Something big is happening to marketing
Two weeks ago, Matt Shumer wrote an essay about AI that went viral. It was grounded in how he could see AI impacting his day to day as an engineer, but also provided ways for people to think about how to prepare for the inevitable new era.
Reading his essay, I began to think about how there’s something big happening to the field of marketing. The conversation about AI and marketing is stuck between two useless poles: panic pieces about robots taking our jobs, and vendor listicles about which AI tool to add to your workflow. Neither captures what’s *actually* happening.
AI is collapsing the distance between strategy and output, and when that distance disappears, brand and taste are all that’s left. As someone who has been in tech marketing for 15 years — having built and led marketing teams at Coinbase, Block, Blockchain.com, and now at the Stellar Development Foundation — here’s what it means for marketers and what you should do about it.
The layers are collapsing
For decades, here’s how medium and large marketing teams have worked: A CMO sets strategy. VPs and Directors build plans and campaigns. Managers write creative briefs or comms docs. Individual contributors execute. Each layer exists because translating “what we need to accomplish” into “the thing that ships” required human judgment and curation at every step.
AI is compressing those layers. Not hypothetically. On my team. Right now.
Take a thought leadership piece for a CEO. The old way: weeks of interviews, multiple drafts going back and forth, and once the copy is locked another few days to build the posting plan, creative briefs for design, social copy by channel, and then publish. Call it two to three weeks from kickoff to live.
With Claude Code and a deep library of contextual inputs — voice guidelines, past work, strategic docs, audience context — that first draft now takes minutes. The team reviews it live and asynchronously for a few hours. The full content plan and design comes together in a day or two. AI gets us 80% of the way there in minutes. We spend our time on the 20% that matters: sharpening the strategy, elevating the creative, and figuring out the best way to distribute it.
That’s not a productivity hack. That’s a structural change to how marketing organizations work. The people who set direction, define quality, and make judgment calls are about to be supercharged. The people whose value was executing tasks that senior leaders couldn’t get to are in a precarious spot. Entry-level marketing roles — the coordinator jobs, the “write three versions of this email” jobs, the “pull together a competitive analysis deck” jobs — those are compressing fast.
As I think about marketers on my team, there’s no doubt that each and every role is rapidly evolving to something radically different than the same role a year ago. Not everyone has the desire or capacity to entirely change how they work, but those who embrace the tools will accelerate their impact and ability to advance in their career.
The responsibility of any marketing leader right now is to be direct about what’s changing, invest in helping people adapt, and not pretend that “everything will be fine” when the ground is moving.
Production is free. Mostly slop.
Marketing isn’t rocket science, but is remarkably hard to do well. Strategy is hard. Taste is rare. Put simply, what makes it hard isn’t production.
When production costs approach zero the gap between good marketing and bad marketing widens. The team with sharp strategy and genuine creative instinct produces high quality storytelling that resonates. The team without it produces a thousand nothings.
The scarcest skill in marketing is no longer execution — it’s direction. Knowing which projects to pursue, what message to put in front of which person, and holding a quality bar high enough that what ships actually matters. The best leaders have always done this with teams of people. Now every marketer needs to do it with teams of agents. We’re all becoming managers.
Brand is THE moat
Here’s what I keep returning to: when building becomes fast and free, and every individual armed with Claude Code can make any product or piece of content on demand, what is the moat? How will companies and products and teams differentiate?
Brand.
Not distribution. Not channels. Not tactics. Brand — the accumulated weight of every promise kept. When your competitors have access to the same AI tools, the same production speed, the same ability to test and iterate, the thing that separates you is whether people trust you. Whether they choose you when they have a hundred other options.
Great marketing has always been the same thing: understand the person you’re trying to reach, understand what makes your product meaningful to them, and connect the two. AI changes the speed at which you can make that connection. It doesn’t change whether the connection is real.
I’ve believed this my whole career: every interaction you have with a customer or potential customer is either keeping a promise or breaking one. The accumulation of kept promises is what builds trust. Trust is what builds a brand. This includes product — every feature, every experience, every interaction is a promise kept or broken. AI doesn’t change that equation. If anything, it raises the stakes because AI makes it trivial to break promises at scale. To overproduce slop.
The brands that win in the next decade will be the ones that feel human in a world of machine-generated sameness. Real people behind the work, with a real point of view, making real decisions about what they stand for.
People go to basketball games to see Steph Curry push the limits of human ability, not to watch a robot sink three-pointers. The same principle is coming for brands: when everything around you is AI-generated, the things that feel irreducibly human will command a premium.
What to do, right now
This week, take the next project you’re working on and try to get as much of it done with just you and AI. A Q2 strategy doc. A landing page. A positioning doc on something highly technical. An email campaign. Whatever’s on your plate, start there. Don’t stay in a single chat window. Set up a project folder, feed it as much context as you can — your voice guidelines, past work, strategic docs, audience context — and see how far you get.
Then focus on making the output better. Update its memory. Build a markdown file with your voice, your strategy, your previously shipped work. The AI gets better the more you teach it about your voice, your strategy, and your standards. This is a compounding investment — the tenth project you do this way will be dramatically better than the first.
Once you’ve got the basics down, start building purpose-built tools for the parts of your job that eat your time. An agent that scours X every morning and gives you a digest of what customers and competitors are saying. An agent that consolidates your meeting transcripts and gives you coaching feedback or a prioritized list of action items. An agent that monitors competitor campaigns and flags what’s changed. These aren’t science fiction — they’re afternoon projects with the right tools.
You don’t need to be technical to do any of this. My stack started with Cursor as my IDE, a Claude Code Max plan, and help from friends and YouTube. Before building anything, I go back and forth with Claude on a brief. Before working on a written piece, I draft a plan with Claude. Before building an agent or a website, I make a product requirements doc and ask Claude to help me improve it before I start. Whatever you can imagine would help you with your job, you can build an MVP and go from there.
It used to be you could only have two of the three: fast, cheap, or good. Now you can have all three with the right tooling and the right direction.
Design your career for what’s coming
A wise friend once told me to not think of your current job as the end: the final job, the one where you’ll make life-changing returns, or the one that you’re meant to garner accolades. Instead, think of what job this job will set you up for.
When I was a head of product marketing, I designed each role to prepare me for the next one. I still think about the gaps that will make me a better CMO, leader, and subject matter expert.
Right now, designing for the next role means one thing above all else: get fluent with AI as fast as you can. Learn to direct agents. Set strategy that accounts for what these tools make possible. Compound what you learn by building on each project, each workflow, each experiment.
Careers are long and the pace of technological change is breakneck. Extend your time horizon for success, but be urgent about building the skills that will matter when the next opportunity shows up, whether that’s in a month or in years.
We’re heading into a K-shaped economy. Those who learn to work with AI will accelerate. Those who don’t will fall behind. The divergence is already starting and it will widen fast. AI isn’t going to take your job. Someone great at using AI is going to take your job.
Learn, build, ship
It sounds smart for a moment to be a pessimist, dismissing AI or insisting your function is special and immune. But pessimism doesn’t build anything. The world is built by optimists who prove the pessimists wrong.
Marketing isn’t going away. It’s going to matter more than ever. When anyone can build anything, the ability to tell the right story to the right person becomes the most valuable skill in business. The question is whether you’ll be the one telling those stories, or the one reading someone else’s.
Be the author, not the reader.