The Industrial IoT transition from manual quality checks to AI-driven automation can perhaps be described as a "Technicolor moment" – a shift from a monochromatic world of error-prone processes to a vibrant, data-rich reality. Leading this charge is Telit Cinterion, whose collaboration with NVIDIA has turned high-performance edge computing into a cornerstone of modern manufacturing. By integrating 5G connectivity with the NVIDIA Jetson platform and the low-code power of NVIDIA TAO 6.0, Telit Cinterion is effectively removing the "coding barrier," allowing OT technicians – the unsung heroes of the factory floor – to deploy sophisticated visual inspection tools without needing a degree in data science.
In my interview with Telit Centerion, Head of Marketing Strategy Joe Braga pulls back the curtain on how the deviceWISE AI Visual Inspection platform is transforming the industry. From the "Wizard of Oz" style paradigm shift in defect detection to the "Field of Dreams" being built through 5G-enabled robotics, Braga discusses how the synergy between robust connectivity and AI horsepower is creating a new era of autonomous industrial operations. We explore why industries like automotive and pharma are leading the way, and how Telit Cinterion ensures that as AI becomes more powerful, it also becomes more accessible to the people who keep our production lines running.
Carl Ford: How long has Telit Cinterion been collaborating with NVIDIA? What was the first project that has led to your further integration?
Joe Braga: We started collaborating in 2022. We started with the integration of our 5G cards with the Jetson Xavier leading to the integration of more 5G cards with the Jetson Orin which led to the start of the work with the Telit deviceWISE Industrial IoT platform integrating the 5G connected Jetson Orin with the platform for visual inspection.
CF: When you first released deviceWISE AI Visual Inspection, what customer pain points were the drivers? Did they have visual inspection tools? Were they postproduction?
Joe Braga
JB: deviceWISE AI Visual Inspection was launched to solve speed, accuracy, and integration gaps in visual inspection workflows. It targeted manufacturers who needed real-time defect detection, ease of deployment, and AI without complexity—whether for inline production or postproduction quality assurance.
The launch of deviceWISE AI Visual Inspection was driven by manufacturers’ need to overcome persistent quality challenges. Traditional visual inspection methods—often manual and inconsistent—couldn’t keep pace with modern production demands. Customers faced high defect rates during new product introductions, costly rework, and delays caused by slow, error-prone checks.
deviceWISE addressed these pain points with an AI-powered, no-code solution that integrates seamlessly with industrial cameras, PLCs, and enterprise systems. It enables real-time defect detection, automated feedback to operators, and rapid deployment without specialized coding or data science skills. While primarily designed for inline inspection, the platform also supports postproduction use cases such as final quality checks, logistics verification, and predictive maintenance—helping manufacturers reduce warranty claims and improve overall efficiency.
CF: When I think about visual inspection, I think of the movie “The Wizard of Oz” where they go a bleak monochromatic world to the colorful land of Oz. Has the ability to do visual inspection improve that much that the recent years have seen an equally dramatic transition?
JB: Your analogy to The Wizard of Oz is spot-on—the shift in visual inspection technology over the past decade has been nothing short of transformative. Historically, inspection relied on human eyes or basic rule-based machine vision. These methods were slow, subjective, and prone to fatigue-related errors. We’ve moved from a world of limited, error-prone checks to a vibrant landscape of intelligent, automated, and predictive quality control—a true paradigm shift.
CF: A big driver bringing us to this land of OZ has been NVIDIA. How does collaborating with NVIDIA processing impact your AI’s comprehension of flaws and irregularities?
NVIDIA has been the yellow brick road. By harnessing NVIDIA accelerated computing, deviceWISE AI Visual Inspection transforms quality control from basic detection to intelligent comprehension. High-performance GPUs and edge platforms like Jetson enable real-time inferencing, while toolkits such as TAO and TensorRT optimize deep learning models for speed and precision. This synergy allows deviceWISE AI to spot microscopic flaws, interpret irregularities in context, and deliver predictive insights—all at production-line speeds. It’s not just faster; it’s smarter, turning what was once a manual, error-prone process into a dynamic, AI-driven ecosystem that redefines manufacturing quality.
