Everybody’s speaking about style. I’ve been promoting it for 20 years.
Once I first wrote about this in February, “style” was having a second.
Paul Graham had simply lit up tech Twitter arguing good style is actual and measurable. Founder occasions had been treating it because the unofficial theme of the evening. A wave of Substack essays had been getting severe traction on the subject.
Since then, the dialog has solely accelerated.
In late February, Sam Altman, a day earlier than saying OpenAI’s $110 billion funding spherical, went on report saying essentially the most beneficial factor a non-technical candidate may convey to his firm was style.
“We consider one of the best analysis groups are constructed by means of context, style and an actual really feel for the place the sphere is headed subsequent,” he instructed the world.
Cloudflare’s CTO declared it the differentiator in engineering for 2026. Fortune ran it as a jobs story. Each week now, a brand new piece lands making some model of the identical argument.
I’ve been watching all of this with a mixture of pleasure and déjà vu.
Not as a result of I’m smarter or sooner than anybody within the room, however as a result of I’ve spent the final twenty years constructing and scaling companies that had been, at their core, style companies. We simply didn’t at all times name them that.
And I believe the present dialog, as energising as it’s, is barely attending to half the perception. Sure, style issues. It at all times has.
The tougher query, the one nearly no one is answering, is: what does the infrastructure for style appear to be within the AI period?
What we actually bought
Within the 2010s, I used to be the co-founder and CEO of Sound Alliance, which grew into one in all Australia’s most influential impartial music and tradition media firms.
We constructed and ran manufacturers like inthemix, FasterLouder, and Mess+Noise. We partnered with and represented worldwide platforms like Last.fm, Pitchfork, and Hype Machine within the Australian market. It was a second.
On paper, we had been a media firm. We constructed audiences round high-value niches and bought promoting round them.
However for those who regarded underneath the hood, the precise asset we had been monetising was style.
inthemix didn’t simply cowl digital music and golf equipment. It filtered digital music by means of a selected lens, a neighborhood with shared values and a collective sense of what was good, what was fascinating, and what was price your Friday evening. FasterLouder did the identical for reside music. Every model had a perspective. Every was, in its personal approach, a style engine.
The identical was true of our companions. Pitchfork wasn’t only a music overview web site. It was a judgment machine. The ten-point scale was actually a style sign, a approach of claiming “belief us, this issues.”
Hype Machine aggregated music blogs, however the actual worth was the style layer it positioned on prime. Last.fm mapped your listening habits, however its magic was in connecting you to individuals with related style.
None of those manufacturers marketed themselves as “style firms.” However that’s precisely what they had been.
And for those who zoom out, you’ll see this sample repeating throughout each period of media and tradition. Condé Nast bought style on shiny paper.
Broadsheet has constructed a complete enterprise round telling Australians the place to eat and drink. When a creator like Andy Cooks builds a large following round meals, they’re probably not promoting content material. They’re promoting their style, their means to type by means of infinite choices and say “this one, not that one.”
The enterprise mannequin was by no means actually promoting, sponsored posts, or occasions. The underlying asset, the factor audiences had been really paying for with their consideration and loyalty, was the boldness that somebody with logic had finished the filtering for them.
The abundance precept
Right here’s the sample that retains repeating, and it’s the important thing to understanding why style is having its second proper now.
Each main expertise wave will increase the quantity of stuff on this planet. And each time that occurs, the worth of style goes up, not down.
Napster and broadband gave everybody entry to primarily infinite music. Did that kill the necessity for curation? It created an explosion of curators. Pitchfork, inthemix, Hype Machine, music blogs, all grew immediately out of that abundance.
Social media platforms gave everybody a publishing software. The amount of content material turned unmanageable, so algorithmic feeds emerged to deal with the flood. However one thing fascinating occurred alongside these algorithms: the rise of the creator economic system. Particular person voices constructed huge followings exactly as a result of they provided one thing the platforms’ personal advice engines couldn’t.
Style. Viewpoint. A way that somebody actual was making selections in your behalf.
The creator economic system is, whenever you strip away the model offers, a style economic system. Individuals comply with creators as a result of they belief their judgment about what to cook dinner, what to put on, the place to journey, what to learn.
