How I'm building a personal branding AI agent on Claude Code (Part 1)
You're good at your job. You have real opinions. But online, you're invisible. Let's fix that.
Let me describe a person I know well.
They’re quite good at their job. They have real opinions about their industry. They’ve seen things work and seen things fail. If you got them in a room and asked them about their niche, you might leave with three things to think about.
But online? They’re hard to find on Google. Sporadic LinkedIn posts. Nothing that would tell a recruiter (or an AI) what they stand for.
That person is me, and probably some of you reading this right now.
I could go on in a whole article about how professionals get discovered is shifting (actually yes, a piece on that is coming up) but first, I’ve decided to build a system that gets me visibility, inbounds, and treated as a credible voice in the field — for me, that’s being an AI-charged marketing leader for the tech space.
A personal branding agent — running on Claude Code — that takes raw reflection and routes it through a structured identity layer before generating anything.
💡 The idea
The problem isn’t that marketers like us don’t have things to say. It’s that there’s a gap between the thinking and the publishing — and most people run out of energy in that gap.
You have a sharp observation on Tuesday. By Friday, when you actually have time to write something, it’s gone cold, stuck inside a note app no one else gets to read. So you post something generic instead, or nothing at all.
I wanted to close that gap with a workflow. Not a content scheduler, not an AI ghostwriter — a system that captures raw thinking and routes it through a structured identity before generating anything. Stories and tasks in the same place. Capture once, activate everywhere.
Here’s what a typical session looks like in practice: I open the agent at the end of the week and answer one question: what happened, what did I notice, what am I thinking about that I haven’t seen anyone else write about?
No structure required. I might type three sentences or three paragraphs. The agent reads what I give it, classifies it, checks it against my identity file, and generates a LinkedIn draft, a newsletter angle, a content hook, and a follow-up question.
The whole thing takes about ten minutes. The editing takes longer. But the hardest part — getting from blank page to something real — is already done.
One design principle runs through the whole thing: the agent never invents your identity. It only activates it. Everything it generates flows through positions you’ve already stated you believe. That’s what makes it different from an AI writing tool.
⚙️ Why Claude Code?
Quick context if you haven’t used it: Claude Code is Anthropic’s agentic tool that runs in your terminal. You talk to it in plain English, it writes and executes code, reads and creates files, and chains multi-step tasks together. Think less chatbot, more dev collaborator who lives in your command line. You don’t need to be a developer to use it, but a comfort with tinkering helps.
I chose it over no-code tools for two reasons.
One: I needed control over the identity layer. A persistent context that encodes my specific positions, not just generic brand guidelines. Template-based tools don’t handle opinionated, specific context well and are built for output, not identity.
Two: this workflow isn’t a single output. It’s a chain: capture, classify, generate, score. Claude Code lets me build the chain, not just individual steps.
🪞 Start here: the visibility test
Here’s a two-minute exercise that will either reassure you or ruin your afternoon.
Open Claude, ChatGPT, Perplexity, or whichever tool you like, and run these four prompts:
"Who are the best [your specialty] creators or practitioners?"
"Can you recommend someone who [does what you do]?"
"I'm looking for a [your title/role] who specialises in [your niche]. Who should I follow?"
"What experts should I learn from if I want to understand [your topic area]?"If your name comes back, great, drop me a message — I’d love to interview you.
If it doesn’t, welcome to the club. Membership is free, the coffee is mediocre, and there are a lot of us.
What’s coming in Part 2?
The full build. How I structured the agent across four layers, the prompts I used in Claude Code, what’s working, what’s half-finished, and what the workflow looks like end to end.
If you want to run this yourself — or just understand how an agentic workflow actually gets built by a non-developer — Part 2 is for you.
In the meantime: run the visibility test. Hit reply and tell me what you find.
The AI;DR
Elsewhere in the AIverse
OpenArt launches Worlds. Asingle prompt or photo goes in, explorable 3D environment comes out. Drop characters in, move through the space freely, capture production-ready shots without losing scene consistency.
Google upgrades AI Studio into a full-stack vibe coding platform — generates production-ready apps with multiplayer capabilities, real-world service integrations, and an agent that picks up where you left off. The barrier to building your own AI workflows just got lower again.
Gamma just overhauled its core product. The headline feature is Gamma Imagine, which generates on-brand posters, logos, and visuals from a prompt. As they approach 100M users, they’ve also rebuilt the template experience so you can update entire decks without breaking the design.
Meme of the week
