What actually are AI Agents?

Hey guys,

By the end of this year, I believe 50%+ of what creative strategists do today will be done by AI agents, if not more.

But I’ve noticed that there’s a lot of misinformation out there about exactly what AI agents are. Today I want to clear that up and share how we plan to use AI agents and workflows throughout the rest of 2025.

Let's get into it...

Agents ≠ Automations

Many people use the term 'agents' when they're actually just talking about AI automations. Here’s how I think about them...

• Automations are like having a robot follow a strict recipe - great for repetitive, predictable tasks

• AI workflows are like having a sous chef who can improvise a bit with the recipe

• AI agents are like hiring a full chef who decides what to cook based on what's in your fridge, your dietary preferences, and what you've enjoyed in the past

Agents interpret trends, make recommendations, and evolve based on feedback. They're dynamic partners, not static tools.

In the context of ad creative, an AI automation might look like this:

1. Extract performance metrics from ads manager into a Google Sheet every morning.

Where as an AI agent might look like this:

1. Pull performance metrics from ads manager daily.

2. Detect trends and anomalies to identify performance patterns.

3. Flag underperforming ads to kill and suggests scaling high-performing ones.

4. Analyze ad performance to recommend tweaks or new creative concepts.

5. Provide daily summaries with actionable insights and visualisations, assigning tasks to the relevant team members.

6. Learn from feedback to improve future recommendations.

The Future of Ad Creative Is Agent-Driven

Now imagine this, a media buying agent, working alongside a creative strategist agent and a creator communicator agent - building and optimizing ads for you 24/7.

That's the power of AI agents. It's scary, but also very exciting.

What This Means For Your Creative Strategy

If it’s your job to make ads for your brand or clients, here's what you need to know:

1. The tech isn’t quite there yet - A lot of the ideas mentioned above aren’t possible with the tech available today. Start learning about AI agents now (I'd recommend talking to ChatGPT) so that you’re ready by the time the tech is there.

2. Think about your creative process - A much better way to approach AI agents and workflows is think about your current process and work out which parts AI can improve/automate. This is a lot more defensible than becoming reliant on all of these tools that promise a full-stack ‘infinite ads machine’ at the click of a button.

3. Start building workflows today - The companies that start experimenting with workflows and agents now will have a massive advantage. You can start as simple as creating flows in Poppy or Gumloop (like Jimmy and I covered in this video).

And the beauty of these agents is that they get better over time as they learn from more situations and outcomes - just like a good creative strategist.

The use cases will be endless, but the window of opportunity to get ahead of the curve won’t last forever.

If you've been experimenting with AI agents for your ad creative, I'd love to hear about your experience.

And by the way, I dropped a podcast with Andrew Faris on the crossover between AI and creative recently which covers a lot of the topics above in more detail. Go and give it a listen here if you haven’t already.

See you next week,

Alex and Jimmy