AI abundance has a coherence problem. Paging product marketing.

“Whoever spends the most money wins!” said no CFO ever.

Actually, that’s possibly not true. The CFOs at Amazon, Anthropic, Google, Microsoft, and OpenAI may very well be whipping up the frenzy to acquire more and more compute capacity. Damn the torpedoes! Although I imagine them more likely waking wide-eyed in the middle of the night murmuring a Talking Heads lyric, “My God! What have I done?!”

But here on Earth, below the gods of AI’s Mount Olympus, treating money spent as a metric to be maxxed, divorced from any measurable return — you know, the “R” in ROI — seems kinda crazy.

Although for a brief moment, it didn’t. The tokenmaxxing movement rewarded employees for spending as much as possible on AI consumption, independent of any other variable. There were even leaderboards, with top spenders earning a badge of honor, while those at the bottom got the hint that they weren’t AI-ing hard enough. “Use more AI!” was the edict. To what end? Shrug.

Thankfully, that moment is passing.

To be fair, it did kickstart AI adoption and learning. Probably worth it. But now that the engine is running and the bills are rolling in, people are snapping back to their senses.

It’s not just tokenmaxxing. AI washing, slapping an AI label on anything that moves, has grown tiresome. Cramming more AI “features” into products and processes that nobody asked for is more likely to annoy than amaze. And AI slop is now regularly being shamed and banished. (To quote Monty Python and The Holy Grail, “And there was much rejoicing.”)

This is not an anti-AI revolt, although that sentiment is certainly out there. It’s simply tempering raw enthusiasm with good business sense. Real innovation is happening with AI in marketing and martech, and you should embrace it. This includes “vibe coding” (in the broadest sense) by marketers, marketing ops, and GTM engineers. But judge it by its impact, not its AI-ness.

Vibes without value are vanity.

Yet at software companies — especially martech companies — marketing’s bigger AI challenge may not be what it can build itself, but what it has to do with everything product and engineering can now build faster.

AI coding — developers wielding Claude Code, Codex, and Cursor to crank out more software, faster — is arguably the leading example of AI delivering real productivity gains. But those gains are measured primarily in the volume and velocity of things shipped.

Now, “shipping” has long been a rallying cry in software development, because shipping was hard. Getting something built and deployed to production was the bottleneck for most ideas. Go-to-market teams were often waiting on product. So “ship, ship, ship” was a mantra to march to, kind of like “ro, ro, ro” was for Norway in the World Cup.

But with AI… shipping is getting a lot easier. That’s good in many ways. But it reshuffles the dynamics of everything around it.

For one, the lower effort required to build things has reduced the sense that ideas need to be carefully considered first. Just build it and see what happens. MVP-palooza! I’m a big fan of experiments, prototypes, and agile iterations to discover and refine customer value. But building isn’t the only cost. And when building becomes fast and cheap — relatively speaking — those other costs emerge as the new constraints.

One of those costs is maintaining what gets built. The rule of thumb is that the initial build is 30% of the engineering work, while post-launch maintenance is 70%. Granted, maintenance is accelerated by AI too. But it’s not all coding. It’s also debating and resolving all the interaction effects and compatibility issues that an expanding and shifting product stirs up. From first-hand experience, I can attest: that opinionated and political human process can be a time suck.

However, development is only the first mile. When it comes to customers of commercial software products, it’s rarely an “if you build it, they will come” Field of Dreams.

Taking a new feature or product to market is a lot of work. Go-to-market teams need to first understand it, then fit it into all the relevant positioning, pricing, marketing content, sales enablement, support documentation, etc. There’s internal training and external communications to be orchestrated. There’s new content to be created and old content to be revised or replaced. There’s legal and compliance reviews. Possible updates to systems ranging from accounting to the CRM, onboarding workflows, the website, the help desk, and a growing array of customer-facing agents.

Sure, the scale of that work depends on the scale of the release. Small features are less work than big ones. But none are zero. Because if a feature ships in the woods and nobody knows about it, does it make a difference? (Answer: no.)

The larger a company’s footprint is, the larger that task becomes. There are more teams, more customers, more touchpoints, more moving parts, more competing and confounding priorities. The ripple effects compound.

All that go-to-market work is in the service of helping people outside the company absorb what’s being released. Yet even with the best GTM, that’s a long last mile to traverse.

Existing customers are busy and used to the way things have been working. Asking them to stop and learn something new — much less change the way they do things — is frankly a big ask. Even when it’s to their benefit. It takes time and persuasion for them to absorb changes into their thinking and their workflow. Their time, attention, and mental bandwidth are scarce resources that approximately a billion other things are competing for. Throwing more and more at them, at an ever faster rate, becomes counterproductive.

Exhibit A: note the slow adoption of many AI features that have been jammed into SaaS products this past year. They may be good features, but their install base can’t absorb them at the rate SaaS investors really wish they would.

(One of the legit risks behind the SaaSpocalypse is that it can often seem easier to customers to adopt new products, to do new things, in new ways, than it is to reconfigure their mental model of existing products in their stack.)

And customers aren’t the only constituency that struggles to make sense of a vendor’s feature fluidity. Prospects need a clear grasp of what a company does, which often requires updating priors set months or years ago. Partners need to know what changes mean for them — new opportunities or new competition? Analysts need to categorize and compare offerings within a somewhat stable framework. And so on for agencies, consultants, investors, media, influencers.

Every new feature, product, agent, edition, bundle, and positioning shift asks all of those people to update their understanding of what the company does and where it fits. As the velocity of product change accelerates, coherence becomes harder to maintain.

This is Martec’s Law applied to shipmaxxing. AI steepens the curve of what product teams can ship, but the curve of what the rest of the organization, its customers, and the market can absorb doesn’t bend at AI speed. The widening gap is the new constraint.

For all the hubbub around GTM engineers and AI-pilled demand generation — apologies, you’re probably as sick of the phrase “AI-pilled” as I am — what the world needs now is love, sweet love really great product marketing. At least if you’re a software company (or a service-as-software company) moving at AI speed.

I’ve long been a champion of marketing ops and martech teams, and I’m delighted to see them in starring roles now, albeit with a mélange of new titles. But at this particular moment, I think product marketing needs the spotlight.

Product marketing can sit in the catbird seat at the crossroads of product, marketing, sales, customers, and the market. Yet in many companies, that strategic position has been squeezed into a supporting function, downstream of the real decisions made by product, growth, and demand gen.

When product teams can create change faster than the rest of the organization and the market can absorb it, somebody needs to orchestrate the whole product-to-market flow. It’s more than communications. It’s bringing broader market context and narrative judgment to decisions about what deserves to ship, when it should reach the market, and how it fits into the larger story. It’s owning the throughline from what gets built to what gets absorbed.

Product marketing can be that somebody, and companies should increase its authority to curate, coordinate, and constrain across the constituencies and touchpoints strained by product abundance.

Product marketers should be the keepers of coherence.

As someone who spends his days reading a lot of martech company emails and websites, I can attest, our industry needs it.

Scott