Industrial AI can help reduce production losses -- but there's something you need to know

It’s impossible to miss the rapid rise of Artificial Intelligence in manufacturing, or Industrial Artificial Intelligence. It seems like every other technology-related article -- and practically every single vendor vying for our attention -- uses terms like AI, machine learning, digitization, automation, Industry 4.0, Industrial AI, etc... almost as a form of punctuation.

As a result, manufacturers' views of AI range from unrealistic (“AI can solve all our problems!”) to skepticism any time the words “Artificial Intelligence” are even uttered.

But, as with any technology, the truth is really somewhere in between.

AI can be extremely effective -- in the right context.

That’s the key. Understanding those contexts, and the kinds of AI technology that are relevant to them, is the only way to set realistic business goals for AI adoption.

Introducing The Industrial AI Quadrant

Watch: How to select the right Industrial Artificial Intelligence solution:

As a rule of thumb, AI works best when it is applied to solving a specific problem, or a very closely-related set of problems.

Artificial Intelligence is not a silver bullet; no solution will solve all, or most, of your problems. “General AI” is something to be wary of: if a vendor claims to do everything, then they probably do nothing particularly well. When it comes to AI in manufacturing specifically, there are numerous potential applications and use cases, each of which requires a unique type of Artificial Intelligence.

Sounds simple enough, right?

There’s just one problem: most manufacturing executives don’t have a PhD in data science. Even if they do, they certainly don't have the time to pour over the technicalities of whether each vendor is using the optimum type of algorithm.

Fortunately, you don’t need to do that. Seebo’s Chief Data Scientist has invented a highly effective methodology to select the right Industrial AI solution to address any given manufacturing challenge. We call it “The Industrial AI Quadrant”.

The Industrial AI Quadrant:

The focus: reducing production losses that hurt your bottom line

The Industrial AI Quadrant focuses specifically on perhaps the most significant application of Industrial Artificial Intelligence: reducing the losses incurred during the manufacturing process -- things like quality, yield, waste and downtime. These losses hurt manufacturers’ bottom line and are a major source of headaches.

To address this challenge, manufacturers are increasingly turning to Machine Learning -- a specific category of Artificial Intelligence -- which can provide insights, predictions and recommendations to prevent these losses.

Of course, there are multiple types of losses, occurring to varying degrees and with varying frequency across different industries. “Machine learning” is a broad term (despite being a sub-category of the even broader term “Artificial Intelligence”); there are different types of Machine Learning, which are each suited for a particular type of problem.

Process-driven vs. asset-driven

One of the most common crossroads when choosing Industrial Artificial Intelligence technologies to address production losses is the question of where those losses are rooted.

Asset-driven losses, for example, are a common cause of losses due to downtime. In recent years, a variety of Predictive Maintenance solutions have arisen, which address this specific question using Unsupervised Machine Learning: when is my machinery going to break down or otherwise cause losses?

But the AI technology used for Predictive Maintenance -- while highly useful in that context -- isn't suited for solving a different pressing challenge many manufacturers face: process-driven losses; things like quality, waste, yield and throughput losses. These types of losses are caused by inefficiencies within the production process, rather than asset-related problems. Since many manufacturing processes are highly complex (particularly in the world of continuous process manufacturing), process experts and engineers can't possibly figure out all the root causes of these inefficiencies themselves.

To address these process-driven losses, manufacturers need an altogether different kind of Machine Learning AI. -- one that understands the unique complexities of each individual production line, and can deliver the insights production teams need to prevent production losses. That's what Seebo's proprietary Process-Based Artificial Intelligence is designed to do (see here for more details).

Choosing wisely

Leading manufacturers are already using the Industrial AI Quadrant as a simple but effective methodology to navigate this complex ecosystem, and zero-in on the technology that can solve their specific business problems.

Download the guide to discover how:

Case Study: How a Global Chemical Manufacturer Reduced Quality Losses from Toxic Side-Products by 65%

Quality and yield losses are a perennial problem for most chemical manufacturers; from quality variabilities, to impurities, to incomplete reactions, to losses during separation and purification -- to name just a few. These production losses are costly, hurting manufacturers’ bottom line and sucking up previous time and resources from their manufacturing teams.

What if you could prevent these losses from occurring in the first place?

That’s what one global manufacturer of Ethylene Dichloride managed to do -- saving nearly one million euros each year on a production line that, until then, had been suffering significant annual losses.

The primary problem was a toxic side product (trichloroethane) that kept forming during the production process. Despite the best efforts of the company’s process experts, the cause of this process inefficiency remained frustratingly elusive for years.

The limitations of generic AI and analytics tools

Part of the problem was that the production process in question was relatively complex, with some 4,400 data tags. The analytics tools used by their process experts, advanced as they were, still limited them to conducting ad hoc analyses of a select number of tags. This was of little help, since once their existing theories were exhausted they had no idea where to continue looking for the root cause of the problem!

Due to the significant financial impact of this inefficiency, the company decided to invest in Industrial Artificial Intelligence.

But attempts to use generic Artificial Intelligence solutions repeatedly failed to yield useful or accurate results, due to the complexity of their production process. These solutions weren’t made specifically for continuous manufacturing, and were therefore unable to cope with the unique complexities of the process and resulting data.

