AI in Manufacturing

1. EPISODE SUMMARY

This episode explores why most manufacturing AI initiatives fail and what companies must do to build a foundation for AI-native industrial intelligence. Craig Scott, Founder and CEO of Fuuz, an industrial intelligence platform, shares insights from nearly a decade of bridging the gap between shop floor data and enterprise systems. The conversation reveals why the missing "shim" between operational technology and enterprise systems is the root cause of unreliable data in manufacturing, and why model-driven approaches are essential for scaling AI across industrial operations. Craig explains how organizations can achieve a single source of truth by implementing a persistent contextualization layer that governs data before AI ever touches it. Whether you're struggling with fragmented point solutions, evaluating industrial data platforms, or preparing your data infrastructure for AI, this episode provides a practical framework for building scalable industrial intelligence.

2. KEY QUESTIONS ANSWERED IN THIS EPISODE

• What is fundamentally broken with current manufacturing data infrastructure and how does it impact AI initiatives?
• Why do most AI pilots fail to scale in manufacturing environments?
• What is a model-driven approach to industrial data, and why is it superior to in-line data transformation?
• How do you balance enterprise governance with plant-level flexibility in industrial data architectures?
• Should manufacturers adopt industry-standard data models like ISA-95 or build custom models?
• What is the difference between a data lake and an operational intelligence platform?
• How can manufacturers prepare their data foundation before investing in AI technologies?

Traditional MES platforms were built for a manufacturing world that no longer exists.

They assume stable product lines.

They assume you have time for lengthy implementations, tolerance for complexity, and operators who can navigate digital forms while running production.

But here's the challenge. Today's manufacturing reality is different:

⇨ Markets demand the flexibility to shift from 1.5-liter bottles to 1-liter bottles overnight

⇨ Low volume, high mix production is now the norm

⇨ Tribal knowledge is retiring faster than it's being captured

⇨ Workers stay 2-3 years, not 20, making traditional training models obsolete

The cost of this disconnect?

❌ Frontline workforce unable to contribute operational intelligence at scale

❌ ROI delayed by complexity, not capability

❌ Two-year deployment cycles for basic systems

❌ Digital initiatives stuck in pilot purgatory

That's why leading manufacturers are rethinking execution from the ground up, shifting from monolithic systems to AI-native, human-centric platforms built for today's workforce reality.

This new approach is effective because it’s built with an AI-native mindset, not a digitized version of paper-based processes

✅ AI-generated SOPs from video, cutting engineering time by 80%

✅ Learning systems that surface troubleshooting guidance from historical fault data

✅ Human-centric design that captures operational intelligence without disrupting workflows

✅ AI-powered interfaces that enable natural interaction; think voice, not dropdowns

✅ Rapid deployment measured in weeks

✅ Scalable without complexity; connect thousands of machines without lengthy integrations

The companies winning today aren’t planning more; they’re executing faster and adapting continuously.

In this episode of the AI in Manufacturing podcast, I speak with Mickey Shaposhnik, Founder and CEO of Next Plus, about how practical, AI-powered frontline execution is redefining operational excellence.

Watch/Listen now

Agentic AI Framework for Manufacturing Operations AI in Manufacturing Podcast Show Notes

Episode Guest: Gilad Langer, Head of Digital Transformation Practice at Tulip Interfaces Host: Kudzai Manditereza Publication Date: [Insert Date]

Episode Summary

Manufacturing systems are complex adaptive systems that require a fundamentally different approach to AI implementation than traditional monolithic architectures. In this episode, Gilad Langer draws on 30 years of manufacturing experience—including PhD research that laid the groundwork for Industry 4.0—to introduce a composable agentic framework specifically designed for frontline operations. He explains why adaptability has become a competitive necessity in today's disrupted markets and how multi-agent systems can transform innate factory equipment into intelligent, communicating entities. The conversation covers practical implementation strategies, the artifact model for structuring manufacturing data, and why cultural change remains the biggest obstacle to agentic AI adoption.

Key Questions Answered in This Episode

• What is an agentic AI framework for manufacturing and why do factories need one?
• How do complex adaptive systems apply to manufacturing operations?
• What are the five pillars of composability in manufacturing?
• How should manufacturers structure their data for AI agents using the artifact model?
• What is the difference between staff agents, builder agents, and artifact agents?
• How do you implement agentic AI in a brownfield manufacturing facility?
• Why do traditional MES systems fail to deliver the adaptability modern manufacturing requires?

Context isn't static.

