Building a car-sharing platform for São Paulo

Summary

Kypcar is a peer-to-peer car-sharing platform in São Paulo: car owners rent out their idle vehicle to other people, and renters save up to 40% versus traditional rental agencies. I was brought on as a freelancer to design the entire product from scratch, solo, from the first flow sketch to the final screen.

Challenge

A peer-to-peer marketplace has two sides to convince at once: the person looking for a car needs to trust the host and the process before paying for days that haven't happened yet; the person offering the car needs to feel the platform protects their asset. Kypcar's answer runs through AI-powered identity verification (facial recognition and driver's license validation), full insurance included, and professional cleaning between rentals. All of it needed to be visible and clear in the experience, not just sitting in the terms of service.

With one person on design and a freelance timeline, the product also had to launch whole: onboarding, search, filters, host profile, scheduling, checkout, post-booking and chat. There was no shipping with half the flow.

Objective

Design the platform's complete MVP, from onboarding to booking, creating an experience trustworthy enough for an owner to hand their car to a stranger, and for a renter to choose a private car over a rental agency.

Constraints

A freelance project with a single designer and a tight timeline, with no previous product to build on. With no connected hardware in scope, the key handoff happens in person: the meetup between strangers had to be part of the experience. And trust had to be built through design, without the established brand of international players.

Market analysis

Before designing the first screen, I mapped the competitive landscape: how traditional rental agencies, digital-first rental agencies, and peer-to-peer marketplaces each solve (or don't) the same mobility problem.

Localiza and Movida have scale, but depend on owned fleets and physical presence; Turbi digitized the rental experience but kept the same fleet model. On the other side, Turo and Getaround proved peer-to-peer in other markets, each solving the friction differently (Turo with insurance and reputation, Getaround with hardware that eliminates the key handoff). The model simply had no proven equivalent in Brazil. That gap grounded several product decisions: if nobody here had solved trust in a peer-to-peer marketplace yet, it had to be the center of the experience, not a detail buried in the terms of service.

Experience analysis

After the market, I dug into the experience: how do established apps design the contact between two strangers? I analyzed the interface and flows of products where the transaction depends on that encounter, and what each one solves well became repertoire for Kypcar.

The common thread: the interface doesn't ask for trust, it shows evidence: identity, reputation, and rules visible at the right moment. In Kypcar, I combined all three repertoires: Airbnb's listing, Uber's encounter, and DogHero's track record.

Interactive prototype

This prototype is fully navigable: search for a car, pick your dates, complete a booking, and chat with the host.

Approach

1. 01

   Mapped the end-to-end journey before designing a single screen: from first opening the app to the conversation with the host after the booking is confirmed.

2. 02

   Built a dark, high-contrast visual system (purple and lime) that reads as secure and tech-forward without looking like a generic fintech app.

3. 03

   Designed both sides of the marketplace's trust equation: AI identity verification and a verified-host badge on the supply side; included insurance, professional cleaning, and a visible cancellation policy on the booking side.

4. 04

   Prototyped the entire flow as a navigable product, not standalone screens, to validate the end-to-end experience before any production code was written.

Design decisions