Why do manufacturers still rely on human eyes for inspection when AI vision systems can see more, faster, and cheaper? In this episode of The Connected Factory Podcast, Alexander sits down with Heiko Wirtz, co-founder of ANTICIPATE, to unpack how computer vision is moving from hype to shop-floor value. They explore what it takes to integrate AI into legacy environments, why not every inspection needs deep learning, and how to start small with pilots that actually scale.
Topics Covered
- From rule-based vision to deep learning: where AI adds value
- Why lighting, variance, and product mix break traditional inspection
- Integrating AI into decades-old machines and PLCs
- DIY vs. vendor solutions: where in-house teams succeed—and stall
- The hype around generative AI vs. real shop-floor AI use cases
- Reducing cost-to-value for SMEs and the Mittelstand
- Practical first projects: replacing manual inspection with AI
- Using open-source tools for quick wins and internal buy-in
Key Takeaways
- Start Where Humans Still Inspect
Manual visual checks are prime candidates for AI—fast ROI, less error, and easier scaling. - Traditional Vision Still Has a Place
Not every project needs deep learning; simple rule-based inspection is cheaper and faster when variability is low. - Legacy Integration is the Real Hurdle
Connecting AI to 20-year-old machines and protocols is harder than the modeling itself. - Ease of Use Matters More than Model Accuracy
Tools that automation engineers can deploy without PhDs will define adoption. - Beware the DIY Trap
Building in-house seems easy until edge cases and maintenance costs surface—balance experimentation with reliable tooling. - Shift the C-Level Conversation
Executives hear “AI” and think ChatGPT; engineers must reframe the value in terms of quality, uptime, and data flow. - Prototype, Show, Scale
Run a quick open-source pilot, secure sponsorship, and demonstrate ROI—momentum beats tech perfection.
IT / OT Integration Platform for Industrial DataOps
Connect all your machines and systems with our Open-Source IT/OT Integration Platform to make all shop-floor data accessible at a single point.
Chapters
00:00 Introduction
01:15 From RWTH Aachen to Anticipate
02:42 Computer Vision and AI Hype Cycles
04:40 State of AI in Manufacturing
06:51 Limitations of Traditional Visual Inspection
07:28 Integrating into Legacy Systems
08:58 Role of Automation Companies vs. In-House AI Teams
11:07 Challenges of Homegrown AI Approaches
14:06 AI at the C-Level vs. Shop Floor Needs
15:48 Building Scalable AI Systems for Manufacturing
17:36 Getting Started with AI in Quality Inspection
19:13 Open Source Tools and First Projects