As their VP Manufacturing noted:

“These vendors touted some powerful AI technology. But continuous chemical manufacturing processes like ours produce uniquely complex and messy data that their algorithms simply didn’t understand. So the end results either made no sense or were clearly inaccurate.”

The solution: Process-Based Artificial Intelligence with embedded process expertise

Things finally turned around when they turned to Seebo.

Seebo’s proprietary Process-Based Artificial Intelligence is designed specifically to solve complex continuous manufacturing process inefficiencies. First, Seebo’ automated root-cause analysis reveals the hidden causes of their losses. Next, the Seebo solution provides predictive recommendations to reach their optimal process settings to avoid losses and maximize capacity. Finally, proactive alerts enable their manufacturing teams to prevent losses before they occur.

Using the Seebo solution, the company’s process experts were finally about pinpoint the primary causes of their yield and quality issues, as were armed with the capabilities to maintain optimum process settings and prevent similar problems in the future.

The results were impressive: a 65% reduction in the amounts of toxic side products - which translated into annual savings of one million euros for that single production line.

Download the full case study to learn how they did it.

2 Trends Manufacturers can Harness to Turn the Coronavirus Crisis into an Opportunity

Over the past half a year the manufacturing industry -- like most other industries -- has been focused on how to cope with the Coronavirus crisis.

This has meant adapting to new, challenging realities like remote working, supply-chain disruptions, more stringent health and safety requirements and so on. It has also meant adapting to sudden, major shifts in consumer behavior for the short term: for example, the collapse in demand for the automotive industry (which is already on the road to recovery -- pun unintended); or the rocketing demand for hygiene products; or indeed the radical shifts in demand within the food industry, as people panic-bought, stocked up on non-perishable goods, and generally changed where, how and what they ate, particularly during lockdown.

But the current public health crisis is about more than the immediate sense of "crisis" itself, and the major, yet largely transient changes in consumer behavior it triggered. Indeed, many of the early trends have all but faded away (at least in many places), as populations, businesses and governments learned to adapt. However, some longer-term changes are emerging that manufacturers would do well to recognize, in order to gain a competitive edge.

Accelerated evolution of the manufacturing industry

Once such change is to the manufacturing industry itself. In a recent interview with FoodNavigator, Seebo CEO Lior Akavia explained how the crisis trigger a wave of technological innovation among manufacturers.

Meeting accelerated demand, even in the short-term, can have long-term impacts on a company's market share -- be it food, hygiene products, over-the-counter medicines, or anything else. A recent study by PwC revealed that during a crisis, consumers are more likely than ever to switch brands based on availability alone. Simply put: if they can't find their favourite brand on the shelf, they'll just grab the next one. More significantly still, the study shower that they are likely to stick to that new brand in the long-term. This makes it all the more urgent that manufacturers meet demand during a crisis.

Paraphrasing Akavia, the Food Navigator article noted how, in the face of unprecedented spikes in demand for certain products (both short-term and longer-term):

"...manufacturers were left with two choices: to either build new production lines to meet demand -- but 'this takes years' and 'huge investments', said Akavia -- or optimise their processes, which can 'quickly translate into increased capacity.'"

"Optimization" in this context means reducing the process inefficiencies that lead to production losses; e.g. waste, quality, yield, throughput and so on. If manufacturers are to meet rising demand in a scalable way, every ounce of productivity and efficiency counts.

This wave of innovation is still ongoing, and it's clear that manufacturers who successfully adopt the right technologies and methodologies to thrive in a crisis environment, will emerge with a significant advantage over those that do not, even once the crisis is firmly behind us.

As a recent paper by Mckinsey notes, the COVID-19 crisis “creates an opportunity to reimagine the way work is done”:

The coronavirus will have long-lasting—perhaps permanent—effects on manufacturing organizations, forcing companies to restructure their operations to maintain production while protecting their workers…

...By accelerating the adoption of new digital technologies and by drawing on the flexibility and creativity of their frontline staff, companies have the opportunity to emerge from the crisis with manufacturing operations that are safer, more productive, and more resilient.

Crises and challenges -- economic, natural or otherwise -- can always occur. Those manufacturers who respondwell to the current crisis will also be best equipped to cope with and emerge stronger from each subsequent crisis.

Changes in consumer behavior

There have also been many important changes to consumer behavior, many of which are likely to have long-term impacts which -- if correctly identified -- can offer an opportunity for manufacturers to capitalize on.

For example, demand for baking ingredients doubled during the height of the crisis, according to a General Mills study. This can be attributed to a variety of factors: people seeking comfort foods to deal with stress and anxiety, or abandoning diets for similar reasons. Still others had to find ways to fill additional spare time stuck at home, not to mention find new ways to occupy their children. This appears to be part of a wider trend of people preparing more of their own meals -- particularly at dinner time, in part to recreate the “restaurant” experience they were missing. As restrictions are lifted, people are indeed returning to the restaurants and cafes they love -- but will many of these people continue their newfound cooking hobbies as well? How will that impact long-term demand?

