SheTaxis

Problem

In 2014 roughly one percent of New York yellow cab drivers were women. For a rider who wanted a woman behind the wheel — late at night, or simply by preference — that was not a choice the market offered. Founder Stella Mateo started from her own daughters asking her to stay on the phone during late rides because they did not feel comfortable sharing a closed space with a stranger.

The service she wanted to build had two sides that had to launch together and stay in balance. Riders would not use it without drivers available, and drivers would not sign on without fares. Neither side tolerates a rough first version: a car service that cannot find you a car is not a product with room to improve, it is one someone deletes.

Approach

Realm 8 delivered the whole product surface — the brand, the website, the native iOS and Android applications, and the server infrastructure and API behind them.

Two apps, two audiences, one system. Riders request a car and confirm at the point of request that there is a woman in the party. Drivers receive and accept work. Between them sits the dispatch API, which is where a car service is actually won or lost: it has to know where drivers are, decide who gets an incoming request, and do it fast enough that a person standing on a kerb does not give up and hail something else.

Dispatch ran on an algorithm of our own rather than simple nearest-driver assignment. It weighed proximity together with driver rating, so the closest car was not automatically the one sent. That matters more here than on a general car service: the product's promise is that the ride feels safe, and a rating is the only signal the system holds about whether previous riders found it so. Routing purely by distance would have optimised for pickup time and quietly ignored the thing people were choosing the service for.

Building the brand and the product together was deliberate rather than incidental. The service's whole proposition is that a rider trusts it on sight, so the identity had to arrive at the same moment as the app rather than be fitted afterwards.

Outcome

The service launched in September 2014 and reached more than 100,000 users.

It was covered by the New York Times, The New Yorker, The Economist and CNBC among others — unusually broad press for a launch of that size, and a fair measure of how sharply the idea landed. Those articles are linked alongside this page.