I don't know why more ecom founders aren't using AI to write their ad copy, landing pages, and email flows. Here's the exact workflow I use to build creatives across all my businesses:

1. Download all your customer reviews, your website, your competitors' reviews, and your top-performing advertorials. Everything goes into one place. The more raw input you give it, the better the output gets on the other side. Codex and Claude Code can build a trust pilot scraper within one prompt.

2. Run deep research with a proper prompt (e.g., “create me everything I need to know about my customer to do future marketing”). Remember that your prompts probably need to be twice as long as you think in this day and age.

3. You now have a customer doc - this is your source of truth for all future copy, so treat it like one and keep coming back to it.

4. Take that customer doc and give it the structure of the advertorial you want to write with a proper prompt, including a proper mechanism and solution mechanism.

5. Don't give it one massive prompt. My YT script prompting cycle is 14 prompts long because AI doesn't handle large blobs of data well. Give it 12 angles first, select the best ones, then prompt it with questions about what you already know about your customer, feed the answers back, and build from there.

6. Loop the results back in. The top creative in your ad account is the new input, and that performance data goes straight into the next round of prompts.

The problem with ecom used to be time. Finding products took weeks, building stores took weeks, shooting content took weeks.

But AI broke that barrier. You can now build 10 advertorials in a day, and that volume negates luck.

TLDR: Use more AI.

A public operating system for how I work — Metrics, Analysis, Action. How I use AI, how to work with me, and how you can write your own portable context so you can move conversations across Claude, ChatGPT, Gemini, and any agent.

I use AI agents live to write articles, publish blog posts, build knowledge bases, and run SEO audits. I share my 4-stage Content Factory, the Task Library, and the magic phrase "according to" that makes AI follow my process.

I'm building an AI marketing team of one.

Here's the thing about being a one-person marketing & sales team at an open-source software company (JobRunr): you don't have time for everything.

So I asked Patrick for help.

Patrick is my AI executive assistant, built on OpenClaw. And this week, he built me a LinkedIn Ads review system. From scratch.

Let me walk you through how we got here.

Step 1: We did the research first.

Before writing a single ad, we analyzed our actual data. Not vibes. Data.

We pulled company visitor data from Scarf and cross-referenced it with HubSpot. IP matching, domain lookups. That gave us a ranked list of which industries and companies are actually visiting our Pro and Pricing pages.

Finance. Insurance. Government. Not what I would have guessed.

We built company lists from that intelligence and matched them to LinkedIn audiences.

Step 2: We built a messaging framework.

Patrick analyzed all our existing customer data. Hubspot data, email conversations, intake questionnaires, support questions, ads,... Then we used that data to fill in a creative strategy map and come up with 3 angles per target-audience.


Step 3: The visuals.

I showed Patrick our existing ad visuals. He reverse-engineered the visual language and wrote image generation prompts matching our brand. Generated with Gemini. I could approve, reject, download, tweak in Photoshop, re-upload. All from a review tool he built.

Step 4: The review tool.

Patrick built a web app where I review everything:
→ Ad copy with live LinkedIn preview
→ Image variants (approve, download, replace, regenerate)
→ Comments and feedback
→ One-click approve to push live

No SaaS subscription. Just me and my agent.

The proof? One ad Patrick created by analyzing my best performers and generating a variant is now my highest CTR ad at 2.65%. AI-generated. Outperforming all my manual ones.

The difference between "write me an ad" and this? Real context. Your CRM data. Your analytics. Your brand. Your actual performance numbers.

And most importantly: Your taste. You are the creative director who shows the way.

Full stack: OpenClaw, LinkedIn Marketing API, HubSpot, Scarf, Gemini, and a custom review tool. Fully built using Telegram conversations.

Happy to share more details. Or just roast my AI visuals. Either works.

My CMO gave me three weeks to fix a campaign that had a 0.8% click-through rate.

I didn’t hire a consultant. I didn’t run another A/B test. I opened Claude.ai and treated it like the most senior marketing strategist I’d ever worked with.

The campaign hit 4.3% CTR. We added 2,100 leads in 18 days.

This is not a “AI is amazing, go use it” article. This is the actual playbook — the prompts, the frameworks, the mistakes, and the surprising moments when Claude did things no human marketer had thought to suggest.