Demand for salty snacks also increased for similar reasons, as a coping mechanism for stress and boredom. For example US demand for potato chips grew by 30%, while sales of popcorn and pretzels rose 47%. Similarly, sales of instant coffee have surged worldwide. Again, consumer behavior will likely moderate once the crisis ends, but by that point, will they have developed a newfound taste for these (somewhat addictive) products?

It’s impossible to know for certain how these trends will feed into post-crisis consumer habits. But clearly those manufacturers who fill that demand now will be best positioned to capitalize on any long-term impact.

Of course, these trends aren’t limited to the food industry. For example, the crisis has triggered higher hygiene awareness globally. While the current levels of hyper-awareness aren’t sustainable, it is reasonable to assume that many people will come out of the crisis more conscious about cleanliness -- whether at home or in the workplace. Demand for certain cleaning products will therefore probably remain higher than before as well.

Indeed, a report by Fortune Business Insights forecast that the "Hand Wash" industry will reach $4.56 billion by 2027, as demand rises significantly. In great part this demand is being driven by "increasing government initiatives to promote the usage of personal care products", as well as by "regulatory bodies and NGOs [who] are also encouraging the usage of hand wash and sanitizers through social media platforms, websites, print media, television, and radio." Another report forecast a similar rise in demand for the hand sanitizer market -- which is expected to reach $1.87 billion by the end of 2020.

According to that report:

"The hand sanitizer market size was anticipated to reach USD 1.35 billion in 2020 before the Covid-19 outbreak. However, on account of the present scenario, it is likely to generate USD 1.87 billion this year. In addition to this, it would rise tremendously from an annual growth rate of 5.06% to 45.71% in 2020." [emphasis added]

The takeaway for manufacturers: now is the time to turn crisis to opportunity

The Coronavirus crisis won't last forever (even if it sometimes feels that way). When the dust settles, many things will return to the way they were before.

But it's clear that for manufacturing, things will never be quite the same. The accelerated pace of innovation -- especially in the area of digitization; the long-term changes in consumer behavior; the rise and fall of various products and even industries; all of these factors are coalescing to form a new reality that is more efficient, more demanding and far more competitive.

Those manufacturers who adapt and innovate now, will emerge from 2020 with a clear competitive edge.

Download our free white paper below to learn more:

Root Cause Analysis in the Age of Industry 4.0

Every manufacturing process has its inefficiencies — and these inefficiencies often lead to losses that harm their bottom line.

These losses take many different forms: be it from yield, quality, waste, throughput, energy, emissions, factory downtime, or something else. Usually, the more complex the process, the more complex the problems tend to be — and that’s where root cause analysis (RCA) comes in.

Process experts or process engineers use root cause analysis to trace a particular loss to its cause, in order to eliminate it. Of course, while that sounds pretty straightforward, the reality is often quite different.

Why is it so hard to find the root cause?

Sometimes, it’s relatively simple to solve a particular problem – particularly when that problem stems from a single cause, or a factor that’s easy to spot. For example, if the temperature at a particular point in the line strays from the permitted range, then a process expert will usually be able to identify and solve the problem themselves fairly quickly.

But very often this is not the case. Sometimes, the problem is caused by a complex combination of factors, making the root cause more difficult to understand. This is particularly true if the cause is rooted in the interrelationships between numerous tags, and their place within the process. This can mean no single tag is behaving problematically, but that the problem is rooted in the process itself – for example the speed at which the raw material travels from Tag A to Tag B; or its temperature or concentration at a specific point within the process; or the pressure it is subjected to between point A and point B, and so on. The options are endless.

Then of course there is the fact that the problem identified is actually just a symptom of a more fundamental issue, and the root cause may actually be the second or third derivative. Perhaps the issue began further upstream in the production process, and only became noticeable later on.

In any of these scenarios, nothing is amiss to the naked eye. This is why most manufacturers simply come to accept a certain amount of production losses: they’ve applied their smartest, most talented process experts and advanced tools to the problem, and simply couldn’t see what was wrong. What else can they do?

Shortcomings of traditional root cause analysis

As mentioned above, the general approach currently used by many manufacturers when it comes to root cause analysis is to rely on on-site expert knowledge, aided by a range of analytics tools.

Experience and process expertise is of course invaluable.

The problem is, for many complex processes, it isn’t humanly possible to analyze all the combinations of all the data tags on a production line, all the time.

In our conversations with hundreds of manufacturing executives, including many from leading global brands, the same limitation was consistently highlighted: advanced analytics are excellent for validating existing theories, but are much less useful for discovering the hidden root causes of persistent production problems. That’s because, even with the most sophisticated such platforms, there is always an inevitable blind spot, as the process expert needs to select a handful of tags based on their own human intuition and biases.

It’s natural that even experts can be biased towards certain ideas. Even if the root cause of the problem is roughly identified, there may be inaccuracies in the definition of the problem, making it difficult to come up with an intelligent and lean solution.

Other disadvantages of manual root cause analysis include:

  • Often, most RCA information isn’t shared across manufacturing sites, as manual analysis doesn’t scale. This leaves factories of the same company – or even individual lines within the same factory – to repeat each other’s mistakes, leading to losses that could have been avoided.
  • Manual RCA is conducted on an ad-hoc basis. But as manufacturing processes are dynamic, the data is constantly changing, so the analysis can quickly become redundant.

