Corporate video with AI: what can be done today, what can't and when shooting is still worth it

What AI can do today in a company video

It's the doubt we hear most: why spend money on a shoot if AI can already do it? We'll give a proper answer, because at Bostak we use AI daily, and that's exactly why we know its limits.

Today there are four clear uses. The first is avatars: a virtual presenter, created from a real person or from a catalog, reads your script before a camera that doesn't exist. HeyGen and Synthesia are the best known. The second is generating images and video from text, with Sora, Veo, Runway or Kling, which create scenes that were never shot. The third is editing: removing silences, transcribing, subtitling, translating or extending a shot. And the fourth is voice: synthetic voice-overs and dubs in many languages.

How do you make a corporate video with AI? Essentially the same way as a normal one: a clear idea, a script and a decision about style. Then, instead of shooting, you generate: you choose an avatar or an aesthetic, produce each shot, edit, add voice, music and subtitles. What changes a lot is production time and what you can change afterwards. And a fair warning: the tools change every few months, so what you say today may fall short in six months.

What hasn't changed is the most important part: what you want to say and to whom. If your script is weak, AI will make a weak video much faster. That's the part that still needs the most work, and here the judgment of someone who's been doing this for years makes a lot of sense.

What it's good for, and what it isn't

These are the uses where AI works very well at a company, and it's true that it saves a lot.

Training and procedures. Videos explaining how to use an internal tool or a rule, which need updating often. An avatar gets changed in an afternoon; a shoot doesn't.

Versions in other languages. For a company with teams in several European countries, being able to repeat a message in six languages without shooting six times changes the numbers. Always check the translation and sector terms: we cover it in how to adapt a video into several languages.

Short-lived pieces. An internal announcement, a week-long product update, a test of an idea before producing it.

Impossible images. A visual metaphor, something invisible, a place you can't go or a world that doesn't exist. This is where it works best with creativity, as we explain below.

And these are the cases where we'd recommend thinking twice.

Presenting your company. People want to see who's behind it. If your presentation video uses an avatar, many will notice and draw a conclusion about you. In company presentation video we explain why this video weighs so much.

Testimonials and client cases. A generated testimonial isn't a testimonial. As well as not being credible, it can lead to legal problems.

Showing your product, factory or team. AI makes up what it doesn't know. If your machine has a specific shape, or your lab is a certain way, what gets generated may look similar but not be the same. In technical sectors, that error shows right away.

Attracting talent. Nobody imagines working with an avatar.

The risks to look at before using it

There are four matters worth reviewing. They're not there to scare you, but because they get resolved easily if you think about them first.

Trust. More and more people recognize a video made with AI. Often that doesn't matter, for example in an internal or funny piece. But if your brand lives on trust, as in health, finance or professional services, it's worth weighing up carefully.

Image and voice. To use someone's avatar or voice, you need their written permission, and that includes employees, and what happens when they leave the company. It has to be agreed before starting.

Rights. Not everything a tool generates can be protected in the same way or used with the same peace of mind in a campaign. Read the tool's terms and, if it's an important project, consult a legal adviser. References come into it too: don't ask for an imitation of a specific actor or brand.

Transparency. Europe is regulating how content generated or manipulated with AI should be disclosed. Our recommendation is simple: say when it's been used if someone could think it's real, and check the updated rules with your adviser. Besides the legal side, there's also a brand reason: being found out is much worse than saying so.

And there's another risk that gets talked about less: sameness. If everyone uses the same tools with the same presenters, all the videos end up looking alike. The same avatar can turn up at companies in different sectors. So the difference is still made by the idea and art direction, just as before. In creative corporate videos you'll find ideas for breaking out of the usual.

What we do with AI: two real cases

Here are two projects you can see, because they help more than any list.

The Bunker Christmas is a campaign made with AI, with a world inspired by Willy Wonka, generated characters and settings, and two thirty-second spots designed for social. The usual Christmas aesthetic didn't fit a street brand, and shooting something like that physically would have been impossible. What makes it work is what was behind it: a concept, a very carefully worked art direction, and an edit and music conceived for that piece. AI gave us the possibility, and we brought the idea.

Canta Coco is a children's song format in which we developed the character, the visual universe and the animation, with generative AI for image and voice, plus traditional post-production, art direction and human music production. It has passed half a million views on YouTube and has more than ten episodes. We're now taking it towards a new format aimed at companies.

On top of that, in our day-to-day work with podcasts we have our own system, built by Albert Murillo, that turns a horizontal episode into a batch of vertical reels. That saves hours of trimming and leaves the team what really matters: choosing which moment deserves to go out.

And when do you need to shoot? With Qilimanjaro, a quantum computing company, we film real people in their setting, because credibility with investors and talent is what's at stake. AI can help in the process, with transcription or subtitles, but the image of those people has to be real.

The most practical approach: combine shooting and AI

This isn't about picking one of two sides. On most projects, what works best is using each thing for what it's good at.

Shoot what needs to be true: your company's people, your product, your premises, a testimonial. With a well-planned one-day shoot you already have material for many pieces.

Use AI around it: to test ideas before the shoot, for example a generated storyboard you can show the client in a day; to extend backgrounds or clean up images; to create shots that would be impossible or very expensive to film; to translate and dub into other languages; and to produce versions.

Here's an example. A company wants to present a product in five European countries. The team and the real product get shot in a day, and the main video is edited in one language. Then, with AI and human review, versions are made in four more languages, with adapted subtitles and captions. Finally, some supporting shots are generated to illustrate the problem the product solves. The client sees real people, and the project costs much less than shooting five times.

Here's a practical question to help you decide. If people could tell it was made with AI, would it hurt you? If the answer is no, go ahead. If it's yes, shoot. And another: who's going to review the result? Because the typical mistake isn't the tool's. A video gets published that nobody has looked at critically, and it contains strange hands, badly written text or a figure the AI made up. Someone with judgment has to give the go-ahead.

How to decide it with us

At Bostak we won't tell you everything has to be a shoot, because we use AI daily, or that AI solves everything, because we know it well. Albert Murillo, our creative director, comes from directing film and music videos and uses these tools daily: that mix is exactly what's needed to decide what gets shot and what gets generated.

We work with companies in Barcelona, the Baix Llobregat and the rest of Europe. We can do anything from a fully generated piece to a shoot with AI parts, and help you get the most out of versions for other markets. If your project can be solved with a cheap tool and without us, we'll say so too. We'd rather you came back when you need something that's really ours to do.

If you're considering a video for your company, see our corporate video page. And if you want to know what makes sense in your case, book a free diagnosis: on a video call you tell us what you need to say and to whom, and we tell you what we'd shoot, what we'd generate with AI and what we'd leave for later. You can also show us a video you've already made with AI, and we'll tell you what works and what shows.

FAQ

Got a project in mind?

We tell stories in video end to end — production, filming and post-production.

Keep reading