Paytm’s Prism AI: India’s #2 Global Breakthrough in Enterprise Text-to-SQL Ushers in the Age of Autonomous Data Intelligence

In a landmark moment for Indian artificial intelligence, fintech leader Paytm has developed Prism, a proprietary self-organizing multi-agent “swarm” system that has secured the #2 position on the prestigious Spider 2.0 Snow Leaderboard for Text-to-SQL capabilities.

As the only Indian company featured on this elite global ranking, Prism outperforms numerous entries from top research labs and international technology giants, demonstrating that world-class, production-grade AI innovation is emerging from India.

Solving the Real Enterprise Data Crisis

Enterprises today are data-rich yet insight-poor. Databases routinely contain thousands of tables and columns, convoluted business logic, multiple SQL dialects, temporal reasoning requirements, and complex multi-step workflows.

Traditional text-to-SQL tools falter in these environments, while even advanced large language models like o1-preview achieve only around 17-21% success rates on realistic enterprise benchmarks.

Spider 2.0, an ICLR 2025 Oral paper, was designed to reflect this reality. It comprises 632 genuine enterprise workflow problems drawn from actual data applications. The Snow track on Snowflake tests not just query generation but robust interaction with rich metadata, documentation, and cloud-scale complexity.

Prism’s ability to deliver high-80s to low-90s accuracy in this unforgiving arena marks a genuine leap forward.

“”The future belongs not to organizations with the most data, but to those that can transform data into decisions at machine speed.””

How Prism’s Swarm Architecture Works

At its core, Prism employs a collaborative multi-agent swarm that autonomously handles the entire text-to-SQL lifecycle.

Unlike single-prompt or basic agentic systems, Prism dynamically assembles and coordinates specialized agents that debate, negotiate, and reach consensus.

Planner

Decomposes natural language requests into structured subtasks and reasoning chains.

Retrieval

Navigates massive schemas, metadata, and documentation to surface relevant tables, columns, and relationships.

Proposer

Generates sophisticated SQL queries involving complex joins, aggregations, window functions, and dialect-specific syntax.

Validator

Executes queries against database stubs or schema representations to catch logical, syntactic, or semantic errors.

Refiner

Iterates intelligently on failures, refining outputs until high-confidence results are achieved.

This self-organizing design, powered by high-context models such as Claude Sonnet variants and techniques like Deepthink, ensures reliability, explainability, and minimal human intervention.

The system produces not only the final query but complete decision traces, building the trust required for mission-critical enterprise use.

Led by Soham Acharya (Senior Manager, Data Science) and Anshul Chauhan (Lead Software Engineer), Prism exemplifies deep Indian engineering excellence applied to one of the hardest problems in applied AI.

Global Recognition and Technical Contribution

Prism’s #2 ranking on the Spider 2.0 Snow Leaderboard, alongside earlier strong performances, places Paytm among a select group of frontier players.

It validates the effectiveness of swarm architectures over monolithic approaches in complex reasoning domains and advances the broader field of agentic AI.

By open-sourcing ideas while keeping core code proprietary, the team is contributing valuable insights to the global community.

Strategic Importance and Transformative Potential

Prism’s true significance extends far beyond benchmarks.

It directly addresses the global shortage of elite data talent by democratizing access to complex analytics. Business users, analysts, and executives can now obtain reliable answers from petabyte-scale databases using natural language, dramatically accelerating time-to-insight from days to seconds.

For Paytm itself, Prism delivers internal superpowers in real-time fraud detection, customer analytics, risk modeling, personalization, and operational decision-making.

In regulated, data-intensive sectors such as fintech, healthcare, and supply chain, its validation loops and low-hallucination design are particularly valuable.

“”Trustworthy AI is not merely about generating answers. It is about generating answers that organizations can confidently act upon.””

Market Tailwinds and Massive Growth Opportunity

The timing is impeccable.

The AI agents market is projected to grow from approximately $7-8 billion today to $50-180+ billion by 2030-2033, with compound annual growth rates exceeding 45%.

Embedded analytics, self-service business intelligence, and generative AI-driven productivity gains, estimated at $2.6-4.4 trillion annually across industries, all rely on trustworthy data interaction as a foundational layer.

Prism’s commercialization and expansion potential includes:

* Internal scaling across Paytm’s vast ecosystem.

* Productization as a platform or API for other enterprises.

* Vertical applications in banking, e-commerce, logistics, healthcare, and retail.

* Partnerships with cloud data warehouse providers such as Snowflake and Google BigQuery, along with business intelligence platforms.

* Evolution into proactive, multi-modal autonomous intelligence systems.

Future versions could incorporate real-time streaming, unstructured data, proactive anomaly detection, and seamless orchestration with broader agentic workflows, moving from reactive querying to autonomous enterprise intelligence.

Future Value Creation

Prism represents more than efficiency gains.

It is a foundational capability for the next generation of data-driven organizations. By unlocking the latent value trapped in enterprise databases, it enables faster innovation cycles, superior customer experiences, better risk management, and entirely new data products and revenue streams.

For India, this achievement is both symbolic and strategic. It proves that homegrown AI can compete at the absolute frontier of applied technology, inspiring talent retention, investment, research, and policy focus on high-impact domains.

For Paytm, Prism strengthens its AI-first positioning and deepens its long-term competitive moat.

The Road Ahead

Paytm’s Prism is more than a leaderboard victory.

It is a blueprint for reliable, scalable, and trustworthy enterprise AI that solves painful, expensive problems today while laying the groundwork for autonomous data ecosystems tomorrow.

As organizations worldwide race to monetize their data assets, innovations like Prism will increasingly separate market leaders from laggards.

India now has a compelling new flagship in the global AI narrative, one that does not merely follow trends but helps define the future of intelligent enterprises.

The world is watching what Paytm builds next.

The era of autonomous data intelligence has begun.

Strategic Reflection

Prism’s achievement may ultimately be remembered not for a leaderboard position, but for demonstrating a broader truth: the next wave of enterprise transformation will be driven by autonomous systems capable of reasoning across data, context, workflows, and decisions. In that future, organizations that successfully combine trustworthy AI, domain expertise, and operational execution will enjoy enduring competitive advantage.

India’s emergence as a creator of such frontier technologies signals an important shift in the global innovation landscape. The nation is increasingly moving from being a consumer and implementer of advanced technologies to becoming a creator of globally significant intellectual property and enterprise platforms.

(Major General Dr. Dilawar Singh, IAV, is a distinguished strategist having held senior positions in technology, defence, and corporate governance. He serves on global boards and advises on leadership, emerging technologies, and strategic affairs, with a focus on aligning India’s interests in the evolving global technological order.)