The Power of Automated Root Cause Analysis

Automated root cause analysis harnesses the power of Machine Learning — a subfield of Artificial Intelligence that focuses on developing and researching algorithms that learn from data. The algorithms exist in the form of models which are trained with historical data in a way that allows them to make predictions and decisions based upon new data.

Thanks to significant advances in machine learning and Big Data analytics, root cause analysis can be performed using automated methods. These methods are unbiased and based purely upon historic and real-time data from the production floor, infused with process expertise (more on that later).

Just as importantly, they take the Sisyphean task of analyzing and interpreting data away from the people on the factory floor, thereby enabling them to focus on actually optimizing the processes and improving performance.

Download the complete white paper to learn how Automated Root Cause Analysis works, and the real impact it’s having on manufacturers just like you.

Watch: How to Embed Manufacturing Process Expertise into Artificial Intelligence

For Industrial Artificial Intelligence to have a real impact in solving production problems, the algorithms must have a deep understanding of your production process. Watch our latest Vlog below to learn how it's done - and why it's so important.

WATCH: Talking Smart Manufacturing -- The Predictive Quality & Yield Series: Vlog #2


[Eden] Hi everyone, it's Eden from Seebo. Welcome again to our V blog. Our subject for today is how to embed process expertise into artificial intelligence and why it's so important. Here with me again is the co-founder and COO of Seebo, Liran Akavia.

[Liran] Hello, Eden.

- Hi, Liran. Liran is an expert in the field of industrial artificial intelligence and reducing manufacturing losses, and has extensive experience working with the world's leading manufacturers. Liran, we all know that process expertise is important. This is the key to meaningful process analysis. Many manufacturing executives that speak with us are asking, "Can algorithms solve my process inefficiencies when there is no process expertise in the algorithm?"

- So it's a true concern. Honestly, algorithms without process understanding will have no value to the manufacturing team. Just algorithms to analyze a production line will provide insignificant results.

- Okay, Liran, I get that. So can process expertise be embedded into the algorithm?

- In short? Yes. And I'd like to show you something. In my opinion, it's very interesting. I'd like to speak a little bit about how the algorithm is thinking about the data that is coming from the process and what is the big difference that is made when the algorithm actually understands the process. Let me take you through a quick journey.

- Okay.

- So a traditional algorithm, let's say just for this example, we look at three temperature data sets or tags. This is temperature data number 532, and this is temperature data number 667, and temperature data number 780.

A normal algorithm will simply look at this data and will try to understand if it's within the value or a little bit about its behavior, and in most cases, will come back with no meaningful results. Now, let's take a look on how Process-Based Artificial Intelligence is looking at the data, or how algorithms that actually understand the process look at the same piece of data.

So just for this example, let's say that there are three assets, asset A, B and C, and now when we understand the assets, we also know the physical order of those temperature sets. So we know that 780 is the first one and 532 is the second one physically, like the material is going first through asset A and then through asset B. With that knowledge, we can suddenly make a lot of beautiful calculations. For example, we can calculate the difference in time and material flow between those two temperature sensors, and by using this data, we can calculate the delta T, in this example, how much the material got hotter or colder during the time that it went between 780 to 532.

Related Video: How to Select the Right Kind of AI to Solve Your Manufacturing Losses

With that, we can calculate the first derivative that tells us the beautiful story of how the temperature is increasing or decreasing. Now with that value, we can maybe realize that the problem is not the absolute temperature parameter. The problem is the speed of increase and speed of decrease. So this is just one example of why for the algorithm to understand the plant topology and the flow of material is so dramatically important.

- Okay, Liran, that sounds really interesting, but very difficult in theory. Every production line is unique, so do you need to reinvent the wheel every single time?

- So you're right again. This is extremely difficult to do that. And when we built Seebo, we were concerned exactly because of that.

On the one hand, we put for ourselves a clear goal to provide value to manufacturers reducing the process inefficiencies. On the other hand, we understood that algorithms with no process expertise within them will be meaningless. And on the other hand, we understood that every production line is different.

So you actually need a tool or technology that is able to take a flow diagram, a P&ID, a production description, and somehow inject it into the algorithm, and I'd like to quickly show you how it works and what makes the Seebo technology so unique in understanding the actual manufacturing process.

So the journey starts with the manufacturer's data. Usually, we look at process data, raw material data, quality data and sometimes other data sources, for example, the ambient temperature. Then we build a unified multi-source schema, so we actually take all the data sources and put them inside one database while they're synced and connected to each other, and then we do something that no one else does: we build a digital model of the process, technically taking the P&IDs and converting them into a digital appearance, and then using our proprietary tools, we automatically convert it to code and inject it into the core process algorithms.

Now, the core process algorithms has in them the data and the plant topology, the flow of material, and the algorithms can do incredible calculations, for example, calculate dynamic traceability, understand buffers or production loops or even parallel processes. We can suddenly clean the data in a meaningful way. We understand the batch and changeovers and many, many other things, and the result is being digested through a straightforward, ready-to-use interface for the production team.

