Trust, localization and the ability to adapt, be agile and lean seem to be the secret ingredients for manufacturing during this turbulent COVID-19 period.

The COVID-19 business impact is large for manufacturing, with fewer people allowed in factories and reduced mobility.

For customers, trust in products is more essential than ever.  During this COVID-19 crisis, are those manufacturers who are still open placing more emphasis on maximizing production, even if this may risk the actual safety of the products they are circulating in the market? How are we sure that, with the current rapidness of production, safety regulations can be adhered to? According to the 2020 Edelmann trust barometer, respondents recognized the need for specific brand action to help address the societal challenges posed by Covid-19, such as protecting the well-being of employees to shifting products.  Tracing the safety of products and the health-safety of human workers along the entire supply chain becomes a key priority for manufacturers. Blockchain technology can support such a traceability need easily (Matthieu Hug, CEO and Founder of Tilkal)!

For manufacturing, COVID-19 has disrupted the supply chain with the need to re-sync where products are produced potentially requiring new business models (Eric Prevost, Oracle).

Furthermore, COVID-19 has increased the need for fast adaptation and manufacturers are looking for AI to increase the ROI (Rachel Alexander, Omina Technologies). But ROI can only be maximized by using a lean AI and human-centered AI methodology.

There are four challenges:

• Challenge 1: Trust in products is from the past
• Challenge 2: A bottle of milk is rarely just milk in a bottle: traceability means aggregating data throughout transport

• Challenge 3: the supply chain is complex and distributed
• Challenge 4: Unknown supply chain actors lead to uncontrolled and harmful behaviors. 

Nobody knows exactly who their tier 3 suppliers are. This leads to illicit trade such as child labor.

Having a sensor on a single product is not enough for traceability.

Blockchain technology can overcome all 4 challenges.

Blockchain is essentially a distributed database allowing for decentralized data collection. Every actor has a copy of the data. Every actor can contribute data by making a declaration to this distributed database. Data is immutable.  If the data that was declared had a mistake, then an actor can make another declaration to fix it.
If one participant changes the data, all other people in the blockchain are notified: tamper-proof data.
It does not really ensure whether the data is true, but it allows you to audit the data: Who said what at what time?

Tilkal created a software platform for traceability that is based on blockchain technology.

• transparency to the customer:

Supermarket Casino wanted traceability for poultry: eggs: from which farms do they come from, what is the situation of the chickens in this chain.
Farmers were afraid if their name and location would become public: afraid of theft and of violence by vegan/pro-animal organizations.

June: sells cosmetics and diapers for babies: where does the plastic come from?

• operational control:  improving inventory: Danone: traceability of formula milk on unit-level from China upto tier 3. View in real-time on inventory.
• compliance:  work with a supplier for Airbus: The supplier can apply for a specific tax regime. Typically the tax agency asks for prove of being entitled to the beneficial tax regime long after the application has been filed. The supplier had to be able to prove up to two years after application that parts had only be used for planes that solely used in civil air flights.

Industry 4.0 trends require manufacturers to rethink their digital strategy and platforms to transform their challenges into opportunities.

We are now at the fourth industrial revolution.

The 4 industrial revolutions:

• 1st industrial revolution: human work simplification:

1 person at 1 machine, the aim was to simplify the work of human but not to replace the human

• 2nd industrial revolution: electricity.

Standardize work, simplify work in a linear way: the conveyor belt.

The conveyer belt allowed multiple brains to work together.

• 3rd industrial revolution: automation and nonlinear work orchestration.

Automating some part of the process where it does not need a lot of decision making.
Manufacturing robots
Business ERP: automate processes for finance, procurement, supply chain
Put the brain of the human in a machine.

• 4th industrial revolution: continue to reinforce our production efficiency.

Local manufacturing as opposed to distributed manufacturing: split manufacturing in different parts. COVID-19 requires to resync where the product is produced.
Lancome: small manufacturing, buy skin product directly in the store.

