AI Workflows That Actually Work for Marketers

My screen looks like I’m hacking

Three terminal windows, a browser with 14 tabs, Claude running in the background on a 40-page competitor analysis. If you walked past my desk you’d think I was breaking into something.

I’m doing marketing.

Most “AI for marketing” advice falls into two buckets: too basic (“use ChatGPT for subject lines”) or too abstract (“AI will transform your strategy”). Neither tells you what an actual Tuesday looks like when AI is woven into how you work.

This is what mine looks like.

Research that used to take a day

Last week I had 10 tabs of competitor analysis open. Annual reports, product pages, case studies, pricing pages, three different review sites. The old version of this task is a full day of reading, note-taking, and trying to see patterns.

I dumped it all into Claude. 20 minutes later I had a positioning map, a messaging comparison, and a list of gaps none of the competitors were addressing.

Here’s what I actually typed:

I'm researching [client's] competitive landscape.

Here are my sources:
[Paste everything: articles, reports, notes, screenshots]

Synthesize into:
1. How each competitor positions themselves (one sentence)
2. What benefits they emphasize
3. What objections they address
4. What's missing that we could own
5. Messaging patterns across all of them

Be direct. Skip caveats. If sources conflict, note it.

The output isn’t final. It’s a starting point I’d have spent hours getting to manually. I spend my time refining the analysis instead of building it from scratch.

Turning one piece into ten

A client did a 45-minute webinar. Good content, 200 attendees, then it sat on their YouTube channel doing nothing.

I took the transcript, fed it to Claude, and in an afternoon we had: a long-form blog post, a 5-part LinkedIn series, three email snippets, and a list of pull quotes for social. The quality wasn’t publish-ready (it never is), but it was 80% there. The editing took an hour instead of two days of writing from scratch.

The pattern is always the same: AI gets you to a solid draft fast, you add the 20% that makes it sound like a human wrote it.

Data that talks back

I paste campaign data (raw CSVs, messy tables, whatever I have) and ask questions I’d normally spend an hour noodling on alone.

“What patterns do you see? What anomalies stand out? Give me three possible explanations for why this metric dropped 40% in week 3.”

It doesn’t replace your judgment. It replaces the lonely staring-at-spreadsheets phase. AI generates hypotheses; you decide which ones are worth pursuing.

What AI is genuinely bad at

This is the section most AI content skips, and it’s where credibility lives.

Strategy. AI can inform a decision, not make one. It doesn’t know your board is nervous about the budget, your best salesperson just quit, or that your biggest customer is about to churn. Context is everything, and AI doesn’t have yours.

Brand voice. It can mimic. It cannot maintain. Every piece still needs a human ear. Does this sound like us, or does it sound like a machine trying to sound like us?

Original creative. AI recombines existing patterns. It’s a brilliant remix artist and a terrible original songwriter. Breakthrough ideas still need a human brain doing something AI can’t simulate: having a weird thought in the shower.

Knowing what matters. AI treats all information as equally important. You know that one sentence in the quarterly report is the one that changes everything. AI doesn’t.

My setup

Claude for thinking. Long conversations, research synthesis, anything that needs reasoning over large amounts of text.

Local models for client data. Anything confidential stays off the cloud.

Terminal for everything. I run most of my AI workflows from the command line, not chat interfaces. It’s faster, it’s scriptable, and yes, it looks like I’m hacking.

I wrote more about this setup in How I Use AI.

Getting started

Pick one workflow. Use it for two weeks. Then add another. That’s it.

Your first prompts will be mediocre. Save the ones that work, refine them over time, and build a personal library. Within a month you’ll wonder how you worked without it.

Want to walk through your specific use case? Book an hour and we’ll figure out where AI actually fits your workflow.