About
Most manufacturers don't have an AI problem. They have a "random acts of digital" problem…
Articles by Adrian
Activity
11K followers
-
Adrian Pask shared thisGood read from Greg Riche. The line that sticks with me is that "narrowness is a form of safety". Most of the AI I see on plant floors is pretty boring. A camera checks welds. A model flags a part that's out of spec. It cant do anything else because nobody trained it to, and that's how it should be. The scary scenarios need a model with broad reach and nobody deciding where it stops. In manufacturing, deciding where it stops is the job.
-
Adrian Pask reposted thisAdrian Pask reposted thisWe're Hiring! Join the Accenture Next Gen Robotics Team Exciting news — the Next Gen Robotics team is expanding across North America and India, and we're looking for talented individuals to join us! If you're passionate about pushing the boundaries of robotics and want to work on cutting-edge technology that's shaping the future, this is your opportunity. We're building a team of innovators, problem-solvers, and forward-thinkers who are ready to make an impact. Why join us? - Work on next-generation robotics solutions at the forefront of Physical AI: Humanoid, new Physical AI models training (VLM, VLA, ...), Digital Twin, new robotics platforms, drones, ... - Collaborate with a diverse, multidisciplinary team across the world - Be part of a growing organization with real career development opportunities We're actively looking for candidates across a range of roles and experience levels. Whether you're an engineer, researcher, or operational specialist — we want to hear from you. Ready to be part of something big? Stay tuned for open positions, or reach out directly to learn more. The future of robotics starts here: North America : Senior Leaders : https://lnkd.in/eHQ5iy4X North America : Next Gen Robotics Engineer : https://lnkd.in/euCC6ymk India : Senior Leader : https://lnkd.in/ejbGeHDX #PhysicalAI #Robotics #Accenture Adam Burden Prasad Satyavolu Maria Danninger Falk-Moritz Schaefer Alen Petrunin William C. Mary Hamilton Elias Treis Henrique Almeida e Sousa Victor Chevalier Michaël Guerrier Edouard Mathon
-
Adrian Pask reposted thisCome join a fast-paced, innovative business in a growing market!Adrian Pask reposted thisCould the work you do every day help more people get better care? At Philips, we’re hiring in Plymouth, MN across teams that bring healthcare innovation to life - from R&D, software and data science to quality, service, clinical, operations and project roles. Different expertise. One shared purpose: helping deliver better care for more people. Explore open roles in Plymouth and apply today 👉 https://lnkd.in/e8pPrsWT
-
Adrian Pask shared thisExcellent explanation about the value of knowledge graphs in manufacturing (and really valuable context about how they relate to your UNS architectural approach).Adrian Pask shared thisI’ve spent a lot of time over the last couple of weeks playing with knowledge graphs... and specifically with Flow Software's Atlas. Here’s the thing -- manufacturers need to start paying attention. We’ve spent years solving the connectivity problem. Connect the PLCs, databases, MES, ERP, historians... then organize that information so we can see the current state of the business, and UNS does that really, really well. But connected data is not the same thing as understood data... and that’s the next problem. It’s one thing to know that two things are related. It’s another thing to know HOW they are related -- and the how is where meaning comes from. A UNS can tell me Pump 7 and Line 3 exist in the same infrastructure. A knowledge graph can tell me Pump 7 FEEDS Line 3... Motor 12 DRIVES Pump 7... Order 4471 WAS PRODUCED ON Line 3. That is a completely different thing. The unified namespace is basically 2D plumbing -- parent, child, sibling... structure, current state, events, context. A knowledge graph lets me extend those relationships in three dimensions and apply meaning to them. This is where it gets really important for the manufacturer. The operator knows what that temperature means. The maintenance engineer knows what usually happens before the bearing fails. The process engineer knows that what happened three steps upstream may be the reason quality is out right now. Most of that knowledge still lives in people’s heads... or an SOP... or a spreadsheet. That knowledge is incredibly valuable, but it isn’t queryable. We do this naturally. Agents don’t. An agent needs something to traverse. It needs the relationships. It needs the meaning. It needs a scaffold that tells it where to go and what matters so it can get to the right context and reason through the problem. Flow has built a type-based knowledge layer that gives us the ability to apply meaning and knowledge to the infrastructure... and make it usable by people AND agents. Think about what that means -- tribal knowledge becomes queryable, root cause can be traced across systems and agents can navigate relationships to find the context they actually need. That is a big deal. And I want to be really clear because this question keeps coming up -- knowledge graphs do not replace Unified Namespace. They work together. UNS gives you the live, semantically organized flow of information and current state of the business. The knowledge graph gives you the map of meaning across that flow. At the end of the day, this is all about turning data into better decisions... helping operators understand what is happening faster, helping maintenance get to root cause faster and helping organizations stop losing knowledge every time someone retires or changes jobs. The next phase of digital transformation is not about connecting more data... it’s about making the data we already have understandable to people and machines. 🙏