There are three types of AI. AI that is custom-built, AI that is embedded and you do not know it is part of the product (e.g. recommendations in Netflix), and AI that is generic/industry-specific.

Typically there is no need to build your custom AI solution for a lot of the manufacturing problems.

1. Identify your use case. AI use cases typically match the 4 S of AI.

Some relevant use cases in manufacturing: automate your invoice processing, identify and categorize exceptions (find weak signals that humans would overlook), prioritize exception handling, identify best next action (what-if and recommended actions).

2. Look at what kind of AI solution you need: what are your AI intentions?

3. Check whether there is any existing AI tool solving your problem.

Corporate profit used to be the biggest goal for companies but as companies are transitioning to Industry 4.0 they are realizing that Return on Investment must come hand and hand with societal impact and ethical and environmental responsibility towards customers, employees, suppliers and communities.

The disruption on manufacturing caused by COVID-19 has severe operational, social and financial consequences.

It is forcing manufacturers to rethink risk management and contingency plans, workforce safety protocols, manufacturing operations and new ways of working opportunities, all at the same time.

So far, manufacturing leaders have concentrated on solving the immediate challenges required to keep the business as stable as possible. They’ve formed rapid response teams to gain a better understanding of their production demand changes, labor support challenges and supply chain ecosystem constraints.

It is more imperative than ever for manufacturing companies to 

• Understand the impact of demand disruptions
• Manage workforce safety and flexibility

• Make sure your ecosystem is viable
• Be faster, more accurate and flexible when it comes to your physical assets
• Further build out digital capabilities

• Downtime of machines due to a technical failure
• Late detection of production quality issues resulting in large scrap
• Sub-optimal planning and scheduling resulting in low OEE
• High energy costs
• Uncertainty on demand resulting in too high inventory levels and sub-optimal material use

Everything you can not plan for in advance is a big cost.

The margins in manufacturing can be challenging and although industry 4.0 has a lot of potential, companies are finding increased costs to add intelligent devices/equipment are impacting the ROI of manufacturing IoT applications.

Not all manufacturers have a standard ROI strategy in place, however, making an accurate calculation difficult to assemble. Numerous variables must be factored in with regard to the cost of the project and the total impact on operations. Because each project is unique, establishing a universal methodology for ROI is difficult for many companies. 

53% of manufacturing companies have started Robotic Process Automation projects, however the same study said only 15% of manufacturing companies are using other AI technologies in production, yet they know that to get the best ROI they will need to move beyond the robotic process automation and into AI.

What is holding companies back of implementing AI?

Fear for large investment and need to have an army of data scientists. They have seen a low ROI of PoCs. At Covid time they do not have time to invest in these kind of projects. The reason manufacturing companies are still slow to adopt AI is because they feel daunted by the investment (they don’t have the time or money to take on an army of AI Scientists) and when they do overcome these hurdles they feel like the whole process is taking a long time and not delivering the results they expected. Not only are they not seeing the ROI they expected, but they also don’t feel like they can trust the results and companies more and more want to go beyond ROI.

If you implement AI in the wrong way, then it might eat in your margins/ROI.

The methodology you use to implement AI is very important.

At Omina we developed a system that puts AI in the hands of the business so you don’t need an army of AI specialists. Our AI platform mimics the reasoning of an AI scientist and let’s you compare scenarios so you can do AI in a lean and agile way.  This means you can get production-ready algorithms in half the time which increases your ROI.

• Use Case 1:  Prediction of Machine Breakdown
• Use Case 2:  Prediction of Product Defects
• Use Case 3:  Machine Efficiency Optimization

• Choose AI Use Cases that fit with your strategy.
• Select a limited number of data sources that are of reasonable quality and show with simple AI solutions value of AI
• Have a parallel track to improve data.

Did you like the AI Convention Europe 2020 summary?

Why not read the summary of the Virtual DI Summit 2020 on AI, Ethics and Law?

To receive future event updates, sign up for the Events list in this form.