AI content production is, for many in-house teams, the obvious answer to a simple equation: more channels, more formats, the same budgets. Anyone leading a brand today needs content every week for the online shop, social, newsletter, stores and partners. No team can do that by hand any more.
At a fashion house I led three departments, marketing, content and production, with a team of 37 people. Today I lead a team of 22 across markets in EMEA. My experience: AI only helps an in-house team when it is clear beforehand how the team works.
- AI content production needs clear processes first, then tools.
- New roles emerge: prompts, review and the care of templates.
- Quality over quantity: more content is not a goal in itself.
- Leadership decides whether AI relieves the team or overwhelms it.
Contents
Why AI content production belongs in-house
In-house teams have long been the norm. According to the study by the US advertisers’ association ANA, 82 percent of its members have their own in-house agency (see here: ANA, The Continued Rise of the In-House Agency 2023). The reason is simple: people who live the brand every day produce faster and closer to the business.
With AI this advantage grows. An in-house team knows the products, data and rules of the brand and can feed the tools with them. An external agency has to build that knowledge first. How I weigh in-house against agency in general is covered in the article Building an in-house agency.
Processes before tools
Before a team introduces a tool for AI content production, it should write down its process: who briefs, who produces, who reviews, who approves, where do the files live? This picture quickly shows where time gets lost. Often it is not the creative steps but searching, aligning and reformatting.
These are exactly the steps AI speeds up well. A product text in five lengths, an image in twelve formats, a newsletter in three languages. The creative decision stays with the team, the repetition goes to the tool.
A small test before the big rebuild helps. A team picks one process, such as product texts for the online shop, measures time and quality before and after, and then decides whether to extend it. That builds trust in the tools without the whole team having to relearn at once.
New roles in the team
With AI, roles change. You need people who maintain templates and prompts, people who check results for brand and legal fit, and people who test new tools. These tasks did not exist before, or only at the margins.
I would not hire for these roles but develop them within the team. Anyone who makes good images or texts today can also write good instructions. How teams grow without losing quality is described in the article Marketing organisation.
That includes training that stays close to the work. Rather than general courses, a team learns best on real tasks from its own calendar: the next campaign, the next newsletter, the next collection. That is how judgement grows, and judgement is exactly what a team needs to tell good results from arbitrary ones in AI content production.
Quality over quantity
The biggest risk of AI content production is not poor quality but too much mediocre content. When content costs almost nothing, the temptation grows to fill every channel every day. Customers notice the difference.
That is why I do not measure content by volume but by impact: clicks on products, time on page, sell-through, brand metrics. Fewer, better pieces almost always beat more, arbitrary ones.
In practice this means: before every series, the team defines what a piece of content is for and how its success will be recognised. Whatever does not work after a few weeks is stopped, even if AI would make it cheap to produce. This discipline feels unusual at first, but it keeps the team free for the content that really moves the brand forward.
My conclusion: AI content production belongs in the in-house team when the processes are clear and leadership puts quality before quantity. Then AI relieves the team instead of overwhelming it.