The future of creative marketing - why what's happening in coding will happen to creative

Code is cheap. Creative is next.

The future of creative production isn't being written by creatives. It's being previewed, right now, in engineering.

The world of coding agents is anywhere between 6 and 24 months ahead of the world of creative agents - and what's happened to software development is a remarkably clear blueprint for what's coming for creative next.

Code got cheap

For the first time in history, code is cheap.

Software used to be expensive because writing it was slow, skilled work. A feature could take a team of engineers weeks or months to spec, build, test and ship. Every line was a human bottleneck - constrained by headcount, hiring pipelines and the number of hours in the day.

That constraint is gone. Coding agents can now produce hundreds of lines of code a minute. They can be run in parallel and in the background. A single developer can ship in a weekend what used to take a team of five a quarter.

For a while, many senior developers convinced themselves this was doomed. "Vibe coding" would produce spaghetti code too unwieldy to maintain - eventually it would break and be unfixable.

That narrative no longer holds. Models and their "harnesses" (the way models are connected to the tools and systems they need to work in your codebase) have improved at such a rate that not only is the code they produce less spaghetti-like, but any bad code they create today can safely be left to be fixed by a better model tomorrow.

There is no going back.

The bottleneck moved

Now that code is cheap, the perennial bottleneck of software development - writing the stuff - has disappeared. Except, of course, it hasn't really disappeared. As with any process, resolving one bottleneck shifts the rate-limiting factor to another stage. And for software, two new bottlenecks have emerged either side of writing code: deciding what to build and reviewing what's been built.

Reviewing code has long been part of an engineering manager's job spec. But the sheer quantity of code now being produced has made reviewing it a near full-time job. Myriad new tools, agents and companies are springing up to help. Cursor, just one AI coding firm, have in a matter of months built an 8-figure business out of their Bugbot - an AI that automatically reviews new code for bugs before it's pushed to production. Over time, the output of AI agents will almost entirely be reviewed and approved by other AI agents.

Steering is a more elusive problem. In a world where you can build anything, what should you build? For startups, the idea of racing to feature parity with incumbents is intoxicating. For bigger players, unlocking a backlog of customer requests is pretty appealing. But the teams that thrive aren't just the ones building fastest - they're the ones making the best decisions about what to build.

As Karri Saarinen, Co-Founder of Linear, writes:

Design was never about what the button is or does, or which medium you work in. It was and is about finding the right problem, the right intent, the right vision. The feature you design and build today should be a considered step toward that vision.

Replace "design" with "creative" and you have the thesis of what comes next.

Creative's turn

The parallels are clear.

For the longest time, creative production's rate-limiting factor has been creative resource. Copywriters were needed for email copy, designers for banner mockups, photographers for product shots, editors to clip up social posts. Anyone who has worked agency-side knows the negotiating, grovelling and bribery required to squeeze a request through studio behind a PM's back.

As foundation models and the creative production tools built on top of them improve, this resource bottleneck will erode - just as it has in engineering. And just as in engineering, two new bottlenecks will take its place. In creative, steering and reviewing are better described as Briefing and QC.

QC

In a world with many more assets being produced, checking those assets for brand safety, compliance, and quality becomes a much bigger job. Expect the progression to match code reviews closely: new tools and agents will crop up to help teams check work before it goes live.

But here's where creative diverges from engineering. Code review, ultimately, has a degree of objectivity to it - code either compiles and passes its tests or it doesn't. Creative QC is subjective. Does a banner feel on-brand? Is the tone of that email right for this audience? Is this product shot going to land with legal?

This means creative QC tools can't just check for errors. They need to encode brand guidelines, tone of voice, compliance rules, and visual standards as structured inputs - not vibes. The companies that crack this will build something genuinely valuable. The ones that don't will drown in AI-generated slop that technically "works" but slowly erodes brand trust.

Briefing

This is the big one.

There are two sides to the briefing bottleneck: deciding what to make and giving agents what they need to make it well.

The first is the strategic problem. In a world where production is near-free, the question of what to create and why becomes the entire game. Bad briefs have always tended towards producing bad work. The difference now is that bad briefs produce bad work at scale, instantly. The production cost drops to near zero but the cost of poor creative direction - wasted attention, off-brand assets flooding your channels, confused customers - actually goes up.

The second is the ingredients problem. Creative agents, like coding agents, are only as good as their harness. Brand guidelines, tone of voice, product information, audience context, campaign strategy - if you don't give the agent structured, high-quality inputs, you get generic output. Garbage in, garbage out. But at speed and scale.

This is perhaps the most important takeaway from what's happened in engineering: the brief is becoming infrastructure. Not a formality you rush through before the "real work" starts, but the primary lever of creative quality in an AI-native workflow. What should we be making? And what do we need to make it?

The companies that win the AI-native creative era won't be the ones with the best generators - they'll be the ones with the best briefs, highest quality ingredients, and QC that can keep up.