Everyone’s talking about taste. I’ve been selling it for 20 years.
When I first wrote about this in February, “taste” was having a moment.
Paul Graham had just lit up tech Twitter arguing good taste is real and measurable. Founder events were treating it as the unofficial theme of the night. A wave of Substack essays were getting serious traction on the topic.
Since then, the conversation has only accelerated.
In late February, Sam Altman, a day before announcing OpenAI’s $110 billion funding round, went on record saying the most valuable thing a non-technical candidate could bring to his company was taste.
“We believe the best research teams are built through context, taste and a real feel for where the field is headed next,” he told the world.
Get the best of Startup Daily straight to your inbox
Want to know the latest in startup news? Subscribe to our daily news and analysis coverage on what’s happening to ANZ startups, investors and the broader ecosystem. And best of all, it’s FREE!
By continuing, you agree to our Terms & Conditions and Privacy Policy.
Cloudflare’s CTO declared it the differentiator in engineering for 2026. Fortune ran it as a jobs story. Every week now, a new piece lands making some version of the same argument.
I’ve been watching all of this with a mix of excitement and déjà vu.
Not because I’m smarter or earlier than anyone in the room, but because I’ve spent the last two decades building and scaling businesses that were, at their core, taste businesses. We just didn’t always call them that.
And I think the current conversation, as energising as it is, is only getting to half the insight. Yes, taste matters. It always has.
The harder question, the one almost nobody is answering, is: what does the infrastructure for taste look like in the AI era?
What we really sold
In the 2010s, I was the co-founder and CEO of Sound Alliance, which grew into one of Australia’s most influential independent music and culture media companies.
We built and ran brands like inthemix, FasterLouder, and Mess+Noise. We partnered with and represented international platforms like Last.fm, Pitchfork, and Hype Machine in the Australian market. It was a moment.
On paper, we were a media company. We built audiences around high-value niches and sold advertising around them.
But if you looked under the hood, the actual asset we were monetising was taste.
inthemix didn’t just cover electronic music and clubs. It filtered electronic music through a particular lens, a community with shared values and a collective sense of what was good, what was interesting, and what was worth your Friday night. FasterLouder did the same for live music. Each brand had a point of view. Each was, in its own way, a taste engine.
The same was true of our partners. Pitchfork wasn’t just a music review site. It was a judgment machine. The 10-point scale was really a taste signal, a way of saying “trust us, this matters.”
Hype Machine aggregated music blogs, but the real value was the taste layer it placed on top. Last.fm mapped your listening habits, but its magic was in connecting you to people with similar taste.
None of these brands advertised themselves as “taste companies.” But that’s exactly what they were.
And if you zoom out, you’ll see this pattern repeating across every era of media and culture. Condé Nast sold taste on glossy paper.
Broadsheet has built an entire business around telling Australians where to eat and drink. When a creator like Andy Cooks builds a massive following around food, they’re not really selling content. They’re selling their taste, their ability to sort through infinite options and say “this one, not that one.”
The business model was never really advertising, sponsored posts, or events. The underlying asset, the thing audiences were actually paying for with their attention and loyalty, was the confidence that someone with good judgment had done the filtering for them.
The abundance principle
Here’s the pattern that keeps repeating, and it’s the key to understanding why taste is having its moment right now.
Every major technology wave increases the volume of stuff in the world. And every time that happens, the value of taste goes up, not down.
Napster and broadband gave everyone access to essentially infinite music. Did that kill the need for curation? It created an explosion of curators. Pitchfork, inthemix, Hype Machine, music blogs, all grew directly out of that abundance.
Social media platforms gave everyone a publishing tool. The volume of content became unmanageable, so algorithmic feeds emerged to handle the flood. But something interesting happened alongside those algorithms: the rise of the creator economy. Individual voices built massive followings precisely because they offered something the platforms’ own recommendation engines couldn’t.
Taste. Point of view. A sense that someone real was making choices on your behalf.
