Marketplaces are the hardest business model I know of, which is exactly why I chose one. The easy models are easy because they are crowded; the hard ones, when they work, are defensible because almost nobody gets the early sequencing right. Gravity is a three-sided marketplace, and the third side is the part people ask about most. This is the reasoning behind the design, written plainly enough to be useful whether or not you ever build one.

It is the architecture story behind the complete marketplace guide, told from the founder's seat.

The three sides

Start with who is on the platform and what they get. Users describe a task and run an expert-built agent that completes it in about a minute, paying only for what they use. Builders are the experts who create those agents; they publish on Gravity and earn a share of every run, with no cost to list. Creators are the third side: people with an audience who refer users or builders to the platform and earn on the activity they bring. Each side makes the others more valuable, which is the whole point of a marketplace. More users make building worthwhile; more builders make the catalog worth using; more creators bring both. The full mechanics of how the splits compare across models are laid out in the marketplace splits comparison.

Why three and not two

The default marketplace is two-sided: supply and demand, builders and users. I added a third side deliberately, because the thing that kills most new marketplaces is not a shortage of supply or demand in the abstract, it is the cost of getting the first of each to show up at the same time. Distribution is the hard problem. Creators are a structural answer to it: rather than spending all of the platform's growth budget on advertising into a cold market, Gravity shares the upside with people who already have the trust and the audience to bring users and builders in.

That makes the third side a growth engine wired into the economics, not a marketing line item. A creator who brings a paying user is rewarded out of the value that user creates, so the platform's customer acquisition scales with its own success instead of fighting against a fixed ad budget. This is the same instinct that made bootstrapped economics attractive to me: align the cost of growth with the revenue from growth, and you do not need someone else's money to buy your way to scale.

The cold-start problem

Every marketplace faces the chicken-and-egg question: users will not come without agents, builders will not build without users. The literature on this is unanimous that you cannot grow both sides evenly from zero; you have to pick a side to seed and a wedge to start narrow (NFX, 2024). My read, reinforced by what the early waitlist shows, is that demand for agents arrives first, because people already know which chores they hate. So the seeding strategy is to make a small set of high-demand agents genuinely excellent, prove the 60-second promise on them, and use that proof to give builders a reason to publish and creators a reason to refer.

The wedge matters as much as the side. A marketplace that tries to be everything to everyone at launch has no gravity. One that does a few things undeniably well gives every side a concrete reason to participate, and that concreteness is what overcomes the cold start. Breadth comes later, after the first loop is actually spinning.

The economics that hold it together

Incentives are the structure. Builders earn 20 percent of every run their agent completes, paid as profit, with no upfront cost to publish. Creators earn 10 percent on referrals, and that 10 percent is split, 5 percent funded by the builder and 5 percent funded by Gravity. That split is not an accounting detail; it is the alignment mechanism. The builder funds half because a referral brings them a paying user, and the platform funds half because a referral grows the whole marketplace. When the cost of growth is shared by everyone who benefits from it, everyone has a reason to make it work. The unit-level picture of how this nets out for a builder is in the builder unit economics, and the broader earning model in how to monetize agents.

Keeping the sides in balance

The recurring danger in a multi-sided marketplace is imbalance. Too many agents chasing too few users starves builders and they leave. Too many users facing too few quality agents disappoints demand and they do not come back. Three sides multiply the ways this can go wrong, which is the real cost of the third side and the reason most platforms stay two-sided. The discipline is to treat growth as a sequencing problem, not a volume problem: grow the side that is currently the binding constraint, watch the ratios, and resist the temptation to celebrate growth on a side that is already ahead. A marketplace is healthy when no side is waiting on another, and getting there is a matter of patience and order, the same way describing an outcome only works when the system underneath it is actually ready to deliver.

FAQ

What is a three-sided marketplace?
A marketplace with three participant groups whose value compounds: users who run agents, builders who publish and earn per run, and creators who refer and earn on referrals. Each side makes the others more valuable.
Why three sides instead of two?
Users and builders are the core two. Creators are the third because distribution is the hardest part of a new marketplace, and paying people with audiences to bring users and builders attacks the cold-start problem directly.
How does Gravity solve the cold-start problem?
By seeding demand first, making a small set of agents genuinely excellent, and using the creator side to accelerate distribution. Demand arrives before supply, so the early work is proving value on a narrow wedge.
How do builders and creators earn?
Builders earn 20 percent of every run as profit, with no cost to publish. Creators earn 10 percent on referrals, split as 5 percent from the builder and 5 percent from Gravity. Both are paid on actual usage.
What is the hardest part?
Keeping all three sides in balance while none is self-sustaining. Too many agents with too few users starves builders; too many users with too few quality agents disappoints demand. Sequencing matters more than volume.
Why is the creator share funded by both builder and Gravity?
Splitting the 10 percent aligns everyone with growth. The builder funds half because referrals bring paying users; Gravity funds half because referrals grow the marketplace. Shared cost means shared incentive.

Sources