Bengaluru, May 13: DecisionX, a Bengaluru-based Decision AI company building decision infrastructure for Strategy and Analytics teams, today announced that the DecisionX Agent has ranked 2 globally on Spider 2.0 Lite, the most rigorous public benchmark for enterprise database reasoning. The result was posted on the public leaderboard on April 29, 2026, with a score of 71.84.
DecisionX is the only Indian company in the global top 10. Every other entry is either a large technology corporation with a dedicated AI research division or a specialised academic lab, including JetBrains, Tencent, Samsung SDS Research America, Snowflake, and Tsinghua University. It is the highest finish for an Indian enterprise AI company on a benchmark of this complexity to date.
Why Spider 2.0 Lite is the benchmark that matters for enterprise buyers
Presented at ICLR 2025, Spider 2.0 Lite is built from 547 real-world enterprise database problems. It tests an AI system’s ability to reason over schemas with more than 1,000 columns, generate multi-step SQL across BigQuery, Snowflake, and SQLite environments, and resolve ambiguous business questions with real execution accuracy.
For enterprise strategy and analytics teams, this is the gap that separates an AI that looks impressive in a controlled demo from one that holds up when the schema has 1,200 columns and the business question is genuinely ambiguous. Unlike lab-style benchmarks, Spider 2.0 Lite is designed to expose systems that fail under the real conditions enterprises operate in every day.
“We built DecisionX as multiplayer AI for organisational reasoning: intelligence that compounds across functions, holds under real enterprise data complexity, and gets sharper with every decision. Ranking 2 globally above the research divisions of JetBrains, Tencent, Samsung, and Snowflake is rigorous, independent validation of that architecture. We are built for the hardest version of enterprise reasoning from Day 1. This proves it,” said Ranjan Kumar, Founder and CEO, DecisionX.
DecisionX results reflect two key engineering state-of-the-art achievements.
First, most enterprise AI tools reason cold with no understanding of what a company’s data means, how the business uses it, or what decisions it informs. DecisionX builds semantic understanding before any reasoning begins, across three layers:
-
Data Ontology – Data definitions, business and causal relationships.
-
Domain Ontology – Know-how of your industry, your category.
-
Decision Ontology – A structured record of past decisions.
Together this forms a living causal state graph know-how of every metric, every driver, every decision that grows sharper with every interaction.
Secondly, DecisionX has built specialised agents purpose-built for various kinds of reasoning: root cause analysis, forecasting, and optimisation all operating through the same shared context, compounding institutional knowledge.
The Spider 2.0 Lite ranking is independent validation that this holds under real enterprise conditions. The platform is in production across banking, healthcare, life sciences, consumer, and industrial sectors where strategy teams surface blind spots, track market signals, and act on root causes through chat, grounded in their own enterprise context. This is not a system that works in demos. It is a system that works at scale, on real data, under real conditions.