Synthefy Raises $6.5M for Numerical AI Models

Synthefy Inc. has successfully closed a $6.5 million seed funding round, signaling a significant shift in the artificial intelligence model market. The capital infusion, led by Wing Venture Capital with participation from Haystack, Samsung Next, Canonical Crypto, and Lightscape, along with notable angel investors from OpenAI Group PBC, Microsoft Corp., and Meta Platforms Inc., will fuel the expansion of its novel foundation model platform. This platform is specifically engineered to process and understand numerical data, diverging from the text-centric approach of traditional large language models.

The company pioneers what it terms "Structured Data Foundation Models" (SDFMs). These models operate by digesting extensive numerical datasets, enabling them to discern intricate patterns and relationships within time-series data and tabular information. Unlike conventional machine learning algorithms that often require extensive bespoke training for each new numerical task, Synthefy's SDFMs are designed for rapid generalization and enhanced accuracy in calculations. This approach promises to dramatically reduce the time and resources enterprises currently dedicate to data preparation and model fine-tuning for applications like fraud detection and dynamic pricing.

A key development accompanying the funding announcement is the recent open-source release of Synthefy's first SDFM, named Nori. Despite its remarkably compact architecture, a 30-million parameter version of Nori has demonstrated superior performance compared to Google LLC's significantly larger 1.6-billion parameter TabFM model in benchmark tests. When its advanced "Thinking" capabilities are activated, Nori's efficiency is further amplified, outperforming Google's model while being a mere 2% of its size. This efficiency is a critical differentiator in the computationally intensive field of numerical AI.

Somi Agarwal, co-founder and CEO of Synthefy, highlighted the transformative potential of SDFMs for enterprise workloads. "The traditional approach requires weeks of effort for each new problem, with no accumulated learning," Agarwal explained. "Nori allows teams to address new datasets and challenges with immediate, strong predictions, cutting down evaluation from weeks to minutes." This reusability and rapid deployment capability address a significant pain point for businesses relying on accurate numerical analysis for critical operations.

The market for AI-driven numerical analysis is substantial, with applications spanning financial services, e-commerce, and logistics. Sectors such as fraud detection, demand forecasting, and dynamic pricing are estimated to represent billions of dollars in potential efficiency gains and revenue optimization annually. Synthefy's SDFMs are positioned to capture a significant portion of this market by offering a more efficient and effective solution than existing machine learning frameworks like LightGBM and XGBoost, which often demand laborious manual tuning.

Looking forward, Synthefy plans to leverage the new funding to accelerate its research and development initiatives, expand its engineering team, and enhance the Nori model. The company is also actively seeking strategic industry partnerships. The strong initial reception for Nori, evidenced by over 600,000 downloads in its first few weeks, underscores the market's appetite for advanced numerical AI solutions. Synthefy intends to monetize its technology through premium enterprise offerings, including managed API services, proprietary features, enhanced security, and dedicated support, building a robust commercial ecosystem around its open-source foundation.

Gaurav Garg, founding partner at Wing Venture Capital, expressed strong conviction in Synthefy's vision. "Synthefy is constructing a model platform capable of addressing some of the world's most extensive and valuable datasets," Garg stated. "Its blend of technical prowess, efficient design, open accessibility, and early market validation positions it to define this emerging category of structured data AI." The company's focus on numerical intelligence represents a critical expansion beyond the current wave of text-based AI, unlocking new frontiers for data-driven decision-making.