AI workflows...actually worth it or overhyped?

Hey guys,

You can’t scroll for 30 seconds on Twitter right now without coming across a multi-step n8n workflow that promises to revolutionise your creative process.

And a lot of these workflows can be vibe coded in Claude, meaning that you can create some really impressive things pretty quickly, without needed technical experience.

But are these workflows actually providing tangible business value or are they overhyped for what they are?

As someone who spends a lot of time building AI workflows, here are my thoughts…

But first, here’s what’s I’ve come across in AI this week:

• This guy built an AI slack agent that monitors channels in real-time

• My friend JB built a research agent with Claude that scrapes Reddit, YouTube, and TikTok on command

• Someone on X wrote a fun prompt to get ChatGPT to guess your IQ

AI workflows…actually worth it or overhyped?

I had a consulting call with someone recently who had the most impressive n8n workflows I've ever seen.

He'd managed to automate almost the entire creative process end-to-end.

But his ads sucked and he was getting his ass kicked by his creative agency. Because he'd spent more time learning how to build AI workflows than learning what actually makes a good ad creative.

And this pretty much sums up my thoughts about AI workflows in ad creative right now.

Despite Gumloop and n8n flows being wildly popular, the uncomfortable conversation that no-one wants to have is that barely any of these workflows actually provide any business value.

I think the reason for this is that most people are focusing on the wrong thing.

The hardest part of AI workflows is not the piping.

Building a Gumloop or n8n flow is honestly pretty easy once you watch a couple of YouTube videos. Anyone can connect APIs and set up automations with a bit of patience.

The hard part that no-one wants to focus on is the context.

It's you spending hours creating a 10-page document on how to make great static ad headlines - complete with insights, examples, and reasoning acquired from years of hands-on experience across multiple accounts.

And then doing that same work across every single task in your creative strategy process to build a comprehensive library of context to inject into your workflows.

THAT'S the hard part that actually moves the needle.

What this means for you

If you're building AI workflows (and you should be), here's my advice:

Start with your expertise, not the tools.

Before you touch Gumloop or n8n, document your process. What makes a good headline? How do you identify winning angles? What research methods actually work?

Write it all down. In detail. With examples.

Then build the workflow around that knowledge.

The most successful AI implementations I've seen follow this pattern:

deep domain expertise → systematic documentation → technical implementation.

Not the other way around.

The brands winning with AI workflows aren't just the ones with the fanciest automations. They're the ones who've combined solid technical execution with years of creative strategy expertise.

So before you spend another weekend learning a new automation tool, ask yourself: have you actually documented what you know about making great ads?

Because that's where the real value lives.

The context to inject into your workflows

There are two different types of context you should be adding to your workflows - Brand Context and Domain Context.

Brand context is all of the information about your brand (e.g. brand information, value props, guidelines…).

Domain context is the information about how to be great creative strategist (YouTube videos, tweets, email newsletters, books, podcasts).

You need both of these added to your prompts/workflows for the models to fully understand how to be a creative strategist for your brand.

Sidenote: If you want to learn more about context engineering, we are covering it in the AI Creative Strategist Blueprint live training, it’s not too late to sign your team up :)

I cannot stress this enough…this is the sauce. This is the reason your prompt or workflow will or will not generate actual good ad ideas. And it’s not easy - but neither’s trying to come up with ad ideas - so it shouldn’t be.

So, going back to my original question, are AI workflows actually worth it or are they overhyped?

They’re overhyped in the way that most people use them. But they’re not overhyped in the way they should actually be used. I think the same thing can be said for a lot of AI tools in creative strategy. IF you spend the time training them properly, they can be very valuable.

Technically sound workflows + Legitimate expert context. That’s the formula for AI systems that are actually worth it.

Announcements:

PJ Ace did a killer session on Veo 3 in the AI Creative Strategist Blueprint yesterday. And we’ve got another big one tomorrow on prompting for DeepResearch and Image Gen.

If you haven’t joined the program yet, all replays of past sessions are made available so that you can catch up without missing a training! Click here to sign your team up.

Enjoyed this email? Forward it to a friend who’d find it valuable!

And if someone forwarded this to you, here’s a link to signup. I send tips on how to make more winning ads with the help of AI straight to your inbox every week!

See you next week,

Alex