What it means to be an “AI first” company in a Super App

The Careem app, the pioneer of ride-hailing in the Middle East, uses AI technology with 400 engineers and developers to make people’s lives easier.

Careem, which means “to be generous” in Arabic, started in Dubai as a popular ridesharing app in 2012 and has grown into a full-fledged platform and super app that is spreading across the region. Acquired by Uber for $3 billion in 2020, Careem now offers customers in Dubai 11 services, including ride-sharing, micro-mobility, food and grocery delivery and payments, as well as partner services such as home cleaning and PCR testing. There are 1.3 to 1.4 million captains (drivers) who are registered and working on the app, this number increases by nearly 50,000 every month.

There are 600 million people living in this region from Morocco to Pakistan. Careem believes at least 150 million are addressable in some form. But literally, it affects less than 1% of people on a daily basis. And the other 99% have problems it could fix, but they still can’t access them.

Here is an excerpt from the interview with Selim Turki who is Head of Data and AI at Careem, a mobility company owned by Uber, to discuss the latest trends in advanced analytics and artificial intelligence. The conversation focused on several compelling use cases where Careem is using AI to make a difference in people’s lives. Their teams (400 engineers and developers) are developing the first super application in the Middle East, North Africa and Pakistan region. Interestingly, their hiring strategy is focused on growing a diverse team of data and machine learning scientists to grow their internal experimentation and machine learning platforms. To help their data and AI teams stay abreast of the changes happening in the industry, they have started collaborating with regional academic institutions to solve some of the biggest super-app challenges and to identify new new opportunities for AI innovation. The team contributes also to the open source data communities and offer his work to other AI and ML professionals.

How has the role of AI evolved since the creation of Careem?

We started processing data in real time, using machine learning algorithms and models [ML] to solve some of the key issues in our ridesharing marketplace, including efficiently matching customers and captains, shaping our demand and supply through peak pricing, calculating accurate ETAs for our drivers, and improving our maps and location search functionality. We use several AI techniques depending on the type of service we provide in our awesome app. All these techniques meet three particular needs:

Personalization. Customers have a unique experience with our awesome app based on their preferences and behavior history. For example, we use our understanding of customers’ carpooling patterns to facilitate a one-click widget that lets them choose to drive to their most frequent destinations, such as home, work, or the gym. On the food shopping side, which is usually a more personal experience, personalization includes the cuisines users like, the specific dishes they crave or want to explore, and contextually relevant search content to display. . The more our users engage and transact with us, the better the experience we can dynamically refine and create for them.

AI used for local experience. We serve our customers, drivers and partners better by being local and closer to them. We use AI to take into account prayer times, iftar time during Ramadan and weather conditions to better predict ETA accuracy of when food will be delivered to our customers.

AI used for a safer experience. We use AI to perform targeted facial recognition checks for our drivers with the aim of detecting potential imposters and ensuring that the “captain” driving is the same one who checked in and passed the checks regulations and authorization. On the client side, we use AI to detect genuine or fraudulent registrations or logins and transaction integrity checks to authorize or block super-app transactions.

AI is part of Careem’s decision-making framework. We set quarterly goals to measure and evaluate the use and impact of our ML models on different business flows. We use rigorous statistical methodologies, taking confounding effects into account, to accurately estimate the impact of the model on different areas of the business.

Can you share a recent example of how AI has fundamentally changed the way Careem works with its customers?

Regardless of the digital platform, fraudsters are looking for loopholes to exploit, whether creating accounts with false identities or exploring ways to hijack open accounts. Our team uses advanced AI techniques focused on user identity to detect and prevent losses resulting from fraud. A system we use, called Crazy Wall, uses a RGCN (relational graph convolutional network) to map different data points of a customer’s identity. It also identifies characteristics shared by different identities to detect and block fraudulent patterns in customer or captain activities en masse.

How will AI specifically change the mobility space?

The global mobility space is at a very nascent stage, with considerable opportunities to be solved using AI techniques. At Careem, we have a vision to create an Internet-like network to carry packets of atoms, like the way the Internet carries packets of bits, called “AtomNet”. The AtomNet provides an open network platform that connects, manages and routes multi-mode autonomous vehicles to make transportation ubiquitous. Similar to how packets can travel across multiple transport modalities (Wi-Fi, DSL, cable, and fiber), packets on the AtomNet can travel in motorcycles, cars, vans, trucks, ships , drones and autonomous aircraft. We envision an AtomNet industrial ecosystem with open package protocols to enable package switching and efficient package mobility. With open protocols, coordination costs will drop significantly and local, national, and international transportation gaps will shrink over time.

AtomNet will support Careem’s fast commerce, fulfillment centers, restaurants, grocery stores, transportation, and cross-border commerce. We see the epicenter of AtomNet beginning in the United Arab Emirates due to its progressive regulation and culture of innovation. The current tendency to compromise by improving AI prediction will be reinforced at the expense of short-term factors such as cost, customer experience, and operational excellence. We will continue to invest in our data feeds to help our models learn, build, and manage algorithms at scale.

Our goal is to provide the easiest and best customer experience possible. To simplify things, you have to make them intuitive. To make things intuitively simple, we need to:

  • Know the intention.
  • Understand the context, the brakes, the gains, the needs and the pleasures of the user.
  • Create and implement the right data infrastructure with the right attribution, data provenance, and governance.
  • Build ML models to classify, personalize, contextualize, anticipate, recommend, and learn adaptively.
  • Enable parallel and faster AI experimentation and use our data at scale as a competitive advantage and asset.

We wish to thank the writer of this article for this outstanding content

What it means to be an “AI first” company in a Super App


Take a look at our social media accounts as well as other related pageshttps://www.ai-magazine.com/related-pages/