Synthefy said on August 18, 2026 that it raised a $6.5 million seed round led by Wing Venture Capital as it unveiled a foundation-model platform aimed at tables, transactions, and other structured numerical data rather than text.

Haystack, Samsung Next, Canonical, and Lightscape participated. Angel backers named in the release include Srinivas Narayanan (formerly OpenAI Applied AI CTO), Microsoft Experiences and Devices CPO Aparna Chennapragada, and Meta Superintelligence Labs VP of AI Manohar Paluri.

Nori as the open model layer

Alongside the raise, Synthefy spotlighted Nori, an open-source structured data foundation model released under Apache 2.0. The company said Nori ranks first across 130 public regression benchmarks, that nearly 600,000 Nori models have been downloaded, and that about 5,000 Python installations occurred in its first weeks after launch last month.

Synthefy argues Structured Data Foundation Models can generalize across forecasting, pricing, fraud detection, risk analysis, and infrastructure monitoring without training a separate model for every dataset. Enterprises can use an open model, a managed API, or deploy inside their own environment, with support called out for AWS, Databricks, Snowflake, and Google Cloud.

Co-founder Somi Agarwal said the company is building machine intelligence for numerical problems humans cannot solve at scale, while co-founder and University of Texas at Austin assistant professor Sandeep Chinchali framed a single model layer for the world’s structured data as the long-term goal.

Decoded Take

Most foundation-model capital still chases language and multimodal systems. Synthefy is betting the next durable layer sits under ERP, markets, sensors, and ops tables where accuracy and cost matter more than chat. Open-sourcing Nori builds distribution and a benchmark narrative before paid enterprise packaging hardens. Watch whether Wing-backed traction shows up as Databricks and Snowflake-native workflows, or whether tabular foundation models stay a research curiosity next to XGBoost and classical stacks.