← All tech signals
Creator economy

The Prompt Marketplace

The well-made instruction ships as a packaged skill on app-store-style shelves, with 238,180 counted in one study, yet who owns one, and who pays, is unsettled.

What is it?

A well-made prompt or workflow turns into a packaged good: a skill or plug-in, listed on a shelf that resembles an app store. One study counted 238,180 such skills across three platforms and GitHub, proof supply exists at scale. Demand is harder to read. Whether anyone pays, or keeps using one once it goes stale, is the open part.

Why now?

Once a workflow reliably gets a model to perform, packaging it follows: companies want tailored workflows and skills for their own staff, and stores form to hand them out. The shelf just doesn’t hold its value once the model underneath it changes.

Who’s building it?

Marketplace platforms and sellers build the listings, but model providers decide whether prompts keep working.

Prompt marketplacesModel providersEnterprise plug-in storesPrompt engineersAgent-store curators
The full read
Why now, who is building it, where the market goes, and what stays unsettled.
Get weekly →
01 · The market
Who’s in it
Builders — platforms and sellers packaging prompts, workflows and agents.
Buyers — individuals wanting a shortcut, and companies deploying pre-built workflows instead of writing their own.
Gatekeepers — model providers and app stores, able to delist a listing.
What’s at stake
A single prompt’s value decays as models absorb the trick — the moat is the store, not the sentence.
Whether platforms ever charge for skills, or the model absorbs them.
Copying a proven prompt costs nothing once seen.
What’s unsettled
Whether prompt IP is defensible at all, once seen and copied.
How fast a listed prompt goes stale as models update.
Whether value settles with sellers, the marketplace, or the model provider itself.
02 · How the demand matureda progression, not a timeline
Curiosity
Early users browse for novelty — cheap, low-commitment, easy to abandon after one try.
Time saved
A working prompt beats trial and error, so users return.
Bundling
Sellers package prompts into skills, workflows and agents, worth more than a single line of text.
Platform capture
Stores absorb the category into app-store-style catalogues, some of them private to one company.
03 · Where the market goesthe forks, not a prediction
→Durable asset or disposable trick — decided by how fast model updates outrun what a purchased prompt was tuned for.
→Model providers could fold prompt stores into their own platforms, squeezing out independent marketplaces.
→Prompts bundle upward into paid workflows and agents, where the packaging carries more value than the wording.
→Legal ownership of a prompt stays unsettled until disputes force a clear answer.
Everything here turns on shelf life: whether a bought prompt still works next model version, or whether the buyer is really renting one good guess.
04 · The Programmable Value deskadjacent markets, moving now
The Model Marketplace
Voices and models get licensed too.
ElevenLabs launched a 28-voice celebrity licensing marketplace.
The Decoder
Programmable Money
The rails for paying per prompt are law.
US GENIUS Act became federal stablecoin law in 2025.
The White House
The Consent Economy
Ownership claims extend from likeness to instructions.
Denmark proposed copyright law over citizens’ own likeness.
Tech Policy Press
05 · The Wire
01
Agent skills now sit in app-store-style marketplaces
arXiv (preprint, Holzbauer et al.) ↗
02
Anthropic adds private plug-in marketplaces for company agents
TechCrunch ↗
The panel on this signal
They agreed on
–Real supply has formed, with app-store-style marketplaces built to distribute packaged skills.
–Value sits with the platform, not any single prompt.
–Prompts are structurally short-lived, tied to whichever model version they were tuned for.
The creator-economy analyst disagreed on
–A count of listings measures initial catalogue size, not verified repeat purchasing; without reorder or retention data, the figure is as consistent with a one-time curiosity wave as with an established buying pattern.
Know more about the method →
✦How this signal was built
01 · Panel
A Delphi panel of 4 AI analysts, each reasoning independently.
→
02 · Debate
They challenge and revise each other’s reads over two rounds.
→
03 · Evidence
Every claim is tied to a cited, linked source on The Wire.
→
04 · Synthesis
A moderator writes it into one read, disagreement and all.