In today’s rapidly evolving digital landscape, artificial intelligence (AI) is no longer just a buzzword—it has moved beyond pilots and proof of concepts to become a critical component driving real-world applications across industries. Mukundha Madhavan, APAC Tech Lead at DataStax, is at the forefront of this transformation. With DataStax’s expertise in data technologies and AI, he shares how real-time data technologies and generative AI (GenAI) are revolutionizing sectors such as finance, retail, and automotive. In this interview, Madhavan discusses practical AI applications, scalability, and the groundbreaking innovations helping enterprises stay ahead in this new AI-driven era.
TAM: How are real-time real-time data technologies and GenAI transforming operations across industries like finance, retail, and automotive?
Mukundha Madhavan: 2024 seems to be the year of production AI. We’ve moved beyond pilots and POCs, and now we’re seeing actual production-level implementations. Some of the low-hanging fruits include personalized recommendations, proactive support, and real-time fraud detection. In general, we’re also seeing a lot of automation in workflows and manual tasks. Companies are focusing on automating all of these processes.
Additionally, people are stepping back and asking how to enhance customer experiences. How can we provide a guided, personalized, end-to-end customer lifecycle—from discovery and presales to the buying process, post-sales support, and then repeating the cycle. It will be fascinating to see how these technologies drive new use cases.
From a customer standpoint, I can give you a few examples. Arre Voice, a voice-based social media platform, is leveraging data and generative AI tools to improve its recommendations. By serving recommendations based on this AI, they’ve seen a 40% improvement in user retention. These examples prove that the technology is not only ready for production but is also having a real impact on business outcomes.
TAM: With the increasing integration of AI and real-time data, How are businesses improving their operational efficiency with real-time data technologies and AI, and what new capabilities can organisations unlock for smarter decision-making?
Mukundha Madhavan: If we take a step back and look at all the use cases we’ve discussed so far, we can break them down from a technical standpoint. To perform any unit of work, there are three key steps involved. First is discovery—finding the necessary information or data. For example, if I’m a support agent and a customer comes to me with an issue, I need to find out what the issue is and where it’s documented. This involves searching through enterprise systems, documents, or drives to retrieve the relevant information.
The second step is processing this information. How long does it take to understand what it means and how it applies to the task at hand? Finally, the third step is the actual execution of the work. What we’re seeing across industries is that a lot of time is spent on the discovery and processing steps. This is where having a strong data platform and AI tools can make a huge difference. Having the right data available at the right time, combined with AI, can simplify these two steps, allowing people to focus more on executing the task itself.
Let me give you an example from the healthcare domain. SkyPoint, a healthcare company based in the US, provides technology for elderly care homes. They built an AI agent for care providers to access information about residents. This agent can fetch all the necessary details about a resident, including their latest prescriptions and the recommended care or service they need at any given time. By using this AI agent, care providers are saving up to 10 hours per week. That’s 10 hours they can now spend focusing on caring for residents instead of searching through documentation and trying to interpret what it means. These are excellent examples of how operational efficiency improves. For any task, it’s important to evaluate how much time is spent on discovery and processing. These are the areas where AI and a robust data platform can really make an impact.
TAM: How does DataStax’s technology empower startups and enterprises to optimize their AI initiatives, particularly in the context of real-time data processing?
Mukundha Madhavan: DataStax is a GenAI data company, and that means we focus on two things. First, we deliver a data platform that allows developers and enterprises to get their data ready for generative AI. This includes building their data strategy, modernizing their data, and making it AI-ready. Second, we provide an AI platform-as-a-service (AIPaaS), which enables customers to rapidly build AI applications on this data and take them to production.
That’s what we bring to the table, and we offer this in a variety of form factors. Our data platform is available as a fully managed service, so customers can, with just a few clicks, have a massively scalable, high-performance data platform up and running in about five minutes. Similarly, our AIPaaS is fully managed in the cloud, allowing customers to start building AI applications with just a few clicks. For large enterprises with compliance or regulatory requirements, they can choose to run it themselves since it’s built on an open-source foundation, allowing for self-management wherever they prefer.