So if I'm going back to your core question today, algorithms with no process expertise will probably provide poor results. Embedding process expertise is a difficult task and we took upon ourselves this task to solve it and make it scalable.

- Thank you, Liran, for those meaningful insights, and thank you everyone for watching us and for those incredible questions you are sending. See you again, next time.

To learn more about Process-Based Artificial Intelligence and how it can be used to reduce production losses, check out the Seebo website.

Chemical Process Control - Optimizing Manufacturing with Industrial IoT

Watch: What "Dramatic Impact" did Seebo Make for this Chemical Resins Manufacturer?

Watch the full interview here:

Chemical manufacturing can be a highly-complex business. It's a constant battle to fend-off production losses, particularly in the areas of quality and yield.

The nature of those losses differ from manufacturer to manufacturer -- and even between different factories and lines within the same company. But very often, these losses have one thing in common: they are caused by inefficiencies in the production process itself (as opposed to, say, problems with individual physical assets on the line).

Allnex, a leading resins manufacturer, were struggling with a particularly painful problem. Their production process was very harsh on their equipment, leading to significant periods of factory downtime, as teams needed to be mobilized to clean and restart the machinery. This posed a major challenge in a competitive marketplace, by hampering their ability to meet growing demand.

As Global Operational Excellent Manager Jim Martin explains: "The process is inefficient, and while we're running the equipment is constantly being attacked by the process. The product is growing in the marketplace because of its eco-friendliness, and we wanted to be able to respond to the market... and produce more, more quickly."

Removing AI skepticism, with clear-cut results

To root-out this problem once and for all, Allnex evaluated a number of Industrial Artificial Intelligence vendors, each of which promised to eliminate the causes of their losses -- and every one of which ultimately failed to deliver. It's a story many chemical manufacturers will find all-too familiar, and it led to a sense of skepticism about the practical usefulness of AI technology.

But when they turned to Seebo, that skepticism quickly dissipated. Jim and his team were particularly impressed by Seebo's "process-oriented" approach, in stark contrast to other AI and Machine Learning vendors.

"We had a number of companies that we looked at. Seebo was the only one that was process-oriented. Many of the others were data-oriented, and they did not understand the process," explained Jim.

"Significant financial success"

Jim describes the process as an "iterative" one, that over time has had a "dramatic impact".

"Using Seebo has put us on the path of significant financial success. It's like having a very skilled, very talented engineer watching the process 24/7."

"We no longer have to spend time monitoring and figuring out the interrelations (between tags) -- the Seebo solution is doing that for us," he continued. "It's made the engineers' life much simpler.

"It's eliminating process inefficiencies, it's improved quality, it's improved yield -- and I cant say enough about the collaborative relationship, and their desire to get meaningful results."

The following is a full transcript of the video testimonial above:

There's a lot of confusion and skepticism on AI and what AI means...

Using Seebo has put us on a path of significant financial success. It's eliminating processing inefficiencies, it's improved quality, it's improved yields. It's had a dramatic business impact.

I'm Jim Martin with Allnex, we're a global, leading resin manufacturer.

The product we create at Allnex is very unique in the marketplace. It offers flexible adhesives, non-yellowing sealants, water-based polyurethane dispersions.

The process is inefficient and while we're running, the equipment is constantly being attacked by the process.

The product is growing in the marketplace because of its eco-friendliness, and we want to be able to respond to the market and be able to produce more and more with the equipment.

When we looked at choosing an artificial intelligence partner, we had a number of companies we looked at. Seebo was the only one that was process-oriented, many of them were data-oriented, and they did not understand the process.

We chose Seebo because they were very focused on the process, very focused on business results, and we were confident they could deliver those results.

One of the things the tool does very well is, it's like having a very skilled, very talented engineer watching the process 24/7, watching it the same way every time.

Having that confidence that the process is being monitored correctly, allows us to look at, where else can we use the tool? Where else can we spend the time? We no longer have to spend time monitoring or figuring out  the interrelations -- the Seebo solution is doing that for us.

It's made the daily engineer's life much simpler. There were many interactions with the process, the engineer had to take once a day, once a week, and constantly making interventions. Those interventions have dropped down drastically.

It just gives us the ability to not dwell on the processing issues, but dwell on the processing opportunities.

It was refreshing to work with a vendor that had exactly the same interest that we had.

It's eliminating processing inefficiencies, it's improved quality, it's improved yields,  and I can't say enough about the collaborative relationship and their desire to get meaningful results.

Want to see how Seebo can drive quality and yield improvements for your business? Book a free personalized demo, and we'll show you!

Case Study: How a Baked Goods Manufacturer Reduced Quality & Waste Losses by up to 70%

Process-driven production losses -- particularly due to quality and waste -- are a challenge for any food manufacturer. Sometimes, the cause is fairly straightforward; but what do you do when your process experts can't find the root cause?

This is a common problem for manufacturers in all industries. But for one international manufacturer of breads, crackers and biscuits, shrugging off those losses wasn't an option.

