There is a distinction that rarely comes up in conversations about artificial intelligence: the difference between using AI to create and using AI to automate.
The first method, which everyone is already familiar with, is exemplified by tools like Canva's image generator . The second method is where you can truly save a lot of time. And that's what this article is about: how we create AI-powered workflows that help us streamline and automate a significant portion of digital signage content production.
AI-powered workflows in content production
The problem wasn't creative. It was operational.
We have been working with a client in the retail sector for years, one of the first brands in Portugal to systematically integrate digital signage in-store. The volume of content is considerable: campaigns, communication for various channels, and brochures, many brochures, with different rhythms and schedules, which need to be transformed into video animations to be displayed on the many screens.
Weekly, bi-weekly, by product category. Each with its own format, set of articles, and information. The layout remains the same for months. The content is constantly updated.
What changes in each cycle? The title, the description, the price, the image. Multiplied by ten articles, distributed across various compositions, repeated week after week.
It seems simple. And it is, but the process didn't reflect that.
The problem isn't the work. It's the search!
Each update began with a hunt: finding the right layer in a composition with 50, 60, sometimes 100 layers. Whoever built the project navigates this on autopilot.
But even for those who know it by heart, repetition has a cost. And when it's someone less familiar with that specific structure, that cost multiplies.
Searching is not the same as producing. This was a structural problem, not a problem of competence.
What if the data lived outside of After Effects?
The idea is simple to explain: instead of the data for each item—title, description, price, image—being buried within the layers of a composition in After Effects, they now reside in an external file with a clear structure.
A table that anyone can fill out. One record per item, one field per column.
After Effects is no longer just a place to edit content. It becomes the engine that takes that data and renders it in the correct template.
The designer did the creative work once. From then on, updating a brochure is just filling out a table and running a script.
The logic isn't unique to After Effects. It's applicable to any workflow with high-volume, repetitive production: separate what changes—the data—from what doesn't change—the visual structure.
Where do AI-powered workflows come in and what problems do they solve?
There was a practical problem: nobody on the team was proficient in ExtendScript, the scripting language of After Effects. It's a JavaScript with its own peculiarities, scattered documentation, and a learning curve that doesn't justify the investment when it's not core to the work.
This is where AI came in, not to create content, but to write code.
With a description of the problem, the project structure, and the desired outcome, we were able to develop functional scripts that automate parts of the process that were previously done manually.
We described what we wanted, the AI wrote the code, we tested and iterated. The result was scripts for:
- create compositions and organize projects
- prepare exports automatically
- Populate articles from external data without manually touching the layers
The process isn't instantaneous. There's iteration, there's testing, there are adjustments. But the difference between "we don't have anyone who knows how to do this" and "we can build this" is enormous and is changing the way we approach larger, recurring projects.
What changes in practice, and for whom?
Saving five minutes searching for a layer doesn't seem like much. But five minutes per article, multiplied by ten articles, by several compositions, by several brochures throughout the year, adds up to a significant amount of accumulated time.
And above all, time is a concern for Motion Designers, who are paid to do more creative tasks than searching for layers.
The other change concerns accessibility and ease of management.
With the data outside the production tool, in a simple, user-friendly, and understandable structure, anyone can fill it out and speed up production.
For those working in creative production and wanting to apply this to their reality: the path begins by mapping what is truly repetitive in your workflow. Then, describe this problem to the AI in detail: project structure, what changes, what remains the same, the desired outcome. The learning curve exists, but it's shorter than it seems.
What do AI-powered workflows mean for an agency's work?
There is a recurring discussion about the impact of AI on creative work. Our perspective is practical: AI does not replace creative work, it replaces mechanical work that consumes creative time.
In this case, freeing the team from managing layers means having more time for what really matters in a digital signage project: the communication strategy, the visual hierarchy, and adapting the message by format and moment. The work that no script can do.
We are still developing and refining this approach. But the initial results confirm that the direction and logic behind it are transferable to any team dealing with recurring volume production.
If you work with high-volume content production and want to understand how this approach can be applied to your workflow, talk to us.
Any other ideas about this article? Check the usual places. 🙂
