An indie perfumer sits in a cramped studio, surrounded by blotter paper and half-finished accords. Three AI tools are open on her laptop. Every single one suggests "vanilla amber musk" for a gourmand fragrance. She's tried tweaking the prompts a hundred ways. The AI doesn't know that gourmand in indie perfumery means something entirely different than what mainstream fragrance houses produce. It doesn't know that IFRA regulations have nuked half the classic materials. It doesn't know the inside jokes about "Yankee Candle syndrome."
Meanwhile, a dungeon master building a political intrigue campaign—low magic, no prophecies—watches helplessly as every AI generator insists on inserting dragons and a Chosen One.
Here's the signal: vertical AI just became a $5 billion valuation play in legal. Harvey AI hit that number in June 2025 after growing from $10 million to $100 million ARR in about 36 months. Enterprise gen-AI spending exploded 6x in 2024, from $2.3 billion to $13.8 billion. The application layer—where AI actually touches workflows—grew nearly 8x.
The pattern is clear. The opportunity is open. The creative economy is next.
The Shift: From AI Slop to Taste Engines
Generic AI tools are powerful. They're also taste-blind.
Ask ChatGPT about obscure fragrance materials and you'll get confident-sounding hallucinations. Ask Midjourney to create something in the style of "early 2000s Flash animation aesthetic" and watch it miss the entire point. The models were trained on the internet's center of gravity—mainstream, accessible, obvious. They know what "most people" like. They cannot speak the dialect of weird micro-cultures.
This matters because creative subcultures don't want "most people" results. They want specificity. They want someone who's read the same obscure zines, watched the same forgotten video essays, argued about the same controversies. They want an AI that actually understands their scene—what's cliché, what's underrated, what's sacred.
Future Snoops launched MUSE in late 2025 with exactly this thesis. They didn't build another generic creative AI. They built what they call a "creative intelligence engine" trained on their proprietary 25+ year archive of trend research, image libraries, and cultural analysis. The tool draws exclusively from their curated database—not the open internet. They paired the launch with a whitepaper called The Creatorship Era, arguing that AI should amplify human vision instead of flattening everything into the same beige feed.
Meanwhile, the backlash against AI slop is becoming institutionalized. Books By People launched an "Organic Literature" certification in late 2025, partnering with UK publishers to verify and stamp human-written books. Their founding partners include Galley Beggar Press, Bluemoose Books, and Snowbooks. The first certified title—Telenovela by Gonzalo C. Garcia—carries the certification like a badge of honor. They're not alone: Art Recognition uses machine learning to verify visual art authenticity, Ircam Amplify detects AI-generated music, and CREDO 23 certifies films made with zero AI involvement.
This is structural demand.
The Menlo Ventures 2024 State of Generative AI report found that while foundation models still dominate headlines, the application layer—AI tools embedded in specific workflows—grew from $600 million to $4.6 billion in enterprise spend in a single year. Businesses aren't paying for raw intelligence anymore. They're paying for intelligence that speaks their language.
The global vertical AI market is valued at roughly $12.9 billion in 2024 and projected to reach $115.4 billion by 2034—a 24.5% CAGR. Harvey AI isn't an outlier. It's a proof point. Domain-specific AI that understands the workflows, terminology, and quality standards of a particular field commands premium pricing, earns fierce loyalty, and builds nearly unassailable moats.
Now apply that pattern to creative niches—not billion-dollar law firms, but passionate micro-communities that lack proper infrastructure.
Where the Money Actually Lives
The creator economy is massive and growing. Various estimates put it between $150 billion and $200 billion in 2024, with projections ranging toward $700 billion or more by the early 2030s. More than 200 million people worldwide identify as content creators.
But here's the uncomfortable reality: only a small minority of creators earn serious income. Surveys suggest just 4–7% make over $100K annually. The pyramid is brutal.
What separates top earners from everyone else? Part of it is audience, sure. But increasingly, it's tooling sophistication. According to industry surveys, 84–86% of creators used AI tools in 2024. Top earners use AI more frequently and invest significantly more in premium subscriptions than average creators.
The problem: they're mostly using the same generic horizontal tools—ChatGPT, Midjourney, Canva. These tools help with production speed, but they don't help with creative differentiation. They don't understand the specific craft.
That's the gap a Micro-Muse fills.
Consider the existing beachheads:
Industrial perfumery. Givaudan's Carto system is an AI-powered fragrance formulation tool that gives professional perfumers an interactive scent map connected to a robot mixer. It's trained on Givaudan's proprietary database of molecules and consumer preference data. It's literal AI-augmented R&D for fragrance creation.
Vault-only access.
Join founders who spot opportunities ahead of the crowd. Actionable insights. Zero fluff.
“Intelligent, bold, minus the pretense.”
“Like discovering the cheat codes of the startup world.”
“SH is off-Broadway for founders — weird, sharp, and ahead of the curve.”