Been experimenting with AI for marketing over the past year and the biggest shift for me was stopping the one-prompt-one-output approach. Most people open ChatGPT, type something, get something mediocre, and conclude AI is overhyped for marketing. The real unlock is chaining prompts together — using the output of one as the input for the next.

Quick example: I start by having AI interview me about my customer's frustrations (not their demographics, their actual language). Then I use that output to write the ad. Then I run the ad copy through a "devil's advocate" prompt that pokes holes in it before I publish.

The result is copy that sounds like a customer wrote it, not a robot.

I have been doing digital marketing for 12 years and only in the last 18 months has AI genuinely changed how I work at a strategy level, not just a speed level. I am running a live online workshop on March 22 for anyone who wants to see the full workflow in real time.

It is ₹199 and covers four areas: copywriting, competitive research, customer intelligence, and building a repeatable AI system. Recording included. Happy to answer any AI marketing questions in the comments regardless. Let me know if you are interested.

Two years ago, I used AI to write a caption.

Today, I use it to understand what my audience wants before they even tell me.

That shift happened faster than I expected.

AI in marketing isn't really about robots anymore.
It's about something much quieter and more powerful.

It's about knowing which email subject line will actually get opened.
Which ad will make someone stop scrolling.
Which customer is about to leave even before they do.

68% of marketers say AI is most useful for automating routine tasks.
That's true. But I think we're underselling it.

The real unlock isn't speed.
It's relevance.

When a brand speaks to you like it actually knows you,
that's not magic.
That's machine learning working in the background.

But here's what I keep coming back to:

AI can optimize a message.
It can't originate one.

It can tell you when to post.
It can't tell you why you care about what you're posting.

The marketers who will win aren't the ones who fear AI.
They're the ones who stay deeply human while using it well.

Use the tools.
Learn them properly.
But don't outsource your judgment to an algorithm.

Because AI can make you faster.
Only you can make your work matter.

P.S. If you're a marketer still on the fence about AI then you're not behind yet. But the window is closing.

My mom asked me how I use AI at work.

“For everything” wasn’t a very helpful answer. 😅

So I typed up these steps I often go through (this specific list was for creating web pages, but the process is often similar for other deliverables).

Nobody asked except my mom 😉, but here is my process:


WRITING WEB COPY WITH AI

1. Start with my brain: no AI yet
Dig up existing documents on the topic I need to write about, jot down notes that come to mind, review competitor pages, roughly map page structure...

2. Generate a first draft with AI
- Have AI make a “web page copy brief” based on my notes/outline, plus our Domo product positioning doc
- Using that brief, generate a first draft of the page via a Web Copywriter Agent I made
- Go back & forth with the agent to iterate on messaging

3. Switch into research mode with AI
To confirm I’m on the right track, switch gears and use:
- Gemini’s Deep Research Agent for additional market & competitive context
- My own Voice-of-Customer Agent trained on real customer comments

4. Go back and critique the draft with AI
Feed those findings back into the Copy Agent. Ask it to critique the draft against things like market direction, analyst framing, and customer language. Make adjustments accordingly.

5. Critique the draft with humans
Align with PMs and other stakeholders, and refine again.


I find the best outputs come when I alternate between using AI for creation and critique.

I like breaking the project up into several steps, wiping the context between chats, and using my *very own human brain* heavily throughout the process.

It takes more time, but I think it’s worth it. :)


Any other favorite AI tricks I should add to the list? (It’s for my mom, I promise !!)

After 1,000 hours of AI marketing workflows.

This is what I wish I knew from day one.

The robots are here to stay.

And if they don't kill us in our sleep.

we better make good use of them.

This is how I use AI for marketing.

(I'm 51. If I can do it, you can too.)

1. Record every call → Your customers are writing your content. Every objection, every pain point.

2. Store transcripts in one place → Scattered notes are useless. One folder, every conversation. That's your content goldmine.

3. Extract content ideas → Run your transcripts through AI. Ten post ideas in five minutes.

4. Pull ICP pain points for your ads → Cold traffic responds to their own language. Your calls already have it. Your campaigns should too.

5. Find your hooks in their complaints → The thing your clients hate most? That's your best headline.

6. Build a story bank → Every call adds one story. In six months, you never run out of content again.

If you can't beat the robots, join them.

PS: What's your AI automation story?