One of their production lines in western Europe was suffering persistent losses in quality and waste, specifically:

  • Rejects due to net weight underweight
  • Burned/over-toasted bread
  • Size and shape variability issues

Their process experts were using an advanced self-serve analytics platform -- but still were unable to locate the root cause of these losses. To their puzzlement, in most cases where losses occurred, no clear process  inefficiencies were apparent, and the data tags they investigated appeared to be operating within the permitted ranges.

Predictive Quality & Yield to the Rescue

That changed shortly after they implemented the Seebo Predictive Quality and Yield Solution. Using the Seebo solution, the company’s process experts were able to identify the hidden causes of their production losses, and gain clear recommendations as to how to prevent those process inefficiencies. Furthermore, production teams received proactive alerts as soon as those inefficiencies were detected - enabling them to prevent losses before they occurred.

Related article: Predictive Quality - Why is it Critical for the Food Industry?

This breakthrough eventually translated into a major reduction of waste and quality issues -- up to 70% for some! Ultimately, the company was able to save hundreds of thousands of euros each year on that production line, by preventing some of their most damaging production losses.

Want to learn how they did it? Read the case study below:

Predictive quality analytics in manufacturing

Predictive Quality: Why is it Critical for the Food Production Industry?

The Case for Predictive Quality

Food quality has always been a critical factor in the food production process, requiring food manufacturers to abide by stringent quality regulations, inspections and statistical quality control methods.

The impact of poor food quality is severe, with direct bottom-line consequences. Poor product quality reduces production yield, can damage the company’s brand and reputation. More immediately, quality issues are one of the most common causes of production losses for food manufacturers, together with other, largely process-driven losses such as waste. These losses translate into a significant dent in yearly revenues.

Conversely, meeting high-quality standards can reduce manufacturing costs – internally and externally. Internal costs emerge from problems associated with the product before it is delivered, e.g. shortages, waste, and delays. External costs arise post delivery - through recalls, lawsuits and warranty costs - and constitute a major expense for the food industry, resulting in a staggering $7 billion loss annually.

Food manufacturing operations turn to processes and tools to sustain high-quality production in addressing regulatory compliance, risk prevention, and product traceability. While tools such as statistical quality control (SPC) have been used since the early 1920s to identify issues as early in the manufacturing process as possible, they deal with problems that have already occurred.

Predictive Quality is an emerging category of Industrial Artificial Intelligence solutions, that provide manufacturers with the means to significantly reduce process-driven losses in quality and waste, by pinpointing the root cause quickly and with a high degree of confidence, and preventing those losses before they next occur.

Predictive quality management in food industry

Defining the "Quality" in "Predictive Quality"

In the food industry in particular, "quality" can mean a number of things.

For example, there are laws governing food quality in many countries, as well as international regulations which are important regarding globalization and the increasingly complex food supply chains. These laws and regulations assure food safety and a minimum level of quality for the health and overall benefit of consumers.

While certainly a very important topic, this is not the quality challenge that Predictive Quality addresses. Rather, Predictive Quality tackles production losses caused by inefficiencies in the process itself (rather than due to individual assets for example).

In fact, Predictive Quality isn't limited to quality-related losses, but also addresses other common process-based production losses. In the case of the food industry that typically includes waste and yield. In other industries, Predictive Quality addresses anything from throughput, to emissions levels to energy consumption. The common thread is that these are all losses which - like the consistent quality and waste losses experienced in food manufacturing - are caused by process inefficiencies.

Industry 4.0 and Predictive Quality to the rescue

Fortunately, food factories can overcome many of these quality control hurdles with the use of Industry 4.0 technology, specifically predictive quality.

Predictive quality technology provides manufacturers with 3 key capabilities to optimize their processes, thereby minimizing losses due to quality issues, as well as other process-driven losses e.g. waste, yield, and so on.

  • Automated root cause analysis
    Predictive Quality provides the process engineer with the tools to reveal previously unknown root causes of quality issues in the production line, via automated root cause analysis.Most food production lines have certain recurring losses that can't seem to be traced to any specific visible cause. This is particularly true for more complex food production processes. That's because with so much complex data to analyze, process engineers and experts are largely left to follow their intuition in selecting which of the hundreds, even thousands, of data tags to investigate. Naturally there will always be a blind spot - and the more complex the inefficiency, the less likely even the most eagle-eyed process expert is to find it.By contrast, a predictive quality solution uses Artificial Intelligence to do what human beings can't: continuously analyze all the relevant data (ERP, MES, Quality systems, data lakes, etc) at scale - including complex interrelationships between the different data tags. This provides process experts with the ability to pinpoint the precise combination of factors that are causing a particular quality problem, no matter how complex.
  • Predictive recommendations 
    A key component of Predictive Quality is predictive recommendations. Predictive recommendations use the insights from root cause analysis to identify the optimal process setting, enabling manufacturing teams to minimize quality issues as much as possible.

    Through continuous, multivariate analysis of production data, predictive recommendations will provide the precise optimal range of values for any given combination of tags.

