How launching a feature-packed, AI-driven mobile taxi app put an Italian transportation company in the driver’s seat of their local market, with 78% of customer support requests resolved without human intervention and operational costs plummeting 24% below the industry average.
Industry:
Taxi
Mobile App Development
Software Product Development
UX and Design
AI Development
Uber, Lyft, Bolt, Cabify, Gett, DiDi, etc. — the market for mobile taxi booking apps is competitive. Still, new solutions keep emerging in response to the upsurge in the number of taxi app users.
But how do you make your software stand out from the crowd when the market seems littered with a plethora of similar apps? You have to put a premium on user-friendliness and engaging customer experience.
Our client – an Italian transportation company – decided to win local customers with a taxi booking app that enables riders to choose a cab or a cargo van on the one hand and drivers to receive requests from customers on their mobile devices on the other hand.
Number of users of mobile taxi apps worldwide from 2020 to 2029 (in millions)
Source: Statista
When the client approached us, they had a clear vision for their app. It needed to:
They also set out to transform what most taxi apps get wrong: clunky, rule-based chatbots, which are of not much help for users when it comes to customer support. To break the mold, it was decided to provide a smarter, more human-like experience with conversational and generative AI.
The client turned to *instinctools for a team of mobile developers and AI engineers. Our experts got to work, building a performant, high-ROI, and easy-to-maintain solution that would raise the bar on what a taxi app can be.
With GPS access and smartphone hardware integration being pivotal for taxi apps, we crafted native iOS and Android apps, bypassing the limitations of cross-platform development. This choice also allowed us to stay within platform-specific design, aligning with Human Interface Guidelines for iOS and Material Design principles for Android — ensuring a familiar and intuitive experience for every user.
Given high competition in today’s mobile taxi-booking market, the client opted for a bold strategy: investing upfront to perfect the customer experience and deliver a full-scale product that could truly stand out.
Sticking to *instinctools’ tried-and-true product design framework, our business analyst, product designer, and solution architect collaborated from the onset to ensure first-rate customer experience and speed up the development process.
After gathering client requirements for app functionality and outlining its information architecture, we transformed the product vision into clear, actionable wireframes.
Both drivers and riders have been provided with the following shared features:
An additional set of features for riders includes:
Moving forward with product design, our team got to work on high-fidelity clickable prototypes. Most ideas represented in the wireframes made it through to the final app version without a hitch.
That doesn’t mean though we were locked into a strict plan. Throughout the project, our design team had plenty of “aha” moments” and pitched them to a client. For example, we recommended adding emergency assistance as a must-have feature. The client was all in. Now, with just one tap, users can send an instant alert with their location to local emergency services, which is a big win for passenger safety and an easy way to build trust and loyalty.
We also reimagined the functionality of a “Ride history” feature. In a released app, users can easily repeat a past ride in one click.
Rudimentary, rule-based chatbots with pre-written responses feel more like talking to a wall than getting real help, leaving customers frustrated and craving human intervention. No wonder 86% of users prefer chatting with human agents.
But we knew it didn’t have to be that way. With natural language understanding and processing capabilities paired with generative AI to analyze customer queries and answer them, we could flip the script. By empowering the chatbot to understand and respond like a pro, our AI team helped cut down human agent sessions, slash support costs, and still deliver top-notch communication.
Choosing an LLM engine for the client’s conversational agent, we opted for an open-source Llama 2 that is free for commercial use. After connecting it to the in-app chatbot through API, our AI engineers trained it on the client’s customer support manuals to cover the top three common scenarios:
We accounted for over 100 variations of how users might phrase their requests, even with typos and misprints. Such a meticulous approach ensured highly accurate intent recognition – 97%.
Here’s an example of a conversation with Giulia, who forgot her things in a taxi and needed help. The chatbot easily understands the user’s queries despite some misprints in her input and provides clear, actionable guidance.
A month after the app rollout, the chatbot covered 51% of support sessions without human intervention. After training the bot on less widespread user queries, we hit a new mark – 78% of customer support sessions were resolved without involving human agents.
Here’s how the client’s VP of Engineering describes *instinctools’ contribution to the project:
Instinctools’ team exceeded our expectations in terms of delivering sustainable value. Thanks to their profound AI expertise and proactive approach to app development, we got a user-friendly solution with A-level customer support that enabled our taxi booking app to stand out from competitors.