It's a living layer of knowledge built through problem-solving, conversation, and understanding the complex relationships on the factory floor.

This simple truth is often overlooked in industrial data strategies. 

We’ve been conditioned to believe that context can be predefined; baked into standards, taxonomies, and hierarchies.

But in real-world manufacturing, things change, people think differently, and use cases evolve.

So how can we build this dynamic layer of understanding for industrial AI?

In our latest AI in Manufacturing episode, I spoke with Bob van de Kuilen, Director at Thred, about a more human-centric approach to industrial data contextualisation using Knowledge Graphs.

Thred is a tool that plugs into Ignition Platform, enabling users to visualize their factory assets in a knowledge graph, link related data points, embed domain expertise, and deliver structured, contextualized data to AI and analytics tools.

We discuss:

 ✅ Why traditional approaches to data context often fail

 ✅ Knowledge Graphs act as a mind map for data

 ✅ The practical steps to building context

 ✅ How this new context layer serves as the perfect foundation for AI agents.

SA-95 is a standard that’s often misunderstood, but incredibly powerful.

Is the Timebase free historian getting an AI-Native DataOps component with Knowledge Graphs capability? You’ll hear it here first.

In the latest episode of the AI in Manufacturing podcast, I sit down with Jeff Knepper, President at Flow Software Inc., to discuss the intersection of Information Management and AI in modern manufacturing, plus the exciting announcement of Timebase Atlas launch.

Here’s some of what we cover in this episode:

✅ Why manufacturers struggle to make use of their data

✅ Building reliable pipelines for AI-driven use cases

✅ AI Agents in Manufacturing – Where they fit and what they need

✅ Unified Analytics Framework vs. Unified Namespace

✅ Historization Strategies – Best practices from edge to cloud

✅ Timebase Atlas Launch Announcement: Data Modeling, Pipelines, Knowledge Graphs, and AI interfaces

✅ MCP and Flow AI Gateway: Beyond APIs to Context-Aware Agent Interfaces

Most data-quality initiatives focus on things like freshness or schema. That works for IT data, but not for sensor data. Sensor data is different. It reflects physics. To trust it, you need contextual, physics-aware checks. That means spotting: → Impossible jumps → Flatlines (long quiet periods) → Oscillations → Broken causal patterns (e.g., valve opens → flow should increase) It’s no surprise that poor data quality is one of the biggest reasons manufacturers struggle to scale AI initiatives. This isn’t just data science, it’s operations science. Think of data quality as infrastructure: a trust layer between your OT data sources and your AI tools. Making that real requires four building blocks: 1. 𝐒𝐜𝐨𝐫𝐢𝐧𝐠 – Physics-aware anomaly rules, baselines 2. 𝐌𝐨𝐧𝐢𝐭𝐨𝐫𝐢𝐧𝐠 – Continuous validation at the right cadence (real-time or daily) 3. 𝐂𝐥𝐞𝐚𝐧𝐢𝐧𝐠 & 𝐕𝐚𝐥𝐢𝐝𝐚𝐭𝐢𝐨𝐧 – Auto-fix what you can; escalate what you can’t 4. 𝐔𝐧𝐢𝐟𝐨𝐫𝐦𝐢𝐳𝐚𝐭𝐢𝐨𝐧 & 𝐒𝐋𝐀𝐬 – Define “good enough” and enforce it before data is consumed Why it matters: ✅ Data teams – Less cleansing, faster delivery ✅ AI models – Reliable inputs = repeatable results ✅ Ops teams – Catch failing sensors before downtime ✅ Business – Avoid safety incidents, billing errors, bad decisions In the latest episode of the AI in Manufacturing podcast, I sat down with Bert Baeck, Co-Founder of Timeseer.AI, to discuss time-series data quality and reliability strategies for AI in manufacturing applications.