  • Proactive alerts for real-time action
    Of course, all of this intelligence must be translated into timely action to provide real value. Proactive alerts are delivered directly to production teams in real-time, as soon as a process inefficiency emerges - i.e. a problematic combination of tag behaviors, as identified by the predictive recommendations. This enables front-line manufacturing staff to act to prevent losses before they occur. Ideally, the alert should include as much actionable information - for example, not only identifying the problem, but also providing clear Standard Operating Procedures for addressing them.

Predictive quality analytics in manufacturing

Predictive Quality empowers quality teams to anticipate and proactively address quality problems before they arise - making sense of complex data patterns to determine areas of greatest quality risk and assign production floor resources before risk becomes reality.

The business gains of Predictive Quality are clear and compelling, providing food manufacturers with a competitive edge in an era where every last drop of efficiency counts.

Check out this case study for a detailed look at how one particular food manufacturer succeeded in drastically reducing quality and waste losses, using Predictive Quality:

Watch: How to Select the Right Kind of AI
to Solve Your Manufacturing Losses

Industrial Artificial Intelligence can hold the key to solving some of the most complex and painful manufacturing challenges. But every manufacturing process is unique – as are the challenges that come with them. And different types of manufacturing challenges require different types of AI to tackle them.

With all the noise in the market around “Industrial Artificial Intelligence”, “Industry 4.0”, “Smart Factories”, “Machine Learning” etc, it can be hard to figure out which AI technology is relevant to tackle your specific problems. In this mini-webinar — the first in our new Vlog series — we reveal a simple but effective methodology to help you make sense of it all.

WATCH: Talking Smart Manufacturing — The Predictive Quality & Yield Series: Vlog #1


[Eden] Hi everyone, my name is Eden. Welcome to the Seebo vBlog.

If you are in manufacturing, you’ve probably heard all the buzzwords: Industrial Artificial Intelligence, Smart Factories, Industry 4.0, Machine Learning. Today I would like to talk about a question we frequently hear
from executives in manufacturing: How do you filter out the noise to get what you need? How can you find an Artificial Intelligence technology that actually addresses the business problem?

Today with me is Liran Akavia. As many of you probably know, Liran is the co-founder and the COO of Seebo. Liran is an expert on the field of industrial Artificial Intelligence and reducing manufacturing losses, and has extensive experience working with the world’s leading manufacturers.

Hi, Liran.

[Liran] Hello, Eden.

[Eden] So, Liran, how can manufacturers match their own unique production losses with the right artificial intelligence technology? Is there really a silver bullet out there or do they need to hunt for a specific solution for each individual problem?

[Liran] Manufacturers have diverse types of losses, and there is no silver bullet. There is no one artificial intelligence technology that you could deploy in your factory and will simply solve all the problems.

The good news is that there is a light methodology that every executive can deploy for their own teams to identify the relevant losses, and what is the relevant technology that works with that.

[Eden] Can you elaborate more about that methodology?

[Liran] Yeah, so… Just before we jump into the actual methodology I would like to share with you and with the audience how computers are thinking. Obviously we are not going to do a data-science lecture today, or not even a computer science lecture, but we are going to look in a very simplified way at the ways computers are looking at manufacturing problems, and once we understand that every manufacturing executive will be able to connect between the losses to the right technology.

So let me start with a quick couple of slides.

So let’s first take a look at how humans are solving manufacturing problems. So when we look at manufacturing problems we would differentiate between frequent problems and infrequent problems. When we have a frequent problem we have many examples of the problem and as humans we will try to look for a pattern, we ask what is common between all the problems and we will try to look for the common denominator in all those problems. In other cases, when we have infrequent problems, unfortunately we don’t have enough evidence for how the problem is looking, so we would ask completely different questions. Is it a normal behavior or an abnormal behavior?

Once we understand how humans solve problems, it is very easy to understand how computers are looking at problems.In the computer, or in the professional language,the frequent problems “world” will be supervised learning, and the infrequent problems “world” will be the unsupervised learning.

Obviously this is an over-simplification of the ways computers work. But generally, for computers, when there is a frequent problem, a computer would look at many examples of the problem and will try to look for a pattern, while when there are infrequent problems, the computer will ask a completely different question: Are we in a normal or abnormal situation?

[Eden] Okay, Liran, this is really helpful and very interesting, but how does this answer my question? How do manufacturers still find the right solution for their problem?

[Liran] You are right. Actually, let’s take a look at the methodology: So, the first step will be to list your production  goals, all the losses you are looking to solve. It could be improving quality, or improving yield, or improving concentration, reducing downtime, increasing the asset reliability, process stability, energy consumption. Make sure that you have a very clear list and prioritize it.

With that list, take each one of the problems that you listed and put it somewhere around these quadrants.

So this magic quadrant goes on what drives the problem — process-driven or asset-driven — and what is the frequency of the problem: is it a high-frequency or a low-frequency problem?

Let’s take two classic examples. If I’m looking at bread baking, and on the weight of the bread. This is related to the process and usually high-frequency because the weight changes for every bread (loaf)/ The other extreme is a failing pump. It will probably fail once or twice a year, and the reason is usually the asset, is the pump itself. So the bread weight will be usually a process-driven, high-frequency problem, while the failure of  the pump will be an asset-related, low-frequency problem.

