Toyota Industries innovates its paint shop processes with Azure industrial AI

In high-volume automotive manufacturing, paint defects often surface late—after value has already been added—driving rework, delays, and uncertainty. At Toyota Industries’ Nagakusa Plant, paint quality is one of the most sensitive stages in its bumper production flow. Transitioning from by-sight inspections conducted by a shrinking pool of seasoned paint specialists to a data-driven quality management approach became imperative.

Like many manufacturers, Toyota Industries was already collecting hundreds of data points across its paint operations—from equipment sensors and process parameters to environmental conditions such as temperature and humidity. That data was unified and structured in Microsoft Azure IoT Hub, but teams still lacked a holistic, real-time view of operations to understand which variables were actually driving defects.

As a result, identifying root causes became a slow, resource‑intensive process, with engineers and specialists spending weeks manually chasing correlations—while driving up quality‑related costs. At the same time, increasing regulatory scrutiny created imperative for Toyota Industries to advance its sustainability initiatives.

For leadership, it was clear that incremental analytics improvements would not deliver the step‑change required. “We needed a fundamentally different approach—one that could connect operational shop-floor data, add operational context, and turn insights into action and decision-making quickly for our frontline teams and executives,” says Yoshihisa Watanabe, GM of Automotive Division, Digital Transformation Office at Toyota Industries.

Building a data foundation that works at factory scale

To move from retrospective analysis to near real-time control, Toyota Industries collaborated with Microsoft and partner Sight Machine to build a unified, AI‑ready industrial data foundation on Azure. It’s designed to operate at production scale rather than as a dashboard experiment.

The first step was running a pilot. Teams conducted a three-month proof of concept using its actual factory data. “Through this pilot, we were able to see that data which had previously been difficult to connect could be successfully structured and analyzed through a pipeline,” says Takuya Hatakeyama, Project Lead at Toyota Industries, responsible for IoT and factory data utilization.

The second step was to confirm whether the outcomes it was aiming for—such as improvements in quality indicators and changes in how work is performed on the shop floor—would realistically follow from the technology. “Seeing real operators use data effectively helped us understand that this was not just a technical solution, but one that could change how people work and make decisions," says Hatakeyama.

Accelerating innovation within a strong data culture

In the paint process, many interrelated factors can influence outcomes, and analysis at scale matters. Joseph Fryer, Director of Customer Outcomes at Sight Machine, describes how AI and machine learning were leveraged to analyze nearly 400 variables before narrowing to the signals most correlated with defects—an approach intended to accelerate learning and focus frontline action.

Security and governance were built in from the start by using the Azure platform. “Sight Machine’s software runs inside Toyota Industries’s Azure tenant, so the data never leaves their environment,” says Fryer.

Just as important, the solution was built to be usable beyond data specialists. The project emphasized shared visibility, so manufacturing, production engineering, and related teams could view the same data and align on decisions faster.

“Toyota Industries already has a strong data culture, but this accelerated it,” Watanabe adds. “Engagement, adoption, and excitement were extremely high.”

Measurable impact—and a platform for what’s next

By stabilizing temperature, Toyota Industries reduced seeding-related defects by about 25% during that period. The company also reported an approximately four-fold increase in opportunities to resolve defects. Its daily standup meetings take 80% less time to prepare for, significantly improving day-to-day communications, and the company expects an 18% lower emissions rate during winter painting months.

Most important for leaders, improvement cycles accelerated. “What used to take weeks can now be done in a day; sometimes in less than an hour,” Fryer says, citing an end-to-end bottleneck analysis completed in about 45 minutes from review to actionable insight.

On the shop floor, the change showed up as trust and alignment. “For frontline workers, the biggest change was trust, both in the data and in each other,” says Koichiro Taka, a paint production leader at Toyota Industries. With shared dashboards and near real-time visibility, supervisors and engineers can spot abnormal patterns earlier, discuss potential causes using the same view, and apply countermeasures sooner.

“By partnering with Microsoft and Sight Machine, we can focus on manufacturing excellence instead of spending years building data platforms internally,” says Watanabe.

With a scalable foundation in place, Toyota Industries is considering expanding this approach beyond bumper painting to body painting processes and exploring applicability to other continuous production areas over time.

“What began as a paint shop challenge has become a scalable blueprint—one that shows how industrial AI delivers value when data is contextualized first,” concludes Hatakeyama.

Discover more about Toyota Industries on LinkedIn and⁠ YouTube.