• Alex Cooper
  • Posts
  • I turned Claude Code into a superhuman creative strategist

I turned Claude Code into a superhuman creative strategist

It works for me 24/7 and gets smarter every day...

Author

Alex Cooper
July 28, 2026

Hey guys,

The more I use agentic AI tools, the more I really do believe Claude Code and Codex are the biggest opportunity for creative strategists right now.

Recently I've had Claude Code analyze my ad accounts, find new ad inspiration, and even create ads for me…all without me lifting a finger.

I've also built a living, self-improving brain that gets smarter every time me or my team uses it.

So today I'm walking through exactly how to do this, even if you've never used Claude Code before.

Let's get into it :)

Prefer to watch this on YouTube?

But first, here's what I've found interesting this week:

How to turn Claude Code into a superhuman creative strategist

TLDR:

  1. Install Claude Code and connect your tools (Slack, Notion, Gamma, Higgsfield, Parker) as MCPs

  2. Clone the Parker Brain into a Claude Code folder for your brand

  3. Load the brain and start ingesting your context (Slack, Notion, meetings, client feedback)

  4. Run workflows (static ads, creative reports, competitor gap analyses)

  5. Turn the best workflows into routines that deliver into Slack on a schedule

For the purpose of today's newsletter, I'm going to be explaining my process in Claude Code, but you can assume that everything that I do here can also be done in Codex.

What is Claude Code?

Claude Code is like Cowork, but on steroids.

For those of you who are more used to Cowork or the regular chat, the main difference with Claude Code is that the memory is baked into a local folder on your machine, so you're chatting inside a folder that has all the context about a specific project or brand - instead of memory in Claude.

You can do all the same things here that you can in the regular chat. MCPs, artifacts, visuals, reports, all of it. But because the context lives in the folder, it gets smarter every time you use it.

It’s a little intimidating at first. But once you understand the file structure, it becomes second nature. And when you catch the bug, you really don't have a need to go back to the regular chat ever again.

Step 1: Connect all your tools

Claude Code is already incredibly powerful on it’s own. But it gets really useful when you plug it into every other tool you're already using every day.

You want to connect as many tools as possible. Go to customize → connectors. Add everything you use in your workflow. For me, that’s:

  • Google Calendar (meeting notes)

  • Gmail (for context)

  • Slack (ingesting team messages)

  • Notion (briefs, docs, ideas)

  • Gamma (for reports)

  • Higgsfield MCP (statics and videos)

  • Parker MCP (pulls in your ad account data, competitors, AI tags)

The more you connect, the more context Claude can pull and deliver outputs to the places you already work in.

Before we get into the Brain itself, you'll need the Parker MCP set up as a connector. This is what pulls all your ad account data, competitors, AI tags, and everything else from Parker.

To install: go to connectors, click "add a custom connector", then paste this link:

You now have the Parker MCP inside your Claude Code, and it'll pull in all your ad account data, competitor tagging, and everything else from Parker into every conversation.

Now for the Brain itself.

Step 2: Install the Parker Brain

This is what makes the whole setup self-improving. Without it, you've got a smart chat inside a folder. With it, you've got a system that gets smarter every time you or your team uses it.

The Parker Brain pulls context from your Slack, Notion, and everywhere else you work, and updates itself in your sleep.

Bring it into a Claude Code folder for your brand and type this prompt:

“I want you to clone this brain for my brand [X]. Follow the instructions exactly as is. https://github.com/real-simple-labs/parker-brain.git“

A few things to know:

  • It’ll take a few hours to generate the first time

  • I’d recommend doing it on a 20x max plan to avoid session limits

  • Put it on auto mode so you don't have to click through permissions

Once it's done, open a new chat, load up the brain, and run /get started. Example below:

NOTE: It’s not mandatory to use the Parker brain to get good outputs from Claude code - this is just the part that really makes it self improving…and you get to tap into 170+ creative strategy knowledge docs that we’ve already built :)

Use cases: what this actually looks like in practice…

Here are the three I'd start with.

Use case 1: Monthly creative report in Gamma

Prompt: "Build my monthly creative report. Give me a performance story over time. Tell me what's working, what's not, and ground it in customer evidence."

Gamma delivers the full retro with everything the team needs: performance breakdown, top ads by impressions, customer voice, competitor gap analysis, organic feed opportunities.

At Adcrate we do retros every month. This is now the V1 of every retro, delivered into my Slack for approval. If I push back with "you overindex on ROAS, I care more about the ads that get the most spend", the brain remembers for next time.

Use case 2: Static ads delivered to Slack every Monday

This is my favourite one so far.

Prompt: "Search my customer reviews and Facebook ad comments. Find the emotionally loaded phrases our customers are using. Cluster them into distinct angles. Then use the Higgsfield MCP to make static ads from these."

I now have this on a routine: 25 statics every Monday morning, delivered into our Slack channel, using real reviews and comments.

Use case 3: Ingesting client feedback from Slack

Prompt: "Look through my ‘X’ Slack channel. Ingest all the feedback the client has ever given. Bake it into your context going forward."

Every future query I run for that client already has all the client's feedback, preferences, and pushback loaded. The brain gets smarter with every message that comes into Slack.

You can do the same thing for Notion (every concept you've ever produced), Google Calendar (every client meeting), or any tool you've connected.

Here are some additional ways I’ve used this recently if you wants some more inspo:

Tip: if you want to turn a workflow into a routine, enter this prompt:

"Turn this into a routine and send this to me on Slack every Monday morning."

The Parker Brain sets a bunch of these up for you automatically. Refreshing context, generating ad ideas, running gap analyses, all self-updating.

Setting up Parker properly

The outputs are only as good as your Parker setup on the back end. Make sure you have:

  • Customer reviews connected

  • Post-purchase surveys connected

  • Competitor/inspo brands followed

Then you’ll get access to everything in Parker including AI tagging, ad comments, TikTok and all that other good stuff pulling through into Claude code - which’ll make your outputs even stronger.

Try this for yourself

If you want more info on how to install your Parker Brain in Claude Code, read this document, it includes all the links mentioned above.

If you want to watch the full walkthrough, here's the YouTube video.

There's genuinely no going back once you have a brain set up. I'm excited to see the use cases you guys come up with, there's probably a bunch of things I haven't tried yet.

Hit reply and show me what you've built. I'd love to see the workflows.

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

Miss an issue? Catch up on past editions here.

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

See you next week,

Alex

P.S. If you're spending $100k+ monthly on paid social and want help implementing ugly ad strategies like these, my team at Adcrate is accepting applications to work with us. Check out our case studies at adcrate.co.