So the first step was making a list of your problems. The second step, putting them on the quadrants, and the cool thing comes now. Each one of those quadrant areas is a different technology. Let’s start with quadrant one. In quadrant one we need artificial intelligence that understands the manufacturing process — after all, this is a process-driven problem — and we need supervised machine learning. If you remember what I said two  minutes ago, the supervised machine learning is when we look at frequent problems and find the pattern, and explain why did it happen. On the other extreme, when we look at quadrant four, that was the pump example, it is an asset-driven, low-frequency problem. Here we usually use vibration or acoustic unsupervised learning, since we don’t have many examples, we simply ask if it is normal or abnormal and we do it per asset, not per  process.

Those are usually the most important quadrants. Quadrant number one will be world of predictive quality and yield, quadrant number four will be the world of predictive maintenance.

Let’s take a look at quadrant number two. This quadrant does not have its own technology, but you can convert the problem to quadrant one or quadrant four. So, if you find a process proxy — I will explain what that is in a minute — go to quadrant one. If you don’t have a process proxy, go to quadrant four. Let’s take an example. Let’s take a vacuum system, this is an asset problem, that is failing every second week. It’s a frequent problem. So that would be quadrant two. If we can, for example, look for a process proxy that would be the pressure of the pump, and we know that the pressure of the pump is telling the story of the failing of the vacuum system, that would be a classic quadrant one example. But if we cannot find a process proxy like the pressure, we need to solve it with vibration or acoustic measurements.

As for quadrant (three), process-driven losses that are low-frequency, there is a technology behind that, but usually it is not a real concern for most manufacturers. An extreme situation in the process, very rarely, is not something that many manufacturers are actually concerned about. It’s not happening always and the technology of process-driven unsupervised learning is also not very accurate.

To make a long story short, generally your problems are probably in quadrant one or quadrant four. For quadrant one you will need predictive quality and yield, process-based technology which is supervised learning, and for quadrant four you will need predictive maintenance, vibration or acoustic unsupervised machine learning  per specific asset.

Another thing that is super-important to know at this stage: no one can do it all. There is not one supplier that can tell you “I’m the best in the world in supervised learning and I’m the best in the world in unsupervised learning”. Messi is a great football player but he’s not the best in basketball. You can simply not… You cannot be the best in all.

[Eden] So essentially we would put the Seebo solution in quadrant number one?

[Liran] Yes, you are right. In Seebo, our entire world and expertise is in quadrant number one. Our algorithms are designed to solve problems that are process-driven and frequent. And because of the pure focus that we  have and due to the fact that this is the only thing we do we are the best in the world at that.

[Eden] Thank you, Liran, and thank you everyone for all the great questions you’ve been e-mailing to us. Keep sending them in, the e-mail is right there.

See you next time!

Here’s how Seebo solves process-driven manufacturing losses like quality, yield, waste and throughput.

New Funding & World Economic Forum Recognition
Bring Plenty of Press Coverage for Seebo

It’s been a busy and exciting few weeks at Seebo, capped by two very exciting announcements that have garnered international coverage.

Here’s a summary of the key highlights.

Seebo’s latest funding round

Most recently, Seebo announced our latest funding round, led by Ofek Ventures, together with Vertex Ventures, Viola Ventures and TPY Capital.

That announcement generated significant interest, at a time when manufacturers are working harder than ever to optimize their production processes — particularly in the food industry (see our recent paper “Moving Manufacturing From Crisis to Competitive Edge” for more on that topic.)

Or, to quote a recent interview around that announcement with Seebo CEO and Co-Founder Lior Akavia in FoodNavigator:

Companies are not only looking to cut losses, increase efficiency, and improve their financials… they are also under pressure to maintain supply in he face of unprecedented spikes in demand. “For quite a few of our manufacturers, demand for their food products increased in the dozens of percentages after COVID-19 started,”…

Manufacturers were left with two choices: to either build new production lines to meet demand — but “this takes years” and “huge investments”, said Akavia — or optimise their processes, which can “quickly translate into increased capacity.”

Other publications which covered the funding announcement included industry and tech sites like VentureBeat,,, and, as well as several Israeli business and tech sites, including Globes, Calcalist and

Seebo was also featured a “

Seebo Selected by the World Economic Forum

Another recent, exciting announcement was Seebo’s selection as a World Economic Forum “Technology Pioneer”, for its contribution to the fields of Advanced Manufacturing and Industrial Artificial Intelligence.

As noted in our press release:

This year’s cohort selection marks the 20th anniversary of the Tech Pioneers community.

Throughout its 20-year run, many Technology Pioneers have continuously contributed to advancement in their industries while some have even gone on to become household names. Past recipients include Airbnb, Google, Kickstarter, Mozilla, Palantir Technologies, Spotify, TransferWise, Twitter and Wikimedia.

News of Seebo’s selection, among a handful of other Israeli tech startups, was featured extensively in the Israeli press, spanning national, business and technology publications — including the Jerusalem Post, The Times of Israel, Calcalist,, Israel21C and more.

With the ongoing COVID-19 crisis, the manufacturing industry stands at a pivotal crossroads. Download our free guide to learn how manufacturers can move from crisis management, to building a long-term competitive edge: