Nomagic is turning warehouse picking robots into a production data layer

Nomagic added a $10 million Series B extension on January 28, 2026, led by Cogito Capital Partners, raising total funding above $84 million. The extension followed a $44 million round in February 2025 led by the European Bank for Reconstruction and Development's venture arm, with Khosla Ventures and Almaz Capital continuing support. By then, Nomagic systems had completed millions of picking tasks in 24/7 warehouse environments.

The strongest customer signal came from Brack.Alltron. On May 11, 2026, Nomagic and the Swiss retailer expanded their partnership to include Vision-Language-Action systems in production. Brack.Alltron had already used Nomagic robots for order picking and packing, and the expanded setup supports autonomous warehouse activity during nights and Sunday shifts.

Nomagic's product line centers on Pick, its AI-powered robotic picking system, and Spot, a software layer for decision tracking during warehouse picking. Shoebox Picker handles two-piece shoebox operations and was an IFOY Award 2026 finalist, giving the company a specialized product branch for apparel and footwear workflows where packaging variation can break generic picking systems.

The research layer is now explicit. Nomagic hired Markus Wulfmeier from Google DeepMind as chief scientist in April 2026 to lead work on foundational robotics models and VLA systems. The company connects that work to its Library of Chaos dataset, built from warehouse edge cases collected from deployed robots. Commercially, the data question is whether production failures and odd items become training material quickly enough to improve future picking performance.

The competitive field includes RightHand Robotics, Covariant under Amazon, OSARO, Berkshire Grey-style picking systems, AutoStore integrators, Exotec partners, and conventional warehouse automation providers. Nomagic's distinction is production picking data tied to VLA development, with Brack.Alltron showing the model path inside a live customer operation rather than only a research demo.

Public material does not show robot count by Brack.Alltron site, intervention frequency, pick accuracy by SKU class, uptime by shift, customer retention, or production autonomy logs. The strategic question is whether Nomagic can turn millions of warehouse edge cases into a durable performance advantage. If VLA systems reduce exception handling and expand SKU coverage in live operations, Nomagic becomes a production data layer for robotic picking rather than only a cell supplier.

Have a robotics update Korthos should review? Send news, deployments, product releases, funding rounds, research, or media to tips@korthos.xyz or reach out on X at @agkorthos.