How AI reduced cost by half
Accountants are supposed to count inventory, at least once a year. On January 1st, most people would rather be having a barbecue with their family if they live in Australia, or some hot soup if they live in New York.
But someone must go and count every single piece of inventory. Not so hard for items at the bottom, difficult for the ones on the top shelf. Walmart has solved that problem, and reduced inventory taking from 1 month to 1 day, by using drones.
One of our warehouses in China has 150,000 SKUs. At that level of complexity, managers typically use “average” to manage inventory, and at most look at the top 100 SKUs. From operations in other markets, I knew it was a bad idea to manage inventory on an “average” basis.
Instead, we used AI to analyze all transactional data over 2 years. Every single purchase, sale and inventory movement was studied. When we had ordered more units, even though we had plenty of them, we deployed the 5 Whys methodology to understand the root cause. We reduced forecasting errors and improved the stock replenishment process. As a result, we reduced inventory by 20%. Mckinsey estimates AI can reduce inventory by 50%, out-of-stocks by 65%, and forecasting errors by 50%.
This article is about one of four topics I presented last week, at the World Emerging Industries Summit during the 13th Annual Investment and Trade Fair in China. Detailed below are 3 real-life, implemented examples of How AI can reduce cost by half.
1.AI improves the “Where to manufacture” decision
A typical investment for a facility producing 30 million units per year is $100 Million. Payback is calculated over 10 years. To note, such facilities’ labor cost is “only” $11 million per year. This is because the large investments are on equipment and technology.
Large companies have specialized teams looking into these projects. At the end of the day, however, the decision is made by the Board. The members of the Board, regardless of whether they are sitting in New York or Frankfurt, have personal biases. Intelligence, the artificial one, will lessen the effect of bias such as:
- Personal bias: “I got to have a plant in my country.”
- Market growth bias: “We need 5 new plants. AI will grow at a CAGR of 37%.”
- Perceived quality bias: “We need a plant in the US. Mexican products are inferior.”
AI is better at it because instead of looking at a few pieces of evidence, it sorts through million pieces of data to populate the 400 KPIs used to decide where to locate a manufacturing plant. AI is also able to constantly adjust for foreign exchange and political risks updates.
2.The manufacturing cost of a product. Reduced by half by AI.
For many products, roughly 2/3 of the cost is Raw & Packaging materials. Procurement employs 32 people at a typical $10 Billion company. They also make large investments to gather cost data. Many times, however, smaller organizations, with a much lower purchasing power, buy at a lower rate. This is because while there is a tremendous amount of data, there are no resources to make intelligent use of it.
We have used AI to look at all these data and, together with global sourcing, reduce cost. In one case, we were able to reduce waste by over 30% by studying operational Best Practices from all our manufacturing facilities around the world. Germany gave us insight into China.
Fixed Cost is another critical element. When I meet with Boards in New York or Frankfurt, I remind them that the cost of the rent, of the room where we are sitting, is $2 million per year. If we were able to use that room for more than 10 or 12 hours, we would save cost. With a factory is the same. We want to utilize that $100 million investment 22 hours per day, 7 days a week. Having better data on demand forecasting and product launches improved capacity utilization by 50%.
Finally, productivity will improve with the automation of Support Functions (Finance). Altogether, the manufacturing cost of a product can be reduced by half by AI.
3. Machine Translation grows Exports by 17.5%
When I first arrived at St. Petersburg’s train station, in Russia, the alphabet was a challenge. Today, I can read any document in Russian by scanning a picture to Google translate. For free.
Alibaba translates 200 Billion words per day in 16 languages.
Consumers benefit the most. In the old days, one had to travel. Not anymore. Machine Translation enables customers in Europe to save 50% on purchases. Because Europe is a free trade zone, they can look at the cost of any product on any Amazon site and have it shipped to their home. Consumers are also looking at the cost of a premium Linkedin subscription. It’s $30 in the US, $30 in China, $32 in Europe and $13 in Brazil. So they are buying it through Brazil and saving 60% on it.
Ebay’s machine translation service has increased exports by 17.5%. Both small businesses and consumers are benefiting from it.
Conclusion
These are 3 real-life examples of how AI reduced manufacturing cost and led to a better decision-making process about where to locate a plant. During the convention, we also discussed:
-Will controllers lose their job to AI?
-The limits of AI, and the solution
-How the US & China will benefit from cooperation on AI
The entire presentation is available here:
https://www.slideshare.net/GeorgeBenaroya/ai-global-manufacturing-benefits