CF: Another Movie I want to reference is “Singing in the Rain” and the frustrating transition from silent to “talkies.” The unsung heroes in that movie are the technicians. Who are the unsung heroes of Visual Inspection? It feels like these are not IT but OT technicians. Is that what you’re seeing, or is there another group leading the charge?
JB: Just as the technicians in Singing in the Rain quietly enabled Hollywood’s leap from silent films to talkies, today’s AI revolution in manufacturing owes its success to OT technicians and quality engineers. These are the hands-on experts who install cameras, fine-tune lighting, configure edge devices, and validate AI outputs on the factory floor. While IT sets the stage for connectivity, it’s OT that makes the magic happen—bridging physical processes with digital intelligence to ensure visual inspection systems don’t just exist, they deliver flawless performance at scale
CF: Enabling that group of non-coders is accomplished by your integration with NVIDIA TAO 6.0, enabling low code “Train, Adapt, Optimize” (aka TAO) to ease the technicians' management of the visual inspection. Where do you see the biggest benefit to your customers’ staff as it relates to their own training, adaptation, and optimization?
JB: The biggest benefit of integrating NVIDIA TAO 6.0 into deviceWISE is how it transforms the role of non-coders on the factory floor. Instead of wrestling with complex AI frameworks, technicians can now “Train, Adapt, Optimize” through intuitive, low-code workflows. This means they can fine-tune models for new parts, adapt to changing production conditions, and optimize performance without writing a single line of code. The result is faster deployment, reduced reliance on data scientists, and a workforce that feels confident managing advanced AI tools—turning what was once a specialized task into an accessible, everyday capability.
CF: I think of deviceWISE as a portal to help manage processes, and I believe that NVIDIA TAO 6.0 is a language set that is integrated into deviceWISE. I believe that the terms in deviceWISE are different from TAO 6.0 microservices, which (to me) means that in your integration you had to balance how your existing customers were trained. Are you passing through the microservices and customers need to understand NVIDIAs terminology, or does DeviceWISE enable the low code to mask the interfaces?
JB: Integrating NVIDIA TAO 6.0 into deviceWISE could have introduced a steep learning curve, but the platform keeps that complexity behind the curtain. TAO’s microservices—powering advanced capabilities like auto-labeling and pose estimation—run under the hood, while deviceWISE delivers a familiar low-code interface. Customers configure workflows through drag-and-drop blocks, and interact with intuitive prompts instead of technical specs. The result: end-users gain the full power of AI without retraining teams or adopting new terminology, ensuring innovation feels effortless rather than disruptive.
CF: Where do we stand on adoption? What industries lead the way and what industries would you like to see participate more?
JB: AI-powered visual inspection has moved from experimental to essential, with adoption surging across manufacturing. Industries like automotive, electronics, and pharmaceuticals lead the way, driven by safety-critical requirements and regulatory mandates. Automakers rely on AI to verify welds and paint finishes, electronics manufacturers use it for micron-level PCB checks, and pharma companies deploy it for packaging compliance—all at speeds and accuracies manual inspection could never achieve. As these sectors set the pace, others such as food and beverage are rapidly following, leveraging AI to ensure product integrity and reduce costly recalls.
Despite these gains, some industries remain slow to embrace AI visual inspection. Textiles and packaging still lean heavily on manual checks due to cost sensitivity and slower digital transformation, while specialty manufacturing and biopharma face challenges with small batch sizes and high variability. Yet the potential is enormous: sectors like aerospace, renewable energy, and consumer goods could unlock significant ROI by deploying AI for defect detection, predictive maintenance, and quality assurance. As low-code platforms and edge AI simplify implementation, these lagging industries have a clear path to catch up—and reap the benefits of smarter, faster inspection.