-
Adrian Pask posted thisTechnology adoption requires some part of the operating model to become deliberately unstable for a period of time...which can sit awkwardly with how we normally think about Operational Excellence. Most OE systems are designed to remove instability by defining processes, establishing standard work, reducing variation, make abnormalities visible, and improving from a known work-process baseline. Technology experimentation and adoption asks us to do something different. For a period of time, we may need to challenge standards, run processes differently, change roles, or test whether a decision can happen somewhere else entirely. That instability can make perfect sense to the project team. It can feel very different to the person running the process every day. To an operator, the third pilot in twelve months may not feel like a carefully controlled experiment. It may just feel like another new screen, another temporary procedure, another change that will probably disappear when the project team moves on. That puts a much bigger responsibility on leadership than simply sponsoring the technology. People need to know what is changing and what isn't. They need to know why the experiment exists. How long it is expected to run. What success looks like. And, importantly, what happens when it's over. Did we adopt it? Did we change it? Did we kill it? What is the standard now? Without that closure, temporary instability starts to become the operating model. I also think that the pace of technology change makes this approach ESPECIALLY important. E.g. If a technology can remove 60% of a loss today, it may be better to implement that, learn from it, and improve again than spend two years designing for the perfect end state. Especially when the technology itself may have changed twice by then. The irony is that we often add complexity in the name of future-proofing, when the pace of change makes true future-proofing almost impossible. We need to be good at making bounded changes, explaining them clearly, learning quickly, and then restoring stability around whatever we decide to keep. That feels like a very different definition of Operational Excellence. #OperationalExcellence #ChangeManagement #Manufacturing #DigitalTransformation
-
Adrian Pask posted thisWhy do connected worker platforms exist? Manufacturing needed a layer that could evolve at the speed of frontline work, but MES was evolving at the speed of enterprise systems. That gap created a market MES was built to run production reliably across increasingly complex processes. As it became more capable, it also became harder to deploy, configure and change. Once stable, even relatively small frontline improvements could trigger governance, integration and release cycles That made bolt-on solutions attractive Then tablets, wireless networks, cloud platforms and low-code development changed the economics. Manufacturers could improve work instructions, workflows and data capture without reopening the execution backbone running the plant Connected-worker platforms turned that workaround into a category Now the market is moving again: - MES platforms are becoming more modular, open and better at decision support - Connected-worker platforms are moving deeper into execution - Performance and OEE tools are adding workflow and production capabilities The boundaries are now very blurry My perspective is that: - A stable, repetitive operation may choose to consolidate most frontline workflow into MES - A high-mix or NPD-heavy environment may still value a separate application layer that can change in days rather than months - Specialist applications will survive wherever they solve a specific problem exceptionally well Across a 40-plant network, you'll probably have all three, which creates a different architectural challenge.... ...the question is no longer: Which platform should own frontline work? ...the question is: How do we coordinate work across all of them? That requires enough common meaning across the estate for systems to work together: - Shared definitions of assets, people, work and events - Common rules around authority, qualifications and decision rights - The ability to move context and actions between systems Because work increasingly spans multiple domains: - ERP and planning systems change production commitments - MES executes production - CMMS executes maintenance - Connected-worker platforms guide people - AI agents increasingly recommend or initiate actions within each domain So who coordinates all of that? - Who decides what should happen next? - Who is authorized to make that decision? - How does that decision move across the systems already running the operation? THAT is where I think the real battleground is forming!! Connected-worker platforms emerged because manufacturers needed a faster, more adaptable layer around frontline work. The next phase is orchestration across the entire operating environment The winner may not be the platform that replaces MES (I don't think it gets replaced) It's the platform that makes MES, frontline applications, CMMS, AI agents and people operate as one system #ConnectedWorker #MES #IndustrialAI #OperationalExcellence #DigitalManufacturing #ManufacturingTransformation
