A proposed low-fee marketplace model

There are two traditions of marketplace economics. The app-store tradition takes 30% and justifies it by owning the customer: the distribution, the billing relationship, the review gate, the audience. The cloud-marketplace tradition takes a few points and justifies it differently — the platform already earns its living from the infrastructure underneath, so the marketplace exists to make the ecosystem rich, not to tax it.

Fibric’s proposed model follows the second tradition. The figures below are illustrative planning assumptions. Public publishing, checkout, payouts, negotiated offers, and contractual terms are not available today.

The schedule, plainly

The intended direction is to lower the platform share as a publisher builds its own distribution. Exact thresholds and obligations would need to be finalized and published before this becomes an offer.

The proposed 3% model

The 3% schedule is a commercial design direction, not a currently transacting marketplace contract. The catalog has 48 listings, with 5 live in BearScope and 43 in managed early access. Public purchases, third-party publishing, payout terms, and negotiated volume tiers are not available today.

This is the same logic that leads the big clouds to charge low single digits on marketplace transactions. They understood that a marketplace's job is to reduce the friction of the ecosystem transacting, not to extract rent from it. A high take rate is a signal that the platform has stopped believing its core product earns the relationship.

What this buys the ecosystem

Economics shape what publishers may choose to build. If Fibric opens public publishing, the proposed low take rate is intended to fund support and hardening rather than make the storefront the primary rent collector. Actual fees would require published contractual terms.

A future low fee would not establish connector quality. Listings would still need maturity labels, conformance tests, security review, support ownership, and deployment-specific verification before customers could rely on them.

An alignment you can audit

Low take rates only work if the platform's costs stay low, and ours are engineered to. Fibric runs on a deliberately thin kernel with swappable seams, so the marketplace does not have to subsidize a heavy infrastructure bill. The economics of the fee schedule and the economics of the architecture are the same decision, made twice.

Before public publishing opens, the fee schedule, payout timing, refunds, taxes, support obligations, and any category exceptions must be published as contractual terms. Until then, 97/3 is an indicative product direction only.

Keep reading: A connector in an afternoon · Per-seat pricing is wrong for operational AI