CF: Speaking of Industries, among your integrations with NVIDIA is your mobility modules. In my mind we now have reversed roles, and you have the equipment that feeds NVIDIA Jetson Thor with more import. (Here I am thinking of the movie “Short Circuit” and Johnny Five needing more input). How do your modules supporting higher speeds improve the information flow for NVIDIA’s Jetson Thor?
JB: In the age of edge AI, connectivity is as critical as compute—and Telit Cinterion’s mobility modules for NVIDIA Jetson Thor deliver ultra-low latency (as tight as 1 ms) and multi-gigabit throughput, enabling robots and autonomous systems to stream massive sensor datasets—lidar, video, IMU—while receiving real-time AI model updates without bottlenecks. Think of Short Circuit’s Johnny Five craving “more input”: Jetson Thor’s 2,070 TOPS of AI power needs a constant flow of data to reason and act instantly. Telit Cinterion’s modules provide that feed, supporting URLLC, network slicing, and LTE fallback for resilience. The result? Seamless cloud integration, dynamic edge inferencing, and safe operation in environments where Wi-Fi falls short—from smart factories to wide-area logistics. This isn’t just connectivity; it’s the neural network for next-gen robotics, unlocking true autonomy at scale.
CF: You’ve highlighted industrial solutions first. Does that suggest cellular is seeing wider deployment as private networks in manufacturing? Or is this more about the overall growth of private networks? Or am I off here?
JB: Your review here is correct. Telit Cinterion’s industrial-first positioning reflects broader adoption of private cellular in manufacturing. Manufacturers are steadily rolling out private 5G, particularly to enable robotics, automation, and mission-critical operations—though regional uptake varies. Telit Cinterion’s messaging around Jetson Thor and mobile AI robots reinforces that 5G private networks are not just a theoretical network layer—they’re increasingly foundational for advanced industrial automation.
So no—you’re not off—this alignment reflects both Telit Cinterion’s strategy and the evolving landscape of “factory-first” private 5G deployments.
CF: “Traditional IIoT and other factory infrastructure has been very limited in the way it collects data and reporting issues,” said Martin Krona, President of Services and Solutions at Telit Cinterion. “deviceWISE Intelligence Suite raises the bar by providing active intelligence into the manufacturing floor, that sees, thinks and acts autonomously across every machine, sensor and workflow. The result is faster decisions, reduced downtime, higher throughput and autonomous optimization.” Are we seeing a convergence where advanced platforms like deviceWISE Intelligence Suite are enabling a new era of autonomous industrial operations? And is this driving broader adoption of AI in manufacturing?
JB: Yes. The rise of platforms like deviceWISE Intelligence Suite marks a pivotal convergence of IIoT and AI, ushering in a new era of autonomous industrial operations. Traditional factory systems were passive, limited to collecting data and reporting issues. In contrast, deviceWISE introduces active intelligence—AI agents that see, think, and act across machines, sensors, and workflows. This evolution enables real-time fault recovery, predictive optimization, and seamless integration with enterprise systems, reducing downtime and boosting throughput. By masking complexity behind a low-code interface, deviceWISE accelerates AI adoption across manufacturing, transforming connected factories into self-optimizing ecosystems.
CF: In effect your integrations with NVIDIA have enabled a “Field of Dreams.” What companies, products, or broad categories are you seeing utilizing the combination of Telit Cinterion with NVIDIA to date?
JB: The combination of Telit Cinterion connectivity and NVIDIA accelerated computing is gaining traction across several high-impact sectors. Automotive manufacturers are leveraging it for real-time defect detection and predictive maintenance on complex assembly lines. Electronics and semiconductor producers use it to manage massive image datasets for micron-level inspection. Pharmaceutical and medical device companies deploy the integration to ensure regulatory compliance through automated packaging and labeling checks. Food and beverage processors are adopting it for hyperspectral imaging to detect contaminants and verify product integrity. Even logistics and energy sectors are beginning to embrace this synergy for autonomous systems and edge analytics—proving that when robust connectivity meets AI horsepower, industries build their own “Field of Dreams” for smarter, faster operations.
Edited by
Erik Linask