Now, let me specifically talk about startups. In India, for example, we work with PhysicsWallah, one of the leading EdTech companies. They use our fully managed service to build their AI applications, including an AI-based tutor that helps students learn specific concepts. Students can chat with this AI tutor anytime they need help, whether it’s with worksheets or understanding a problem. They can even ask for explanations in different ways. It’s multimodal and multilingual, which makes it highly accessible. The feedback from students has been very positive.
When PhysicsWallah launched this application, we knew it worked well for individual students and on smaller scales. But the big question was, would it scale to handle a million users? That’s where the scalability and performance of DataStax’s data platform came into play. When PhysicsWallah’s traffic increased 50x in one day during the launch, there were no issues—the platform scaled seamlessly.
This shows that for startups and enterprises alike, our technology is proven. They can trust it to build their applications, take them to production, and scale effortlessly.
TAM: In the Gen AI ecosystem, where rapid data management and scalability are critical, How does DataStax address the growing need for enhanced data management to support real-time AI applications?
Mukundha Madhavan: From a foundational perspective, we are built on an open-source foundation, specifically using Apache Cassandra. This technology stack has been battle-tested for scalability and was designed to address the scaling challenges of traditional technologies. So, scaling is something that comes naturally to DataStax, and we’re well-known for it.
However, while having a highly scalable platform is crucial, it’s not enough on its own. You also need to build something useful on top of it. That’s where the importance of getting your data ready for generative AI comes into play. This means organizing and representing your data in a way that your AI applications can fully leverage.
New technologies like vector search and graph databases play a big role here. We’ve integrated these techniques natively into our platform. Now, developers and customers can not only bring their data into the platform but also take advantage of cutting-edge technologies like vector search and graph databases, all within the same scalable system.
We’ve also built a data API that simplifies database management. If you’re familiar with databases, you know the traditional process involves creating tables, rows, and columns. With our platform, developers can bypass this step—you can just start creating and reading data without setting up all that structure manually.
Recently, we acquired Langflow, which adds another exciting layer to this journey. Langflow offers a no-code agent development framework, allowing developers to drag-and-drop and quickly build AI applications on our data platform without writing any code. This helps streamline the development process and addresses some of the common challenges that developers and enterprises face.
By combining data representation through vector and graph technologies, no-code development with Langflow, and our scalable platform, we are helping enterprises easily build and take AI applications to production. This is how we’re supporting businesses on their AI journey.
TAM: As enterprises adopt next-generation AI and database technologies, what are the key challenges they face, and how is DataStax helping them navigate this evolving landscape?
Mukundha Madhavan: We hear a lot about challenges, but the most common issue I come across when talking to developers and customers is the fragmentation of data sources. There are so many databases and data stored in various places. How do you know where to fetch it from? That becomes the biggest challenge.
Secondly, the existing data often isn’t ready for consumption. This is where concepts like the “system of engagement,” “speed data layer,” or “fast data lane” are becoming more common in the industry. People are starting to build these systems, and they’re critical for supporting the kind of AI applications we’re discussing.
It’s not just about existing data anymore. Generative AI opens up new possibilities by bringing in unstructured data, which was often unorganized in the past. The scale of this data is enormous. Now, when you combine all of this data, the technology needs to support that scale and volume. This is where traditional solutions fall short—those are the common challenges from a technical standpoint.
The solution, while seemingly straightforward, requires the right technology. We provide tools to improve scale and performance. With Langflow and AIPaaS, we offer two key tools. First, they help ingest and prepare existing data for AI, including unstructured data. You can transform it however you need, getting it ready for generative AI. With Langflow, you can connect to existing ecosystems, models, and databases, then quickly build applications using a drag-and-drop interface. Every developer in your company can create AI applications tailored to their needs.
If needed, they can also take these applications to production. What’s exciting is how the time to production has drastically shortened. What used to take months can now be done in weeks. Regarding skillsets, we hear a lot about the need for upskilling. But my perspective is that the space is moving incredibly fast, and it’s hard for anyone to stay on top of everything happening in the industry. That’s where these tools come in handy. We offer a clear path: the essential steps required to build AI applications, take them to production, and see their value. Once you see that value, you can iterate and keep building.