After which there’s Substack. The explosive development of the publication platform is itself proof of the style thesis. Persons are selecting to pay particular person writers immediately, not for info they’ll’t get elsewhere, however for the curation, perspective, and judgment these writers convey.
In a world overflowing with free content material, persons are actually paying for style.
Now we’re coming into essentially the most dramatic abundance shift in historical past. Generative AI has made content material creation successfully free. Anybody can produce a weblog put up, a video script, a advertising and marketing marketing campaign, a track, a picture. The price of creation has collapsed to close zero.
Each era assumes that abundance will democratise high quality. It by no means does. Abundance democratises entry and concentrates the worth of judgment. The individuals and types who can type the sign from the noise, who can inform you not simply what exists however what’s price your time, turn into exponentially extra beneficial.
As The Atlantic fairly properly put it final yr: the flexibility to discern which of infinite AI-generated variations is definitely significant could show to be the rarest and Most worthy talent of all.
4 eras of style infrastructure
Should you step again far sufficient, you possibly can hint 4 distinct eras of how style has been organised and delivered. Each modified the infrastructure. None of them modified the underlying human want.
The gatekeeper period. Earlier than the web, style was bundled with entry. The Rolling Stone critic didn’t simply have good style; they’d entry to albums earlier than anybody else.
The New York Occasions restaurant reviewer had one of the best desk, the platform and the authority. You trusted their style partly as a result of they had been the one ones within the room. The gatekeeper was the style filter, and the 2 issues had been inseparable.
The digital curator period. The web unbundled style from entry. Out of the blue anybody may hear the music, learn the information, discover the restaurant. Entry wasn’t the scarce useful resource anymore. So style needed to stand by itself deserves. That is the period I lived by means of constructing Sound Alliance and later Junkee Media. Manufacturers like Pitchfork, Hype Machine, and inthemix succeeded as a result of they’d real editorial style and the neighborhood belief to again it up.
The algorithmic period. The platforms determined they might automate style. Spotify’s Uncover Weekly, TikTok’s For You Web page, Instagram’s Discover tab. These techniques changed human judgment with behavioural knowledge.
They requested, in impact: “What do individuals such as you have a tendency to have interaction with?” It labored brilliantly for quantity. However one thing was misplaced within the translation.
Algorithms optimise for engagement, not for style. They play to the center of the bell curve. And it’s price noting that the creators who thrived on these platforms did so regardless of the algorithms, not due to them. The very best creators on TikTok, YouTube, and Instagram succeeded by asserting a perspective sturdy sufficient to chop by means of the algorithmic mush. The platforms gave them distribution. Their style gave them an viewers.
The AI period. And now right here we’re. AI can synthesise, summarise, and suggest at infinite scale. It may course of extra info than any human curator ever may. But it surely’s drawing from the identical consensus pool as every part else.
Ask ChatGPT for restaurant suggestions in Melbourne and also you’ll get a reliable record, confidently delivered, that appears remarkably just like what Google, TripAdvisor, and each different platform would floor.
Ask for music suggestions and also you’ll get an inventory that maps roughly to Spotify’s algorithmic output, repackaged in friendlier prose.
The lesson throughout all 4 eras is identical. The infrastructure adjustments. The human want for trusted style doesn’t.
Individuals don’t need extra choices. They need higher filters. They need somebody they belief to say: “Take note of this. Skip that. That is the one.”
The AI style hole
This brings us to what I believe is an important and least mentioned dynamic within the present AI dialog.
AI has concurrently made style extra necessary and tougher to seek out.
There’s a divide I preserve coming again to: information versus opinions. AI is getting excellent at information. Sure, it nonetheless makes errors, it hallucinates, it will get issues incorrect. However these are engineering issues being solved with each mannequin replace.
The trajectory is obvious: AI will quickly be terribly dependable on factual questions. “This restaurant is open till 10pm.” That’s an issue compute can remedy.
However opinions? That’s a human downside.
