Scale your content with AI: my 3 lessons from practice

Why you need a project manager, plus the time and budget required

Do you want to produce content with a bit of help from AI? Moving from experimental to structural AI?

Learn from my mistakes, so you don’t have to make them.

Before we start, the main reasons why it’s so hard to move from experimental to structural.

Change requires courage, from both management and the team
Security and compliance in AI are often handled with fear-based restrictions rather than education, due to limited expertise and capabilities
Management bottlenecks adoption by not allocating time, budget, or resources. Not because it isn’t important, but because of limited in-house expertise and a lack of urgency

3 lessons I’ve learned on integrating AI and making it a natural part of daily work. I’ve applied these with multiple clients, both in my role as a strategist and as part of collaborative teams.

Get a project manager on board
Spend significant time on configuration and training, repeat frequently
Rebuild your AI tech stack every once in a while

The result? You’ll structurally free up time within your team, so you can focus on elevating content quality, not just increasing quantity.

Let’s dive in.

Lesson 1. Get a project manager to become even more efficient

This one is what surprised me the most. I didn’t see it coming, yet it’s so obvious. Having a project manager to ensure the configuration and training take place, to safeguard the processes, and keep track of the progress, makes the difference between good and great.

By using AI for content production, the production process becomes more complicated. You’re adding more steps for reviewing and approval. Next to that, you’ll be working on multiple pieces at once.

At the same time, because you’re using AI, you tend to forget what you did faster and lose track of where you are.

Finding out the hard way

You need someone to keep you on track and accountable. Speaking from experience here. We found out the hard way.

You’ll keep up in the first weeks, but as you progress, it also becomes more complicated. We were working on so many pieces simultaneously and had additional reviewing and editing steps that we lost the overview.

To do: get a project manager

As mentioned, this project manager helps you to progress and keep track. This role is not a side hustle or something you are just adding to the already full plate of a content producer.

It’s a dedicated job that needs focus and time allocation. For an SME that produces and distributes 2 to 4 content pieces a week, I’d say 8 hours a week would do.

My advice is to have a project manager who’s familiar with the content processes and has an interest in AI. By preference, someone from the existing team, whom you can free up for at least 20 to 25% of their time for this role.

Lesson 2. Spend serious time on configuration and training to become more efficient

As most teams struggle for time, it’s our natural behavior to dive in immediately and get things done. However, with AI, the magic is in the configuration of the tools that speed you up. Not in grinding the work.

How much time you should spend on configuration

Good to know that the configuration is the most overlooked part of AI. Mostly, you upload the brand guidelines and think that you’re good to go.

It doesn’t work like that. You must write detailed instructions and give examples to feed AI to get your desired output that takes little review and editing.

From experience, I spend at least two full days on the configuration and another two days on testing and refining, so that I can produce content in minutes. I have done this for multiple clients, and time after time, this time investment pays off double in the efficiency gains.

→ The more time you spend upfront, the more you’ll save on editing and reviewing

For your AI content configuration, there are 3 types of input needed:

Brand voice, lexicon, and writing rules, drilled down to the specific details, whether you use a capital after a ‘:’, using punctuation in lists, and when to use jargon or when not.
Structure of content: predefine the storyline of your blogs, whitepapers, or socials. A clear framework ensures each piece follows a consistent logical flow and narrative, remains recognizable, and maintains the same level of operational detail.

For example, I always use the structure ‘problem – solution – benefit – deeper benefit’ combined with ‘to know – why it matters – to do’.
Use best practices for writing social copy: feed AI what works on social for your industry and audience. For example, I always copy and paste the 20 best hooks for LinkedIn and use them in my instructions.

There’s a lot more to this, and it’s a bit more extended and complicated with instructions, GPTs, and AI assistants, but for the sake of clarity, I simplified it a bit.

Training on data safety, compliance, and ethics

There is no such thing as training once. Keeping your IP, data, and confidential data safe and secure is non-negotiable. Yet, you do need to repeat this message over and over again, just because we tend to forget in the heat of the moment.

→ Besides, as AI improves every day, you’ll need to keep up.

The training is also a great moment to share experiences and insights within the group. Use training sessions to learn from each other, not just top-down updates.

To do; my 2 tips

Spend significant time on the configuration and testing of your configuration. Retest it frequently to keep up to date with new functionalities and features. For an SME, around 4 to 5 days to configure and test, and 1 day every two months to (re)test and update.
For the training, besides the kick-off, I advise having monthly meetings to update on new features, policies, learn, and share experiences.

Lesson 3. Periodically step back and rebuild to become resilient

As AI progresses daily, you should review your AI tech stack accordingly. As we know, new functionalities and features are added, and AI continues to improve, becoming better, faster, and stronger.

→ How you’ve set up the processes and stack can become outdated in a few months.

Stepping back and rebuilding refreshes systems and approaches, ensuring long-term resilience and relevance.

From experience: Q4 2024 versus Q2 2025

How I’ve set up the production process with GPTs at the end of 2024 is suboptimal for how AI works in 2025. Therefore, before handing over the work, I rebuilt everything in a new setup to fit the latest AI developments.

To do: make the project manager responsible

The best thing you can do is to assign the project manager to this. It’s his or her role to keep track of developments and signal timely when the AI stack needs reviewing and rebuilding

Wrapping it up

If you want to scale from experimental to structural use of AI in your content processes, the first thing you need, besides the C-suite buy-in, is a project manager.

The project manager can also help you with the configuration of your AI tech stack. And keeps the overview, keeps track, keeps you accountable, plus signals timely when it’s time to update your AI tech stack.

All in all, a project manager pays for itself because you save time and costs in reviewing and editing, and you stay up to date on the tech, rules, and regulations.