New York-based Sphinx, a company building AI for data, has launched with a $9.5 million Seed round and an AI copilot for data professionals to transform raw information into actionable insights. The round was led by Lightspeed, with participation from Bessemer Venture Partners, Box Group, K5, Impatient VC and others. Sphinx is also supported by Steve Cohen, Naveen Rao, and leaders from Databricks, Windsurf, and Together AI. The funding will be used to continue building agentic AI that natively interacts with data and data science workflows.
“We’re founding Sphinx, an applied AI research firm tackling a major gap in the field: enabling AI to robustly understand and extract insight from business-scale data, from churn models to supply chains to sabermetrics,” said Rohan Kodialam, co-founder and CEO of Sphinx. “Today’s AI excels at language but still struggles with rows, columns, and the real-world logic behind them. Sphinx is changing that. We’re building machine intelligence that reasons natively over data and unlocks a new class of analytical copilots to accelerate journeys from raw information to measurable business value.”
Kodialam said that this is about finding the gorilla in the data.
“Even the best models have blind spots,” Kodialam noted. “At Sphinx, we’re building the layer that helps AI find the gorilla in the data. GPT-5 is finally here, but it can’t find the Gorilla in the Data. It’s taking the lead on many benchmarks, but at Sphinx our focus is data. In our internal evaluations, our copilot + GPT-4.1 is still outperforming GPT-5 on a range of data-centric tasks, including ones that feel trivial to humans. As models advance, their training sets, benchmarks, and implicit optimization goals shape significant blind spots. GPT-5 is a solid coder — but not a great data scientist. Even state-of-the-art models can miss the gorilla in the data, and Sphinx is putting together the missing layer to fix that.”
The way data scientists work is fundamentally more iterative and exploratory than the workflow of software developers. Unlike other copilots, Sphinx Copilot is purpose-built for data, with a focus on building accurate representations, rigorously verifying models, and grounding responses with quantitative evidence rather than rushing to generate code or conclusions.
“If you’re working with messy, complex, or high-stakes data where accuracy matters: let’s talk. We’re looking for partners who need AI that doesn’t just code about data, but thinks with it,” Kodialam stated.”
Sphinx copilot, available today, works collaboratively to transform raw information into actionable insight via autocomplete and agentic reasoning. It refines forecasts, optimizes operations, and can power applications from supply chains to sabermetrics. Built for data professionals, Sphinx integrates a benchmark-leading agent into environments including Jupyter notebooks and VSCode to meet data teams where they already work. Sphinx was founded by Rohan Kodialam, an AI research leader at Citadel, and Jamie Bloxham, an early technology lead at MosaicML.
“AI is driving a paradigm shift for natural language and code, but traditional data has been left behind,” Kodialam said. “Our researchers and engineers are aggressively innovating on the interface between AI and data to drive tangible value for our partners across industries including CPG, retail, and financial services.”
“Sphinx brings frontier AI capabilities to data analysis, redefining how AI reasons with data,” said Bucky Moore, partner at Lightspeed. “Starting with the core workflows of data teams, Sphinx’s agents will continue to handle more of the tedious work that goes into deriving insights from data. It’s more critical than ever for enterprises to glean key information from their data to fuel business decisions and Sphinx enables this at record speed. Rohan and Jamie are on a mission to define how enterprises leverage AI to become truly data-driven.”
Sphinx builds on recent research breakthroughs in reasoning models, with a distinct focus on the interpretation of tabular and semi-structured information and balancing data exploration with value extraction. Sphinx is ready to accelerate over 93 million users of Jupyter worldwide, and to enable a $100 billion market for data insights.