What really makes data and AI innovation teams succeed in manufacturing? In this episode of the AI in Manufacturing Podcast, I speak with Van Tucker, VP of Harbor Lockers by Luxer One, a company that develops and manufactures smart public lockers. We discuss the challenges and strategies for building effective innovation teams in manufacturing. Here are some of the  insights that Van shared: 𝐂𝐮𝐥𝐭𝐮𝐫𝐞 𝐢𝐬 𝐭𝐡𝐞 𝐟𝐨𝐮𝐧𝐝𝐚𝐭𝐢𝐨𝐧 Innovation thrives when people, from the boardroom to the factory floor, believe in the mission. Core values must be lived daily, not just written on posters. 𝐀𝐠𝐢𝐥𝐢𝐭𝐲 𝐨𝐯𝐞𝐫 𝐩𝐞𝐫𝐟𝐞𝐜𝐭𝐢𝐨𝐧 Instead of waiting months for a polished rollout, start simple. Test small ideas quickly, gather feedback, and iterate. Even in hardware manufacturing, lightweight R&D “sandboxes” allow experimentation without disrupting core production. 𝐌𝐚𝐧𝐚𝐠𝐢𝐧𝐠 𝐩𝐫𝐞𝐬𝐬𝐮𝐫𝐞 𝐚𝐧𝐝 𝐛𝐮𝐫𝐧𝐨𝐮𝐭 Burnout shows up in declining quality and disengagement. The best leaders don’t wait, they stay close to their teams, recognize early warning signs, and act before problems escalate. 𝐁𝐫𝐢𝐝𝐠𝐢𝐧𝐠 𝐈𝐓 𝐚𝐧𝐝 𝐎𝐓 The old silos are gone. Effective leaders create environments where engineers from IT and OT collaborate, not compete. Quick collaborative wins build trust and momentum across functions.

Most industrial processes still run on the same foundation: - Hard-coded logic in PLCs that follows predefined rules. - The intuition of process and plant engineers, built from years of experience. This combination has powered industry for decades, but it has limits. When the challenge involves many interacting variables, unknown relationships, and non-linear effects, traditional control starts to strain. Why? Because fixed rules can’t adapt fast enough to changing conditions, and even the best human intuition can only process so much complexity at once. Instead of relying on fixed instructions, RL agents learn directly from real-time feedback. They can: ✅ Adapt continuously to new conditions. ✅ Handle high-dimensional problems with countless variables. ✅ Uncover novel, more efficient strategies that humans might overlook. The result? An optimization layer that works alongside your existing control system, making it smarter, more adaptive, and capable of delivering gains where complexity used to be a roadblock In the latest episode of the AI in Manufacturing podcast, I sat down with Dr. Kyrill Schmid, the Lead AI Engineer at MaibornWolff GmbH, to discuss the application of reinforcement learning agents for optimizing industrial plants.

Can AI agents really make decisions in high-stakes industrial environments? Generative AI agents, on their own, do not have a robust understanding of cause-and-effect for real-world decision-making. However, when combined with Deep Reinforcement Learning, AI agents gain the ability to reason, learn from interaction, and make decisions that solve operational problems in complex, real-world environments, like the plant floor. Case in point. Bryan DeBois and his team at RoviSys developed an Autonomous AI agent to manage a notoriously difficult glass bottle production process, where small disruptions like temperature fluctuations can quickly push the process out of specification. Here’s how they approached it: ✅ 𝐒𝐭𝐞𝐩 1 - 𝐌𝐚𝐜𝐡𝐢𝐧𝐞 𝐓𝐞𝐚𝐜𝐡𝐢𝐧𝐠 They captured the knowledge and decision-making strategies of expert human operators and used this to train the AI agent, essentially teaching it how to respond to different operating conditions. ✅ 𝐒𝐭𝐞𝐩 2 - 𝐃𝐞𝐜𝐢𝐬𝐢𝐨𝐧 𝐒𝐮𝐩𝐩𝐨𝐫𝐭 𝐌𝐨𝐝𝐞 Initially, the agent didn’t control the process directly. It simply made recommendations. Operators reviewed the suggestions and gave feedback using a simple green/red button system. This built trust and allowed the team to validate the AI’s decisions without risk. ✅ 𝐒𝐭𝐞𝐩 3 - 𝐂𝐥𝐨𝐬𝐞𝐝 𝐋𝐨𝐨𝐩 𝐂𝐨𝐧𝐭𝐫𝐨𝐥 Only after months of successful operation in support mode did they enable full automation. Even then, strict safety measures were in place: ⇨ Limited control authority ⇨ Clearly defined operating boundaries ⇨ Automatic handover to human operators if conditions exceeded the agent’s training The Results: ⇨ Human operators typically needed 7–20 minutes to bring the process back into spec ⇨ The AI agent consistently did it in under 5 minutes ⇨ And it maintained safety by operating strictly within validated limits In the latest episode of the AI in Manufacturing podcast, I sat down with Bryan, Director of Industrial AI at RoviSys, to dive deeper into how manufacturers can leverage AI and autonomous agents to optimize manufacturing operations and improve efficiency