The creator economy is, when you strip away the brand deals, a taste economy. People follow creators because they trust their judgment about what to cook, what to wear, where to travel, what to read.
And then there’s Substack. The explosive growth of the newsletter platform is itself evidence of the taste thesis. People are choosing to pay individual writers directly, not for information they can’t get elsewhere, but for the curation, perspective, and judgment those writers bring.
In a world overflowing with free content, people are literally paying for taste.
Now we’re entering the most dramatic abundance shift in history. Generative AI has made content creation effectively free. Anyone can produce a blog post, a video script, a marketing campaign, a song, an image. The cost of creation has collapsed to near zero.
Every generation assumes that abundance will democratise quality. It never does. Abundance democratises access and concentrates the value of judgment. The people and brands who can sort the signal from the noise, who can tell you not just what exists but what’s worth your time, become exponentially more valuable.
As The Atlantic rather nicely put it last year: the ability to discern which of infinite AI-generated variations is actually meaningful may prove to be the rarest and most valuable skill of all.
Four eras of taste infrastructure
If you step back far enough, you can trace four distinct eras of how taste has been organised and delivered. Each one changed the infrastructure. None of them changed the underlying human need.
The gatekeeper era. Before the internet, taste was bundled with access. The Rolling Stone critic didn’t just have good taste; they had access to albums before anyone else.
The New York Times restaurant reviewer had the best table, the platform and the authority. You trusted their taste partly because they were the only ones in the room. The gatekeeper was the taste filter, and the two things were inseparable.
The digital curator era. The internet unbundled taste from access. Suddenly anyone could hear the music, read the news, find the restaurant. Access wasn’t the scarce resource anymore. So taste had to stand on its own merits. This is the era I lived through building Sound Alliance and later Junkee Media. Brands like Pitchfork, Hype Machine, and inthemix succeeded because they had genuine editorial taste and the community trust to back it up.
The algorithmic era. The platforms decided they could automate taste. Spotify’s Discover Weekly, TikTok’s For You Page, Instagram’s Explore tab. These systems replaced human judgment with behavioural data.
They asked, in effect: “What do people like you tend to engage with?” It worked brilliantly for volume. But something was lost in the translation.
Algorithms optimise for engagement, not for taste. They play to the middle of the bell curve. And it’s worth noting that the creators who thrived on these platforms did so despite the algorithms, not because of them. The best creators on TikTok, YouTube, and Instagram succeeded by asserting a point of view strong enough to cut through the algorithmic mush. The platforms gave them distribution. Their taste gave them an audience.
The AI era. And now here we are. AI can synthesise, summarise, and recommend at infinite scale. It can process more information than any human curator ever could. But it’s drawing from the same consensus pool as everything else.
Ask ChatGPT for restaurant recommendations in Melbourne and you’ll get a competent list, confidently delivered, that looks remarkably similar to what Google, TripAdvisor, and every other platform would surface.
Ask for music recommendations and you’ll get a list that maps roughly to Spotify’s algorithmic output, repackaged in friendlier prose.
The lesson across all four eras is the same. The infrastructure changes. The human need for trusted taste doesn’t.
People don’t want more options. They want better filters. They want someone they trust to say: “Pay attention to this. Skip that. This is the one.”
The AI taste gap
This brings us to what I think is the most important and least discussed dynamic in the current AI conversation.
AI has simultaneously made taste more important and harder to find.
There’s a divide I keep coming back to: facts versus opinions. AI is getting very good at facts. Yes, it still makes mistakes, it hallucinates, it gets things wrong. But these are engineering problems being solved with every model update.
The trajectory is clear: AI will soon be extraordinarily reliable on factual questions. “This restaurant is open until 10pm.” That’s a problem compute can solve.
But opinions? That’s a human problem.
