We are excited to bring Transform 2022 back in-person July 19 and virtually July 20 - 28. Join AI and data leaders for insightful talks and exciting networking opportunities. Register today!
During the pandemic, a growing number of manufacturers have begun to pilot — or fully embraced — AI in their organizations. While technical and human roadblocks threaten to slow adoption, manufacturers are deploying AI across a range of maintenance, quality assurance, and production processes. Ninety-three percent of enterprises believe that AI will be a pivotal technology to drive growth and innovation in the manufacturing sector, according to Deloitte. And manufacturing companies are expected to spend $13.2 billion on AI software, hardware, and services in 2025, up from $2.9 billion in 2018.
As the interest in AI among manufacturers grow, a cohort of startups has risen to fill the demand. Founded by Stanford deep learning researcher Andrew Ng, Landing AI is developing a platform that can help manufacturers more readily identify defects in their products. Meanwhile, Elementary has created a system to automate physical product inspections, and Oqton is applying AI to additive manufacturing.
Another vendor in the burgeoning “AI for manufacturing” market is Vanti Analytics, which today announced that it raised $16 million in series A funding. Vanti’s platform allows manufacturing teams to build and deploy predictive models for use cases in several different industries, including semiconductors, automotive, food, energy, and consumer goods.
Applying AI to manufacturing
Nir Osiroff and Smadar David founded Tel Aviv, Israel-based Vanti Analytics in 2019. Osiroff previously worked at Broadcom as a communication algorithmic engineer before joining lidar provider Innoviz to lead the algorithms team. David also worked at Innoviz in the MEMS and opt-mechanics group, where she leads the design, fabrication, testing, and packaging of new hardware modules.
Vanti’s platform is designed to maintain, monitor, and optimize models for manufacturing applications in production. Customers define the use case they want to target, build a model using available data, and optionally train multiple models with different configurations to compare. They then deploy the model and integrate real-time production data using Vanti’s API.
“With the global pandemic, the industry is facing major supply chain challenges,” David told VentureBeat via email. “Also, stringent quality requirements alongside high pressure on costs is something that the industry is facing. In addition, like many other industries, it is facing a major shortage in qualified personnel, so it seeks to move to greater automation as well as finding data analytics talent that can help drive efficiency and innovation through data. Both are challenges that are addressed by Vanti.”
Vanti claims its models are built as a “white-box,” meaning that they provide plain-English explanations for their predictions. Using feedback from the production line, Vanti monitors models’ prediction accuracies and notifies a customer — or disables the predictions entirely — if the accuracy drops significantly. The platform can also detect data drifts, or situations where a model’s predictions become stale and poorly fit to the data that the model is analyzing. In addition to this, Vanti is automatically capable of conducting backtesting, taking the original data that was used to train a model, changing the data to fit a new data structure, and testing to ensure that the model’s accuracy hasn’t degraded.
“Vanti’s architecture builds a customized model for every customer and use case automatically, so part of its capabilities include handling very imbalanced datasets [and] automated detection and treatment of manufacturing events such as disruptions in data acquisition sequences,” David explained. “Adopting Vanti’s solutions can be done seamlessly with any given IT infrastructure and data acquisition and storage that is in place.”
Not all manufacturers are eager to embrace AI in their operations — at least not yet. Legacy systems are restraining the use of AI, particularly for smaller companies with older factories. In a recent PricewaterhouseCoopers poll, 40% of manufacturers cited the inability of existing enterprise resource management and manufacturing execution systems to interface with AI as a barrier to adoption.
Beyond this, there’s healthy skepticism of AI — and concern about its skills requirements — in the industry.
Laying bare the cynicism, a 2021 KPMG survey found that 21% of industrial manufacturing executives and 88% of industrial manufacturing executives believe AI is more hype than reality right now. Seventy-one percent of these executives said that they find it difficult to stay abreast of the evolving AI landscape, while 64% said that their organization struggles to select the best AI offerings.
Twenty-five-employee Vanti claims that it’s already found success with customers including Seagate, Flex, and Innoviz, for its part. Innoviz used the startup’s platform to predict 70% of faults in hardware units, reportedly increasing overall throughput by 9%.
“As noted before, manufacturers are seeking to adopt greater automation and are more open to new technologies being introduced into their operations, so what might have seemed impossible before is now the new norm and the company is very excited about,” David added. “Starting small with limited datasets and surgical injection of value — predictive models — in the manufacturing flow allows the enterprise to see tremendous return on investment and focus only on value generating investments.”
Insight Partners led Vanti’s latest funding round with participation from existing investors True Ventures and MoreVC, bringing the company’s total capital raised to $22 million. Vanti says it’ll use the new capital to fuel growth, expand globally, and invest in its next generation of technology.
VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Learn more about membership.