Centa

Inspiration

What it does How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for Centa Inspiration

AI tools are being built far beyond the markets that most software directories map well. Centa was built for Central Asia, Eastern Europe, the Caucasus, and nearby regions, where buyers need tools that fit local languages, currencies, workflows, and trust expectations, and where local vendors need a credible way to be discovered without paying for rank.

What it does

Centa is a multilingual AI marketplace for discovering, comparing, and evaluating regional AI tools. It combines a curated catalogue, regional search filters, company profiles, side-by-side comparisons, verified company claims, moderated reviews, transparent rankings, vendor analytics, and admin workflows for trust and quality control.

The product is deliberately designed so rank cannot be bought. Sponsored placement and subscription workflows are separate from organic ranking signals. Rankings are computed from verified evidence and published with an audit trail, while commercial workflows remain outside the scoring path.

How we built it

The app is built with Next.js, React, TypeScript, PostgreSQL, Drizzle ORM, Firebase Authentication/Admin, Vercel, Tailwind CSS, and Zod. The repo includes multilingual UI copy, database migrations, seed data, ranking recomputation logic, moderation workflows, vendor/admin surfaces, billing and commercial request flows, and verification scripts for the private beta foundation.

AI-assisted development was used to design flows, stress-test ranking and billing logic, prepare multilingual product copy, review operational risk, and create launch materials. Human review remains central for vendor verification, moderation decisions, and ranking policy. Challenges we ran into

The hardest part was designing trust into the product instead of treating it as marketing copy. Centa has to distinguish demo data from real traction, keep paid placement away from rankings, label unverified listings honestly, and make moderation decisions auditable. That meant building a stricter data model and more operational checks than a simple directory would need.

Accomplishments that we're proud of

We built the private-beta foundation for a regional AI marketplace: catalogue discovery, search, comparison, ranking infrastructure, verification and moderation flows, vendor/admin workflows, payment/commercial paths, and product videos. We are especially proud that the ranking engine has clear eligibility rules and that the app labels demo records instead of pretending seed data is traction.

What we learned

Marketplaces for AI tools need more than listings. Buyers need explainable trust signals, and vendors need a discovery channel where evidence matters more than ad spend. We also learned that regional fit has to be a first-class product dimension: language, currency, country support, local context, and verification all matter.

What's next for Centa

The next milestone is converting the demo catalogue into a live private beta with real claimed companies, real verified reviews, and a repeatable vendor onboarding workflow. From there, Centa can expand into richer market intelligence, launch support, procurement shortlists, and verified demand signals for AI vendors in the region.

Built With

• ai
• drizzle-orm
• firebase
• marketplace
• next.js
• postgresql
• react
• tailwind-css
• typescript
• vercel
• zod