“This is the place that will change how you think about pasta.” That statement requires judgment, experience, and a willingness to have a point of view. It can’t be derived from averaging a million data points. It comes from someone who has eaten a lot of pasta, thought carefully about what makes it good, and is willing to stake their reputation on a recommendation.
No amount of engineering is going to solve for that, because it isn’t a bug. It’s a fundamentally different kind of knowledge.
What makes a good restaurant is a matter of taste. Photo: AdobeStock
The current AI ecosystem treats tastemakers as training data. Their opinions, reviews, recommendations, and cultural judgments get scraped, blended, and served back to users without attribution, without compensation, and without the brand and trust that made the opinion valuable in the first place.
I watched this exact movie play out in the music industry 15 years ago. Napster didn’t just disrupt distribution. It disintermediated the people who had spent years building audiences and trust.
Musicians were told to think of piracy as “exposure.” The economics collapsed, and a generation of artists had to rebuild from scratch on platforms that captured most of the value.
The same thing is happening now, but to curators, reviewers, publishers, creators, and tastemakers of every kind. Their expertise is being absorbed into AI systems that flatten it into consensus.
The very people whose judgment made AI’s training data valuable are being cut out of the value chain entirely.
What comes next
I don’t think this story has to end the same way.
Every era of taste infrastructure eventually found a way to connect people who have genuine judgment with the audiences who want it. The format changed. The underlying exchange didn’t.
The question for this era is whether we build AI systems that empower tastemakers or replace them. Whether the technology honours the human judgment at the core of great curation, or continues to strip-mine it for training data.
Think about what that means in practice. If you’re a publisher like Broadsheet, you’ve spent 15 years building a team with deep knowledge of Australia’s food and culture scenes. That expertise has enormous value, but right now it’s being scraped and summarised by AI platforms that flatten it into generic recommendation lists.
What if instead, your editorial taste could power an AI experience that carried your brand, your voice, your standards? Not a chatbot pretending to be Broadsheet, but an AI tool that lets Broadsheet’s judgment reach people in new contexts, at new scales, while keeping the brand and the economics intact.
Or if you’re a creator who’s built a following around your taste in travel, or wine, or design. Your audience trusts you because you’ve earned it over years of consistent, opinionated recommendations.
Today, an AI assistant can summarise your entire body of work and serve it up to someone who’s never heard of you, never followed you, and will never subscribe.
What if instead, your taste became the engine behind a discovery experience your audience could access on your terms, carrying your name and your perspective?
This is what we’re building at lokol.
The provocation
Everyone’s saying taste is a moat. I agree. But a moat is only useful if you build something behind it.
Here’s the thing the “taste is the new skill” conversation keeps missing: taste already drives a $250 billion creator economy. Billions of purchase decisions, every day, are influenced by the recommendation of a trusted voice.
Someone’s favourite food creator, a publisher whose travel coverage they’ve followed for years, a reviewer whose judgment they’ve come to rely on. That signal is already doing enormous commercial work.
Source: AdobeStock
The problem is where it stops. It travels as far as the content does, then disappears the moment someone opens a new tab, asks an AI, or heads to a retailer.
The provenance of the recommendation, the context, the trust, the person who made the call, all of that gets stripped out. You get the purchase without the why.
What we’re building at lokol is infrastructure that makes that signal portable, attributable, and available everywhere people actually buy. Not a new social platform, not another aggregator, not a generic AI assistant that has digested everyone’s opinions and attributed them to no one. A layer that keeps the tastemaker’s voice intact as it travels into the moments where decisions get made.
The tastemakers have always mattered. The audience has always wanted them.
The only variable, across every era I’ve lived through, is whether the infrastructure gives them a seat at the table or leaves them outside the room.
Right now, they’re outside. That’s the problem we’re fixing.
- Neil Ackland is cofounder & CEO of lokol, an AI startup, currently in closed beta, that turns trusted tastemakers into discovery engines their audiences can talk to. Join the waitlist at lokol.ai or follow them on Substack at lokol.substack.com