“That is the place that can change how you consider pasta.” That assertion requires judgment, expertise, and a willingness to have a perspective. It may’t be derived from averaging one million knowledge factors. It comes from somebody who has eaten lots of pasta, thought rigorously about what makes it good, and is prepared to stake their status on a advice.
No quantity of engineering goes to unravel for that, as a result of it isn’t a bug. It’s a basically totally different form of data.
The present AI ecosystem treats tastemakers as coaching knowledge. Their opinions, opinions, suggestions, and cultural judgments get scraped, blended, and served again to customers with out attribution, with out compensation, and with out the model and belief that made the opinion beneficial within the first place.
I watched this actual film play out within the music trade 15 years in the past. Napster didn’t simply disrupt distribution. It disintermediated the individuals who had spent years constructing audiences and belief.
Musicians had been instructed to consider piracy as “publicity.” The economics collapsed, and a era of artists needed to rebuild from scratch on platforms that captured many of the worth.
The identical factor is going on now, however to curators, reviewers, publishers, creators, and tastemakers of each form. Their experience is being absorbed into AI techniques that flatten it into consensus.
The very individuals whose judgment made AI’s coaching knowledge beneficial are being minimize out of the worth chain completely.
What comes subsequent
I don’t assume this story has to finish the identical approach.
Each period of style infrastructure finally discovered a solution to join individuals who have real judgment with the audiences who need it. The format modified. The underlying alternate didn’t.
The query for this period is whether or not we construct AI techniques that empower tastemakers or substitute them. Whether or not the expertise honours the human judgment on the core of nice curation, or continues to strip-mine it for coaching knowledge.
Take into consideration what which means in follow. Should you’re a writer like Broadsheet, you’ve spent 15 years constructing a staff with deep data of Australia’s meals and tradition scenes. That experience has monumental worth, however proper now it’s being scraped and summarised by AI platforms that flatten it into generic advice lists.
What if as an alternative, your editorial style may energy an AI expertise that carried your model, your voice, your requirements? Not a chatbot pretending to be Broadsheet, however an AI software that lets Broadsheet’s judgment attain individuals in new contexts, at new scales, whereas protecting the model and the economics intact.
Or for those who’re a creator who’s constructed a following round your style in journey, or wine, or design. Your viewers trusts you since you’ve earned it over years of constant, opinionated suggestions.
At the moment, an AI assistant can summarise your complete physique of labor and serve it as much as somebody who’s by no means heard of you, by no means adopted you, and can by no means subscribe.
What if as an alternative, your style turned the engine behind a discovery expertise your viewers may entry in your phrases, carrying your title and your perspective?
That is what we’re constructing at lokol.
The provocation
Everybody’s saying style is a moat. I agree. However a moat is barely helpful for those who construct one thing behind it.
Right here’s the factor the “style is the brand new talent” dialog retains lacking: style already drives a $250 billion creator economic system. Billions of buy selections, day-after-day, are influenced by the advice of a trusted voice.
Somebody’s favorite meals creator, a writer whose journey protection they’ve adopted for years, a reviewer whose judgment they’ve come to depend on. That sign is already doing monumental industrial work.
The issue is the place it stops. It travels so far as the content material does, then disappears the second somebody opens a brand new tab, asks an AI, or heads to a retailer.
The provenance of the advice, the context, the belief, the one that made the decision, all of that will get stripped out. You get the acquisition with out the why.
What we’re constructing at lokol is infrastructure that makes that sign moveable, attributable, and out there all over the place individuals really purchase. Not a brand new social platform, not one other aggregator, not a generic AI assistant that has digested everybody’s opinions and attributed them to nobody. A layer that retains the tastemaker’s voice intact because it travels into the moments the place selections get made.
The tastemakers have at all times mattered. The viewers has at all times wished them.
The one variable, throughout each period I’ve lived by means of, is whether or not the infrastructure offers them a seat on the desk or leaves them exterior the room.
Proper now, they’re exterior. That’s the issue we’re fixing.
- Neil Ackland is cofounder & CEO of lokol, an AI startup, at present in closed beta, that turns trusted tastemakers into discovery engines their audiences can discuss to. Be a part of the waitlist at lokol.ai or comply with them on Substack at lokol.substack.com
