Subscribe Now: Apple Podcasts | Spotify
My View on Using AI in Your Meta™ Ads and Marketing
AI is everywhere right now.
In the online business world, it is almost impossible to avoid. Some people are talking about it like it is going to replace everyone in marketing. Others are avoiding it completely. As usual, I think the truth sits somewhere in the middle.
I want to be clear from the beginning. I am not an AI expert. I am not using it for wildly advanced things in my business. What I do know, after 14 years of running Meta™ Ads and managing client campaigns, is what good marketing looks and feels like. And that perspective shapes how I use AI, where I see the benefits and where I think the line needs to be drawn.
Where I stand on AI replacing humans
I have been hearing more stories from inside the industry about agencies letting go of large parts of their teams and relying heavily on AI to run client ad accounts with minimal human oversight.
And honestly, that concerns me.
Not because AI is automatically bad, but because ads need human judgment. They need human review. They need someone with experience noticing when something looks off before it goes live.
I have seen ads in my own feed from businesses openly talking about running their marketing heavily through AI. Some of those ads had basic issues that any experienced human would have caught straight away, such as terms and conditions copy appearing in the body of the ad and incorrect image sizing in the wrong placements. These are not complex strategic issues. They are simple quality control problems.
To be fair, mistakes can happen with human teams too. Meta™ is constantly changing things in Ads Manager, sometimes even altering settings just before you publish without you realising. That is exactly why I believe campaigns need to be triple-checked by human eyes, not just reviewed once and sent live. When AI is doing the bulk of the work without enough oversight, those checks are more likely to be missed.
My take on the Russell Brunson example
You may have seen the recent discussion around Russell Brunson, founder of ClickFunnels, saying he has fired his copywriting and social media teams and is now using AI for all of that work, reportedly saving over $650,000 a year. On paper, I understand why that sounds attractive.
But I would not make that choice.
Great copy is not just about putting technically correct sentences in the right order. The copy that actually converts has emotional depth. It has creative nuance. It uses unexpected angles. It reflects the specific voice and beliefs of the brand in a way that feels real to the person reading it.
Those things evolve over time. Your audience changes. Your positioning changes. The market changes. The conversations people are having in their heads change. Keeping up with that requires observation, research and judgment. It requires a human who knows the business deeply and is paying attention.
AI can absolutely help with output and speed. But if it becomes the only source of the message, I think there is a real risk that you lose the nuance, originality and emotional relevance that makes marketing powerful.
How I actually use AI in my business
So where do I use AI? Quite a lot, but in a very specific way.
The biggest role it plays for me is helping me organise my thoughts. This podcast is a good example. Each week I know what I want to say. I have the ideas, the opinions, the real stories from client work and the experience from managing ads every day. AI helps me take all of that and shape it into something more coherent and easier for someone else to follow. The expertise and point of view are mine. AI helps me package it clearly.
I use it similarly with ad copy. I know the framework I want to use depending on the objective, the audience, the message and the angle. I know the brand voice. I know the emotional hooks I want to test. AI helps me produce the level of volume Meta™ now needs, because the platform requires far more creative variation than it used to.
That part is important. AI helps me with output, not with replacing strategic thinking and how to write the ads – that is all fed in first.
Why generic AI ad copy is usually poor
This is the problem with using general AI tools casually for ads. If you open a tool and type “write me some Facebook™ ads”, you will usually get generic, flat output. The copy tends to be broad, emotionally weak and full of common phrases with no real angle. It does not know your market, your offer or the way your ideal client thinks. It also doesn’t know the framework for writing ads that convert (believe me I have tested it!)
That is why I often describe generic AI output as slop. It may look polished at first glance, but it rarely has the specificity or tension needed to convert.
The real determinant of quality is not the tool itself. It is the person directing it. The output depends on what you feed in, the clarity of your inputs and your ability to recognise what is good enough to use and what is not.
Why I built Lily LeadGen
Because I work so closely with Meta™ Ads every day, I built my own AI tool specifically for lead generation ad copy. It is called Lily LeadGen.
I built it because general AI does not know my methodology. It does not know the specific formula I use for lead gen ads. It does not know what is working well now versus what worked eighteen months ago. Lily does, because I trained her on it and I keep updating her based on what I am seeing across client accounts and the platform itself.
That ongoing human refinement is what makes the tool genuinely useful. Without that layer of expertise and updating, it would just be another content generator.
Why I’m cautious about the Claude x Meta™ connector
I also want to touch on something newer, the Claude connector that can interact directly with Meta™ ad accounts.
At this point, I am not rushing into it. There are two main reasons for that. The first is security. My clients’ ad accounts are extremely important to me, and anything that touches account access, integrity or security is something I take seriously. Even if it is officially approved, I want to observe, assess and make a measured decision rather than jumping in early.
The second reason is that Meta™ has not exactly been consistent with how it has handled AI-related behaviour in ad accounts. Just two months ago, my own account was flagged twice while uploading ads for clients, triggering video verification so Meta™ could confirm I was a human. I have also heard of businesses having accounts deactivated entirely because they were using AI tools connected to their accounts.
Yes, the Claude connector changes that picture somewhat because it is approved. But I am still taking a wait-and-see approach. I do not think caution is a weakness here. I think it is the right way to approach tools that could affect the stability and security of client accounts.
Final thoughts
AI absolutely has a place in your marketing. It can make you faster. It can help you manage the growing demand for creative volume. It can help you organise ideas and accelerate production.
But it is not a replacement for human expertise. The businesses that will get the best results from AI are not the ones handing everything over to it. They are the ones who know their craft well enough to guide it, shape it and check it properly.
That is how I am approaching it in my business. Not rushing. Not rejecting it either. Just learning, refining and using it in ways that support quality instead of weakening it.
And I think that is the smartest place to be right now.
I would love to hear your thoughts...