-
Adrian Pask posted thisI’m starting to see “human in the loop” as a convenient way of postponing one of the harder questions in AI adoption..."what happens when the system is good enough to make the decision?" We’ve become (maybe too) comfortable when AI generates an insight, content, or recommends an action. But how do we act when it reaches a decision that is currently part of standard work: whether to release a quality hold, change a maintenance priority, or override a production schedule? During design we say "we'll start with human-in-the-loop" and run both. The system recommends, the person reviews, and the human retains final authority. That makes sense to validate a concept but it also becomes a very bad permanent operating model. Imagine: 1. The quality system recommends releasing a hold and the quality engineer disagrees. 2a. The engineer overrides it and the batch is held. Fine. Now flip it. 2b. The engineer follows the recommendation, the batch ships, and there’s a problem downstream. Who made the decision? Was the system wrong? Was the engineer wrong for trusting it? Was the process wrong because the engineer was still expected to independently verify a decision the system was supposedly good enough to make? Parallel decision authority sounds safe, but it also creates a gap in accountability. We’ve been handing decisions to deterministic control systems for decades. A PLC doesn’t ask an operator whether it should execute every interlock. Once the logic is validated, the authority is designed into the system. Generative and Agentic AI is harder because the outputs are probabilistic and, in many cases, less transparent. But that does not removes the question...it's more urgent. At what point has the system earned the right to make the decision? And who actually gets to decide that? My suspicion is that this becomes an Operational Excellence problem as much as an AI governance problem, because ultimately somebody has to change the standard work, the decision rights and the management system around it. If the only thing the operating system knows how to do is protect the current standard, a very mature OE system can become very good at protecting the current operating model....and our humans remain in-the-loop. #OperationalExcellence #ArtificialIntelligence #Manufacturing #AIinManufacturing
-
Adrian Pask posted thisShould "technology adoption" be an Operational Excellence capability? I increasingly think the answer is yes but it creates an uncomfortable tension with the intent of most OE systems to create stability and resilience. So much of Operational Excellence focuses on the definition and delivery of standard work...particularly at machine-into operator-into engineering. The best programs also define standard work for leaders; but definitionally this becomes more ambiguous at higher levels. When we adopt new technology, we normally layer it onto of our standard work....a new dashboard, and improved report, an automated computer vision system. But big technology adoption is hard because we need to experiment. Reimagine workflows. Move decisions from people into software. Ultimately remove some of the processes that are core to the analogy OE methodology. A very mature Operational Excellence system can become extremely good at protecting the current operating model. But technology adoption requires some part of that operating model to become deliberately unstable for a period of time. I don’t mean abandoning process control or experimenting carelessly on a production line. I mean having the capability to identify where the current standard needs to be challenged, create a controlled environment to test something different, learn quickly, and then absorb the new way of working back into the management system. Take condition monitoring. Bolting a sensor onto a critical asset and generating an alert can be done relatively quickly. The OE work starts when that alert proves useful. - Does the inspection route change? - Does a maintenance decision move earlier? - Does the technician's role change? - Which standard gets rewritten? - Which existing control is now redundant? - Do you have the skills to improve the ML model? And how quickly can the organization make those changes without losing control of the process? I'm suggesting that not only does our concept of "Operational Excellence" need to change, but maybe our concept of "resiliency" in 2030 means something slightly different from resilience today. Not simply the ability to hold a known state through disruption. The ability to deliberately change that state when a better way of operating becomes possible — and then make the new way stable. #OperationalExcellence #LeanManufacturing #Manufacturing #DigitalManufacturing #AIinManufacturing #ContinuousImprovement
