Many manufacturers struggle to scale vision solutions beyond the pilot line. This expert roundtable demonstrates how modern AI vision systems achieve consistency and flexibility across multiple sites. Backed by new research, we’ll show how companies, particularly mid-sized manufacturers, are leveraging modular architectures, cloud + edge infrastructure, and adaptive AI models to deploy applications across an entire enterprise.
Overview
As manufacturers race to automate, scaling AI-powered machine vision beyond a single application is a persistent and costly challenge.
This expert roundtable brings together expert perspectives in machine vision and AI to tackle that challenge head-on, exploring how companies — particularly mid-sized manufacturers — can build the infrastructure, architecture and organization needed to deploy vision solutions across an facility.
Panelists will examine where AI inference should actually run — at the edge, on local servers, in the cloud or across a hybrid of all three — alongside the hardware choices, data strategies and feedback loops that determine whether a deployment scales or stalls. The conversation will also address how to integrate machine vision outputs into existing MES, ERP and SCADA workflows without creating new bottlenecks.
Cognex, the world's leading provider of machine vision solutions, anchors the discussion with decades of deployment experience across automotive, electronics, healthcare, semiconductors, and logistics — industries where the margin for error is essentially zero.
Key Takeaways
- Understand the what's needed to help AI vision scale across manufacturing lines
- Identify tradeoffs between edge, on-premise and cloud inference in production manufacturing
- Navigate the data strategy and model monitoring challenges that keep AI vision systems accurate and reliable
- Integrate machine vision into existing MES, ERP, and SCADA systems without disruption
Speakers
Over the past 34 years, Dean Phillips has been a leader in the technology sectors and the world of smart manufacturing. His is a continuous advisor to the smart manufacturing advisory committees with Society of Manufacturing Engineers and PMA Precision Metalforming Association. Phillips has been a speaker on IoT, AI, robotics, VR and AR. He is the creator of reality safe, a VR / AR safety training solution for manufacturing, working with Purdue. Phillips has spent the majority of his time advancing IoT, maintenance and safety to be more predictive and develop outcome based solutions. He has been on the board of directors for SME and an advisor for TTU, MTSU and TCAT. He has been a contributing lecturer on big data and it’s value and need for artificial intelligence to filter the information into actionable items. Phillips provides safety and development to cobot users and assists universities to outline expectations from manufacturers.
Brian started his career in factory automation with Rethink Robotics, helping to define how collaborative robots can work side by side with humans. Since joining Cognex in 2017, he has developed products for multiple vision applications, from simple sensors to complex embedded systems. He has focused his recent efforts on bringing AI into the entire Cognex portfolio – including Dataman Barcode readers, In-Sight vision systems, and cloud based training of advanced AI models through OneVision.
Alvin Clark is an AI Engineer on the Developer Relations team at NVIDIA, where he helps partners build and scale AI solutions using NVIDIA’s vision, language, and multimodal model technologies. His work focuses on industrial automation, smart manufacturing, and AI agents—collaborating with ecosystem partners to bring advanced AI into real-world factory operations.
Greg Tanaka is a computer science and electrical engineer trained at Caltech and UC Berkeley, and he also studied at Stanford University. He founded Percolata, where he raised $10 million from investors including Google Ventures and Andreessen Horowitz to develop edge-based computer vision systems for real-time object tracking and classification. His work spans hardware specifications for edge AI deployments, machine vision quality control, hyperspectral imaging, and industrial data analytics. Greg is also an experienced public speaker, guest lecturer at Stanford and UC Berkeley, and former Palo Alto City Council member.