MobiKwik Developers Write 40% Code With AI, Enhance Efficiency

MobiKwik is advancing its technology roadmap through a combination of generative AI adoption, data platform modernisation, and scalable architecture investments. The company is increasingly using AI in software development; evolving hiring priorities, and moving from experimentation to enterprise-wide AI integration, while maintaining a strong focus on customer experience and operational efficiency. In a discussion with FE FUTECH, Saurabh Dwivedi, Chief Technology Officer (CTO) at MobiKwik throws more light on how the company is leveraging technology.

Excerpts

What are some of the recent digital initiatives undertaken at MobiKwik ?

Broadly, MobiKwik is working across three areas -- market innovation, generative AI & agentic AI frameworks, and data and platform modernization.

We are paying keen attention to customer support. A part of it is AI enabled, utilizing the agentic framework, which offers context-aware assistance in a way that does not require customers to wait or to speak to a human agent in the background. While the initial phase is rolled out, we are in the process of scaling it up.

Under the Lens.ai program, a transactional insight platform, which we began two years back, we have been doing incremental changes. It has now been expanded to a knowledge system, which provides a deep dive into the customer’s tax information. We are also building advanced analytics for the stock segment, which is currently in beta.

We are experimenting with a number of AI use cases. The company is scouting for talent in the market for niche areas like voice. So we are taking a hybrid approach by bringing targeted external talent into the organisation and also trying to internally upskill the team.

To empower developers to improve their coding speed and efficiency, we are engaging our engineering teams towards leveraging AI code pilots. Following initial PoCs, we engaged a large AI vendor. As a result, ranging between 30% and 40% depending on the service, our code is generated today using AI. It has tremendously changed how the developers work today compared to how it used to be in the past.

We are also in the process of reimagining our data and the platform organization. The core data architecture has been strengthened to enable near real-time capabilities. This is a direct consequence of the requirement to scale our services because of the complex product use cases we have launched in the recent past. So we are focusing on staying ahead of the curve in terms of technology to be ready for the increasing scale in the coming years.

What challenges have you faced in working on various technologies and how did you overcome them ? 

Changes at the technology platform level typically involve large technical initiatives. One of the primary challenges is determining the right approach for migration — whether to adopt a lift-and-shift strategy or pursue a more incremental transition to the new platform.

To manage this complexity, the technology and development teams begin by thoroughly mapping out the entire transition process alongside the development work. This includes designing comprehensive fallback mechanisms and contingency strategies to address potential disruptions. The teams also place strong emphasis on business continuity planning, ensuring that critical operations remain unaffected even if unexpected issues arise during the transition.

MobiKwik has developed most of the technology in-house. What process do you follow to implement various technology products and services ?

Most of the strategic initiatives are developed in-house by our core technology group. They are composed of dedicated engineering teams, which research, conduct PoCs and test it before adopting any tech environment to be used in our product or back office funnels. 

The public cloud providers have local teams here in India and they collaborate with us in doing POCs coupled with all the associated work. 

Typically the cloud service providers have a comprehensive products and services suite. While they definitely try to push their products, our internal teams test out all the features. Before adopting them, it has to go through our internal technology team validation, testing, and proof of concept. Followed by entering the cycle of being adopted within the organization. 

For instance in the case of AI code generation at MobiKwik, we did a thorough analysis by doing POCs, interacting with the vendor teams to get a clear picture of the product features, how the technology can be embedded, are there any chances of the technology not adding a value in two-to-three years time frame, etc.

Subsequently, the research to production pipelines are thoroughly vetted and then we decide to go live with any technology.

Given the emergence of artificial intelligence and agentic AI, how are you changing your hiring strategy to include more developers who have a strong AI background ?

At MobiKwik, the engineering workflow has evolved significantly over the past year, with developers increasingly integrating AI tools into their daily work.

Earlier, most developers handled both coding and the associated research independently on a day-to-day basis. The company has now introduced AI-based code generation tools, and all engineers are expected to use them for both routine development tasks and technical research.

According to the company, the adoption of AI-assisted coding has delivered measurable efficiency gains. The time required to refactor certain legacy services has reduced by nearly 30%, while AI-assisted code generation now accounts for around 40% of the code written for some newly built services.

The shift is also influencing hiring priorities. While developers are still expected to have strong fundamentals in programming languages and system design, familiarity with AI-powered development tools such as AI co-pilot and code-generation systems is increasingly viewed as a valuable skill.

This trend, the company notes, is not limited to MobiKwik. Across the technology industry, employers are placing greater emphasis on hands-on exposure to AI development tools. At the same time, companies continue to expect engineers to maintain strong conceptual understanding and technical fundamentals.

What are the plans for the calendar year 2026 ? What are you planning to do as far as AI is concerned?

We are working on three pillars. One is towards data maturity. We have strengthened our data pipelines through which the ingestion of data across multiple funnels and channels is done. Now we are trying to build a more intelligent layer on top of it, leveraging AI and ML and move more towards real-time decision-making. 

Scalable architecture is the second pillar we are working on. Our UPI business is growing fast. Thus we'll be investing on some specific modernization initiatives to prepare our services, decouple them from the core components for us to be more feature ready for the product use cases that are becoming complex by the day. A better stack is now available to offer the right experience. 

The last part is towards AI, which will work hand-in-hand across the first two pillars. We'll be moving beyond experimentation to more integrations. Apart from the success on the code generation front using AI, back office operations is another area where we are planning to leverage agentic AI to streamline many back office processes. 

On the app front, we are building AI native features, to provide better merchant and customer journeys. For instance, going deeper into voice as a channel, pilot projects have enabled in-app payments through voice commands. Our teams are also working on automated voice solutions for the collections team. Accordingly, it can then be scaled to support the current call center collections team.

Return on investment from AI is gaining a lot of debate. As solutions are moved from POC to production stage, how are you getting the ROI from your AI investments?

We are also taking steps towards moving from experimentation to integrating AI at MobiKwik. Until 2025, we were in the initial phases of experimenting. Now we are trying to integrate some of the technologies built in-house. Connecting them with the mainstream funnels or in back office operations.

We definitely see a substantial value-add, in enhancing the productivity of our developer community, by leveraging different AI tools, whether in code generation or anything related to analysis of a legacy large stack at MobiKwik. 

In the area of customer features, we are getting ROI in the form of customer retention by providing the right experience to the customers. MobiKwik as an app is more transactional in nature, however, products like Lens.ai, are experiential in nature. So all those value-add features are definitely bringing in RoI in terms of customer retention.