? Does it match the real product or person? Do we have the necessary rights and consent? Would a reasonable viewer be misled? Does this shot improve understanding or merely look impressive? If the answer to the last question is “merely impressive”, consider cutting it.

What the industry data suggests

Wistia’s 2026 State of Video report, based on a survey of more than 900 professionals and analysis of over 13 million videos, describes growing demand for video alongside constrained budgets. It also reports that AI use is concentrated in pre-production. That points to a practical reality: businesses want efficient content creation, but still need planning, prioritisation and creative judgment.

The filmmaker’s role is changing, not disappearing

The competitive advantage is shifting away from access to equipment alone and towards taste, strategy, technical literacy and the ability to distinguish a credible result from a plausible-looking mistake. Businesses should expect their production partners to understand AI — and to know when not to use it. That is a more useful promise than “everything can be generated”.

A real hybrid example: Cello Voyager

Our Voyager work for Cello Electronics is a useful demonstration of why we do not see AI and traditional production as opposing choices. The film combines studio-based creative production, AI-generated visuals, editing and an AI voiceover. The important decisions still concern the same fundamentals as any commercial film: which features to explain, what the viewer needs to see, whether the product looks right, and how the edit and narration make those details easy to follow.

What AI does not know about your client

A generative tool has no independent understanding of a client's commercial priorities. It cannot decide whether the audience needs reassurance about a technical specification, whether a visual would misrepresent a product, or whether a scene feels out of step with the brand. Those are briefing and direction questions. We have spent more than two decades working through them with clients; AI changes some production options, but it does not remove the need to ask them.

Where we would deliberately avoid AI

We would not replace a genuine customer interview with an invented endorsement. Nor would we rely on a generated close-up to demonstrate a precise connector or interface without checking it against the real product. For a corporate story, real employees and real experiences are often the reason the film is persuasive. AI can assist with archival treatment, translations, visualisation or supporting sequences, but authenticity should remain deliberate, not accidental.

A better question for agencies and clients

Instead of asking, 'How much of this film can AI make?', ask, 'Which production method best serves each shot?' A studio product detail may need controlled photography; a complex feature may need motion graphics; an atmospheric transition might benefit from generation. The answer can vary within the same 60-second film. That flexibility is where a genuinely hybrid production approach becomes useful.

Put it into practice

Before commissioning the next video, run a short planning workshop with the people responsible for marketing, product accuracy and sales. Write down the intended viewer, the decision they are trying to make, the single most important message and the evidence needed to support it. Agree where the film will appear, how long it needs to remain useful and who can approve the finished work. This simple exercise often reveals that one large film is not the only answer: several carefully targeted edits may serve different stages of the customer journey better.

Then make a priority list: essential footage, desirable supporting scenes and optional creative treatments. Cost each group separately and identify what can be captured or created at the same time. Put the most important content first in the schedule, not the most visually exciting. Once the film is live, review its performance with the original objective in mind. If it is not answering the audience’s questions, improve the message and placement before assuming you need another production.

‍