Unifying the data landscape: Insights from Firebolt's CXO mixer on scaling modern platforms

Modern technology leaders are increasingly experiencing tool fatigue. The challenge of managing fragmented systems and multiple query accelerators often undermines both agility and efficiency. To address this issue, Firebolt hosted an exclusive CXO mixer in partnership with YourStory on February 18, 2026, at Conrad Bengaluru. The event, themed ‘Simplifying the Stack: Cutting Through the Complexity of Modern Data Ecosystems’, brought together senior data and technology leaders to exchange practical strategies for streamlining modern data architectures.

The data infrastructure wall

The evening kicked off with a keynote by Sandeep Mathur, Managing Director, APAC, at Firebolt. He addressed the data infrastructure wall that fast-growing companies hit, a problem severely accelerated by the onset of AI agents and large language models querying data. Traditional setups and legacy systems are struggling under the weight of these new demands, where autonomous agents can fire significantly more queries than traditional business intelligence dashboards.

Mathur pointed out the sheer scale of this shift, noting, "the agents are going to fire 10x to 100x queries than what were being done earlier". He added, "So, it's a world change in terms of what amount of pressure your analytical databases are going to come under."

Engineering teams are currently spending excessive time tuning queries and managing infrastructure. This constant maintenance detracts from product innovation and delivering customer value. Furthermore, existing tech stacks often comprise a convoluted mix of systems, combining general-purpose data warehouses, data lakes, and fast query accelerators to handle different workloads.

Firebolt was built to solve this exact problem by combining the functionality of a data warehouse and a query accelerator into one single analytical database. With features like vector search, native AI integrations, and high concurrency support, Firebolt empowers organizations to achieve sub-second query processing without the headache of managing fragmented tools.

Watch the full video here

Abstracting complexity for scale

The core of the event was an engaging panel discussion moderated by Srinivasulu Grandhi, VP Engineering and Site Leader, ex Confluent, and strategic advisor to Firebolt. Reflecting on the core challenges discussed by the speakers, Grandhi observed, "This theme of abstractions, platform tuning, optimization, and the right sizing, the architecture for each workload, is coming up quite often".

The panel featured Alok Sharma, Director of Central Platform Engineering at Meesho; Sandeep Kohli, Founder and CEO of Divyam.AI; and Kamalanathan Viswanathan, Senior Engineering Manager for Data and AI at FourKites.

The panelists explored the evolution of their respective data platforms and the necessity of abstraction. Sharma explained Meesho's transition from a traditional data warehousing stack to a real-time data lake architecture. He highlighted that while a platform must abstract out infrastructure scaling, builders still need granular control. "When I speak to a lot of my platform engineering teams, they need all the nuts and bolts and tunables to be present because they are operating at scale, and they have a lot of constraints that come into play, right from optimizing our TCO to optimizing our performance," Sharma explained.

Kohli echoed this sentiment, drawing from his experience building platforms at Flipkart. He noted that a data platform should shield business users from complexity entirely. "For them, the platform should be like a black box, because I don't believe that they can optimize, and they can really think in terms of optimization, because they are generally thinking in terms of business problems," Kohli stated.

The conversation naturally shifted to the integration of AI within data stacks. Viswanathan introduced FourKites' conversational AI initiative, Foresight, which allows customers to interact with both real-time and historical supply chain data. When discussing data governance in the age of democratized AI access, Viswanathan emphasized the importance of striking that right balance between where strict governance is required versus where we need to free it up so that people can extract maximum value out of the data itself.

Watch the full video here

Real-time AI in action

To bring these concepts to life, Jauneet Singh led a practical demonstration of Firebolt's capabilities for agentic AI applications. The demo showcased a real-time fraud detection system utilizing a multi-agent setup, including a monitoring agent, an analyzing agent, a notifying agent, and a fraud detection agent.

Singh demonstrated how Firebolt functions as the foundational data layer, ingesting transactions and performing rapid vector similarity searches. Even while processing a dataset of 500 million rows on the smallest compute engine, Firebolt handled complex vector and traditional queries in mere milliseconds. This extreme concurrency and low latency are critical. As Singh pointed out during the session, "Because when you have the agent, or you are building any bot on top of it, you cannot restrict your end users to query the data".

Watch the full video here

The message that evening was clear. The era of tolerating disjointed, inefficient data architectures is coming to a close. With unified platforms capable of handling diverse analytical workloads autonomously, modern engineering teams can step away from constant query tuning and return their focus to building truly impactful products.