-
Adrian Pask shared thisOutstanding opportunity to join an incredible company with wonderful leadership!Adrian Pask shared thisAt The Coca‑Cola Company, our ability to deliver safe, reliable, and innovative equipment solutions is critical to fueling growth and ensuring extraordinary brand experiences globally. Equipment innovation is central to how we design, build, and scale the next generation of beverage equipment, creating exceptional consumer experiences. I am looking for curious, collaborative, and forward-thinking talent who want to help build solutions that improve speed, reliability, compliance, and scalability across a global equipment ecosystem. If you are passionate about innovation, operational excellence, digital transformation, and making an impact at scale, this is a great moment to join our journey. Explore our new Global Equipment Platforms roles and consider applying — or share with someone in your network who may be ready to help shape what’s next. Senior Director, Manufacturing Strategy and PLM https://lnkd.in/eU8ehRAU Director, Regulatory & Governance https://lnkd.in/eKWXBQs2 Senior Manager, Operational Excellence https://lnkd.in/epmeVhEa Senior Manager, Product Lifecycle Management https://lnkd.in/enQSySj5 Senior Manager, CADD Designer / PLM Integrator https://lnkd.in/e7r55B4k
Adrian Pask commented on a post
3d
Jay Joyner congratulations on an incredible career. It's been a pleasure chatting with you!
-
Adrian Pask liked thisMaking communities a little sweeter!Adrian Pask liked thisWe are big fans of inspiring sweetness in our communities – and this one is really a touchdown! 🏈 In partnership with the Memorial Park District and the Village of Bellwood with Mayor Andre Harvey, we unveiled the community’s new scoreboard, sponsored with pride by Ferrara, this Homecoming weekend. Here's to many more years of sportsmanship, honest effort, and the kind of character that's built long before the final whistle! #inspiringsweetness
-
Adrian Pask liked thisAdrian Pask liked thisI'm excited to be speaking at The Connected Worker Manufacturing Summit in Chicago this October! To find out more, click here: https://lnkd.in/eU9hahUS #CWM #CWManufacturing
-
Adrian Pask liked thisAdrian Pask liked thisWe’re looking to round out our Site Leadership Team here in Orangeburg! Each role will play a critical part in the startup of our new factory, helping establish a strong foundation for operational excellence, safety, quality, and long-term success. We’re looking for leaders with a strategic mindset, strong operational expertise, and the ability to deliver results in a fast-paced startup environment. Please share this opportunity with anyone in your network who may be a great fit. We’d love to connect with talented leaders who are ready to help build something great from the ground up! Sr. Manager, Factory Supply Chain https://lnkd.in/gyVwxhYS Sr. Manager, Continuous Improvement https://lnkd.in/gVxmdPQB Sr. Manager, Quality amp; Food Safety https://lnkd.in/gV3Tuxqw
-
Adrian Pask liked thisAdrian Pask liked thisSomewhere, a perfectly good IF statement is about to become a six-month AI project. 😬 I get the excitement. Give me a new tool and I want to see what it can do, too. But once “use AI” becomes the objective, something funny happens... we start describing every problem in a way that makes AI sound like the answer. Suddenly, unclear responsibilities become a “decision intelligence opportunity.” A broken process becomes an “agentic workflow.” And nobody wants to suggest a checklist because, well, good luck getting a keynote out of that. 😂 Take manufacturing downtime. Three machines can be sitting idle for three completely different reasons: • One failed without warning. Better prediction might help. • One is waiting for a replacement part. The stocking policy needs attention. • One is waiting for someone to approve a restart. We should probably investigate WHY that takes four people and a meeting. Same metric. Different causes. Very different work required to improve it. AI could contribute to each situation, but we still have to understand what is preventing the result. A more accurate prediction has limited value if nobody can act on it. A faster recommendation doesn’t accomplish much if it joins the same approval queue. Before choosing the technology, I want to know what will happen differently when it works. Who takes action? What changes in the operation? How will we know it helped? And yes, sometimes AI is absolutely the right answer. Go build it!!! Just leave room for the possibility that the best solution is smaller, simpler, and considerably less exciting to announce. A screw doesn’t become a nail because the hammer already has budget approval. 😜 𝐑𝐞𝐚𝐝 𝐦𝐲 𝐟𝐮𝐥𝐥 𝐚𝐫𝐭𝐢𝐜𝐥𝐞: https://lnkd.in/e7dTek4f ***************************************** • Visit www.jeffwinterinsights.com for access to all my content and to stay current on Industry 4.0 and other cool tech trends • Ring the 🔔 for notifications!
