![](data:image/svg+xml,<svg display="block" role="presentation" viewBox="0 0 22 22" xmlns="http://www.w3.org/2000/svg"><path d="M 0 0 L 0 19.127 L 8.709 22 L 8.709 5.347 C 8.709 0 0 0 0 0 Z M 5.532 19.142 L 3.177 18.371 L 3.177 2.631 L 3.592 2.759 L 4.224 2.953 C 5 3.192 5.532 3.934 5.532 4.776 Z M 18.741 6.199 L 18.741 0 C 18.741 0 10.031 0 10.031 5.347 L 10.031 22 L 18.74 19.127 L 18.74 8.753 L 14.529 10.058 L 14.529 13.606 L 15.564 13.306 L 15.564 18.371 L 13.208 19.142 L 13.208 4.776 C 13.208 3.934 13.741 3.192 14.517 2.953 L 15.564 2.631 L 15.564 7.251 L 18.74 6.199 Z" fill="rgb(0, 0, 0)" height="22px" id="iRPdBku8W" transform="translate(1.63 0)" width="18.740741374796762px"/></svg>)
Tulsy is a marketplace where companies publish their AI tools and users discover them through a smart search that cross-references tags, titles and descriptions. The platform serves three roles, each with its own flows: end users who browse and review tools, companies who list and monetise their products through an in-platform subscription, and the Tulsy AI operator who curates the catalogue. The design challenge was to shape a first MVP that carries only the essential features needed to enter and test the market.
End users register for free, search and explore tools, leave preset reviews and comments and build a list of favourites. Companies self-register, upload and manage one or more products separately, pay a per-tool subscription directly in the platform, monitor performance reporting such as clicks and visits, and can request a consultation from the Tulsy AI team to promote a product. The admin reviews product submissions, accepts them, rejects them or asks for changes, adds new tools directly and reads high-level statistics on the searches users run.
Scoping an MVP across three interdependent roles
The hard part of Tulsy wasn't any single screen. It was deciding what a three-sided marketplace can ship first when each role depends on the other two. End users need enough tools to make search worth it, companies need enough demand to justify listing, and the Tulsy AI operator needs enough submissions to curate a real catalogue. The MVP scope, made of free browsing, tag-based search, tool detail pages, preset reviews and a favourites list, is the minimum set that keeps all three loops alive at once, not just a feature list.
Company and admin flows sized to the same constraint
Every role was mapped end to end so the build team had a clear picture before development started. Companies get a self-service area to register, manage multiple products independently and subscribe per tool, with performance reporting on their listings. The Tulsy AI operator's admin view, where they accept, reject, or request changes on submissions, plus check high-level search statistics, was kept deliberately lean. It's quality control for a first launch, not a full moderation system.
