The predictive impact of AI
Infinera’s first supply-chain AI pilot project will go live in the middle of this year, Tuomala says, starting with one manufacturing plant. "We also want to provide availability information to our sales team and customers for all products before the end of the year."
The use of machine learning will speed up the company's ability to make scheduling decisions, he says. In addition, it will allow the company to consider many more factors than it is currently able to do.
Infinera is using supply chain management technology from Intrigo Systems, in combination with AI technology from Splice Machine.
Companies have been getting usable predictions from their supply chain management systems for 30 years, says Monte Zweben, CEO and co-founder at Splice Machine. But it's only recently that the data infrastructure is there to accurately predict things like delivery times, he adds.
"If you're a large manufacturer of network equipment, and you have sales people out trying to sell these large systems, invariably what happens is the sales people are asked, 'Can you get the order to me by this date?' And in most companies, even today with the best ERP system out there, sales people are relegated to saying, 'I will go check and get back to you.' That's an opportunity for the customer to go somewhere else and get a competing quote on the order — after all, they have to wait, anyway," Zweben says.
By having the information available in real time, the sales team would be in a position to negotiate with the customer. Maybe one of the line items isn't available by the required date, but the others are. "And that's a whole different process," he says.
But predicting deliveries is much more than just being able to pull up the manufacturing and shipping schedules. With smart supply chain management technology, companies can look at historical shipping times and manufacturing details, and combine that with external data feeds like weather reports.
"You can start promising against the predicted inventory levels, not the planned inventory levels," Zweben said. "And promising customers based on what's likely to happen, as opposed to what's supposed to happen. Now you're seeing around corners."