Enterprise after enterprise, startup after startup, pitch after pitch, we keep hearing the same idea:

AI is magic.

It’ll automate everything, accelerate every workflow, transform industries // change our lives // make us better parents, lovers and programmers overnight.

This…incantatory language has become the lingua franca of AI marketing, to the point that every company from Canva to Adobe seem to believe that invoking speed, magic, and disruption should be enough to secure customers.

And if it’s not, it’s the customers’ collective fault for not being bullish enough.

But if you talk to the actual buyers - the enterprises who sign contracts, the operations managers whose thankless job it’ll be to integrate these tools, the engineers who inherit the systems, the users waving their new iPhones and Pixels - what you’ll inevitably hear is disappointment, fatigue, or outright skepticism.

The mismatch between marketing promises and customer needs has created a world where the loudest companies win attention in the short term and the entire industry loses credibility in the long term.

And in the process, serious builders are being drowned out by an avalanche of breathless hype that will never (ever) sell a finished product.

Arthur C. Clarke's third law - "Any sufficiently advanced technology is indistinguishable from magic" - is repeated so often it might as well be stitched into the carpet of every venture capitalist's office. But Clarke was describing perception, not strategy. For him, magic was the effect of encountering an unfamiliar but real technology. AI marketing keeps trying to invert the formula: sell magic first, hope the technology will catch up later.

The dot-com boom was rife with companies promising frictionless commerce and instant community online, long before broadband penetration made such experiences feasible. But the AI cycle feels different. Machine learning does accomplish extraordinary feats. Large language models can draft code, synthesise documents, and pass professional exams. Image models can create visuals that would have required entire design teams a decade ago. But the experience of actually adopting these systems is far less enchanting.

Customers want reliability, integration, and compliance. They want predictable costs and clear lines of accountability. They want guarantees that the tool will work as described every day, not just in the demo.

Magic is unreliable by definition.

In the 1950s and 1960s, the early mainframe era was full of utopian promises. General Electric declared that the "electronic brain" would replace office drudgery. IBM marketing films suggested the computer was an almost divine problem-solver. But in practice, adoption was slow and painful. Custom software had to be written for each application. Trained operators were scarce. Bugs and downtime were constant.

The reality didn’t match the spectacle. The companies that endured were those whose “magic” was bundled with service contracts, training, integration, and support. IBM built a dynasty on cultivating trust. "Nobody ever got fired for buying IBM" became a mantra precisely because IBM was seen as the opposite of magic: dependable, staid, boring even.

Contrast that with modern AI marketing.

Instead of reliability, vendors pitch acceleration.

Instead of trust, they promise transformation.

And instead of boring competence, they reach for wonder.

But what customers want is precisely the opposite: a product that works.

Radical, no?

"It'll make everything faster.”

Faster document processing. Faster analysis. Faster customer service.

But speed is not a standalone benefit.

Faster is valuable only when the output is accurate, trustworthy, and integrated into existing workflows.

A race car without brakes is impressive in velocity, terrifying in practice.

A chatbot that produces answers in half a second but hallucinates legal citations is worse than useless in a law firm. A computer vision model that processes x-rays faster but mislabels cancers is not a productivity tool, it’s a medical // legal // moral liability.

The uncomfortable truth is that AI marketing is written for venture capitalists rather than customers. Investors want to hear about world-changing impact, billion-dollar markets, and paradigm shifts.

Buyers are a simpler // more honest crowd.

They want to know if the product integrates with Salesforce, if it complies with GDPR, if the pricing model fits their budget cycle. But marketing departments keep mistaking the language of fundraising for the language of sales.

A CIO who deploys a hyped AI product only to discover it fails basic compliance checks is unlikely to trust the next pitch deck full of promises about exponential speed. And because these failures accumulate across industries, the entire field begins to sound like the boy who cried wolf.

In Goethe's Faust the protagonist (desperate for knowledge) turns to magic after exhausting the limits of science. The attraction of shortcuts and power without process, is as old as literature itself. Modern AI marketing often feels like it is written in that spirit: skip the boring part, leap straight to the miracle.

But Faust pays for his temporary marvels with his soul.

In pragmatic terms, companies who chase marvels without substance are winning early attention but losing credibility, trust, and (ultimately // inevitably) customers.

If calling AI "magic" and "fast" is a dead end, what’s the alternative?

The companies that survived past hype cycles - the IBMs, the Intels, the Oracles - sold boring-as-paint virtues: reliability, standards, service, and integration. They promised to make existing processes more dependable, not to replace them with miracles.

A better language for AI marketing:

These messages won’t go viral on Twitter.

They’re unlikely to thrill investors.

But they speak directly to the anxieties of buyers, and once the first bubble bursts, that’s precisely the point.

The Roman aqueducts were feats of engineering so durable that many still stand today. They weren’t sold to the Roman public as marvels of speed; they were reliable and proven systems used to deliver water every day.

The companies that survive the first AI bubble will be the ones who build that same trust.

They’ll realize that "magic" is an effect customers get excited about after a technology has proven itself, it’s not a label you get to apply in advance.

They’ll market less like Vegas magicians and more like engineers.

AI marketing today is broken because it confuses spectacle with substance. It reaches for wonder when customers want reassurance. It sells speed without reliability, magic without accountability.

The opportunity is in making AI boring enough to trust.

Which means the most radical marketing line an AI company could adopt today would be the simplest: here’s what it does, here’s why, here’s how and here’s the damn proof.