-
Adrian Pask liked thisAdrian Pask liked thisWe're Hiring! Join the Accenture Next Gen Robotics Team Exciting news — the Next Gen Robotics team is expanding across North America and India, and we're looking for talented individuals to join us! If you're passionate about pushing the boundaries of robotics and want to work on cutting-edge technology that's shaping the future, this is your opportunity. We're building a team of innovators, problem-solvers, and forward-thinkers who are ready to make an impact. Why join us? - Work on next-generation robotics solutions at the forefront of Physical AI: Humanoid, new Physical AI models training (VLM, VLA, ...), Digital Twin, new robotics platforms, drones, ... - Collaborate with a diverse, multidisciplinary team across the world - Be part of a growing organization with real career development opportunities We're actively looking for candidates across a range of roles and experience levels. Whether you're an engineer, researcher, or operational specialist — we want to hear from you. Ready to be part of something big? Stay tuned for open positions, or reach out directly to learn more. The future of robotics starts here: North America : Senior Leaders : https://lnkd.in/eHQ5iy4X North America : Next Gen Robotics Engineer : https://lnkd.in/euCC6ymk India : Senior Leader : https://lnkd.in/ejbGeHDX #PhysicalAI #Robotics #Accenture Adam Burden Prasad Satyavolu Maria Danninger Falk-Moritz Schaefer Alen Petrunin William C. Mary Hamilton Elias Treis Henrique Almeida e Sousa Victor Chevalier Michaël Guerrier Edouard Mathon
Experience & Education
Publications
-
Agile
Deliver small chunks of business value in short release cycles. For each cycle, choose whatever will deliver the most value and get it done. Do away with complicated master plans and be evolutionary and adaptive.
Other authors -
Down Time Tips
http://vorne.com/improve/down-time-tips.htm
Down Time is the largest source of lost production time for most manufacturers. Here are some practical tips for reducing Down Time that don’t require a lot of outside support.
Other authors -
Improve the Constraint
Each manufacturing process has a constraint (bottleneck). Focusing improvement efforts on the constraint is the fastest path to improved productivity.
Other authors -
Leadership
Inspire others to do their best work and you leverage your most important resource – people. Leadership is a learned skill. Improve your effectiveness as a leader and you improve your ability to create and accelerate change.
Other authors -
OEE
OEE measures how close you are to perfect production (manufacturing only good parts, as fast as possible, with no down time). It is a great metric for benchmarking progress in improving manufacturing productivity (i.e., eliminating waste).
Other authors -
Short Interval Control
Quick and focused reviews of performance data during the shift enable ongoing course corrections and small-scale fixes that collectively result in significant improvements in performance. Look at the immediate past to guide actions for the immediate future.
Other authors -
Single-Minute Exchange of Dies
Changeover times can typically be reduced to less than 10 minutes (i.e., single-digit minutes). Analyze each element of the changeover to see if it can be separated (moved external to the changeover), converted (modified to be external), or streamlined.
Other authors -
Top Losses
Time is precious, so spend that time fixing the most important things — your top sources of lost manufacturing productivity. Track your Top Losses. Pick the biggest actionable loss. Fix it. Repeat.
Other authors -
IFA (Information x Focus x Actions) = Results
Most companies place too much emphasis on results (looking backward) and too little emphasis on the factors that drive results: Information (the basis for effective decisions), Focus (making effective decisions), and Action (transforming decisions to results).
Other authors -
Vorne: Six Big Losses
Foodproductiondaily.com
Vorne Industries has created a series of Improvement Topics on operational efficiency, which are free to download in pdf format on the internet and has already attracted 24,000 views.
Other authorsSee publication
Recommendations received
8 people have recommended Adrian
Join now to viewOther similar profiles
Explore more posts
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content