For years, I’ve used branding frameworks with my clients and taught them to my students. The process is always the same: research the audience and the market, build personas, develop strategy, define voice, write the manifesto, design the visual identity. It works, it’s also time-intensive, and the quality depends heavily on how much bandwidth you have that day. So I decided to automate it using Claude Code.

To be clear: this isn't production-ready branding. I wouldn't hand these outputs to a client without significant refinement. But as a starting point, as structured first drafts that capture the right questions and force clarity on positioning, the system saves time. And the process of building it made me want to systematize everything.

I’m not a coder. Never have been. But I’m currently teaching a course on Thinking in Systems with AI, and the premise is simple: most of what we do follows patterns. If you can map those patterns, you can potentially automate them. I wanted to test this theory on myself first, using something I knew inside and out.

The result is a brand marketing platform with two connected workflows.

The first workflow handles brand identity. Six AI agents work in sequence, each building on the previous one’s output. Agent one studies your brand and produces audience research: demographics, psychographics, digital behaviors, the whole picture. From there, the next agent builds three detailed customer personas. Personas feed the strategy agent. Strategy shapes voice guidelines. Voice informs your brand’s point-of-view narrative. And that narrative guides visual identity. Think of it like a relay race: each runner passes the baton to the next.

The second workflow handles content creation. Once your brand identity exists, four more agents take over. Market research comes first, with analysis of trends, competitors, and opportunities. Then, a campaign agent develops three distinct directions to choose from. Content generation follows: social posts, email sequences, video scripts, and ad copy. Finally, an image agent produces ready-to-use AI-generated images. By the end, you have comprehensive brand documents and a full content kit that feel like they came from the same brain (much like how actual marketing teams are supposed to work, but rarely do).

Then I turned it into a website so I could hand the process off to anyone with a brand overview. I took these workflows and built a simple click-through interface. Enter your brand information, click through each step, review and refine the outputs, and move to the next agent. No command line, no complicated setup, just a straightforward web app that walks you through the process. I’m not a developer, but with AI coding assistants, I was able to build a functional prototype. The website isn’t polished, but it works. And it proved to me that the gap between “I have an idea for a system” and “I have a working tool” is smaller than I thought.

The process I followed was methodical. Before touching any AI tool, I wrote down exactly what I do when building a brand, capturing every step and deliverable in order. The sequence matters because you can’t write voice guidelines without knowing your audience. For each task, I asked what information it requires. Audience research needs the brand overview. Personas need audience research. Strategy needs personas. This creates your dependency chain. Then each task became an AI agent. I described the role, the context it operates in, and the expected output format. Think of it as writing a really detailed creative brief, something marketers already know how to do.

The output isn’t perfect. AI still needs human judgment. Someone needs to review the personas and say, “this one nails our customer, that one misses the mark.” Someone needs to choose between campaign themes. Someone needs to make the calls that require intuition, experience, and accountability. The platform handles production. Humans handle taste. That feels like the right division of labor for where AI is today.

Building this required no code. What was required was domain expertise, the ability to break down a process into steps, and clear writing skills to describe what each step should produce. You already have the first in your own field. The second is systems thinking. The third is just creative briefing by another name. If you can write a good creative brief, you can build an AI agent.

Building this changed how I see every other process in my work. Once you see one workflow systematized, you start looking at all your repetitive processes differently. You notice the patterns, and you realize how much of what feels like work is actually pattern execution (yes, that includes the parts we enjoy).

I’m documenting this entire process, including how I mapped the workflows, structured the agents, and connected the pieces. If you want a walkthrough, send me a message. And if you have workflows of your own that you’ve been wondering whether you could turn into systems, I’d love to hear what you’re thinking about automating.

— Lauren Eve Cantor

thanks for reading!

if someone sent this to you or you haven’t done so yet, please sign up so you never miss an issue.

I’ve also started publishing more frequently on LinkedIn, and you can follow me here

if you’d like to chat further about opportunities or interest in AI, or this newsletter, please feel free to reply.

banner images created with Midjourney.