Startup Sonibel Raises US$1.6M To Bring Real-Time AI to Weld Quality Control

Sonibel Instruments, an AI manufacturing startup developing a real-time weld quality control system, has closed an oversubscribed US$1.6 million pre-seed round. The round was led by Maple VC, with participation from Champion Hill Ventures, Dorm Room Fund, and other strategic partners.

Sonibel’s system uses an aluminum-cased acoustic sensor mounted on the welding torch, along with machine learning, to detect weld defects in real time and provide immediate feedback to operators. The company says the approach can reduce total project costs by more than 30% in some cases.

The company was founded by University of British Columbia graduates Sophia Millar, George Hollo, and Hooman Pirouz, who have backgrounds in welding, shipyard engineering, and startups, and who personally experienced the cost and downtime associated with weld quality control.

The team also interviewed hundreds of welders, project managers, QC teams, production managers, and fabricators to understand the true impact of weld repairs.

“The more manufacturers we spoke to, the more convinced we were that this is a widespread, urgent problem that’s way bigger than most people think,” said Pirouz, Chief Product Officer of Sonibel.

Investor interest in reindustrialization has grown amid rising demand for more data centers, grid expansion, faster shipbuilding, and defense maintenance.

Sonibel positions its approach around augmenting human welders with AI rather than replacing them.

“We’re the first to give human welders robotic efficiency,” said Hollo, the company’s Chief Technology Officer. “Building Sonibel Instruments wasn’t possible a few years ago. We’re taking advantage of the most recent advancements in edge computing, machine learning, and acoustic sensing, and putting them together in a way that hasn’t been done before and isn’t easy to replicate.”

Sonibel collaborated with fabricators throughout the sensor system’s development, training algorithms on real-world welding conditions. Their input shaped the system to work seamlessly alongside welder judgment, providing an early layer of quality control without changing established workflows.

With the new funding, the company plans to expand development of its weld monitoring system and scale deployment with manufacturers.

A video demonstration showing how the system detects weld discontinuities in real time can be viewed here.