Artificial intelligence is playing a growing role in parcel delivery companies' interactions with shippers and consumers.
FedEx, UPS and smaller companies in the parcel space are tapping into the technology to generate more precise shipping time estimates, limit stolen packages, automate customer service processes and help customers make sense of their package delivery data. The end result is more reliable delivery services and critical information delivered to shippers faster, executives interviewed by Supply Chain Dive said.
"I see the future of AI at FedEx as a true change agent," said Tony Kreager, the company's enterprise VP of data, digital and commercial technology, in an email.
The technology is already a key driver of efficiency in FedEx's internal operations. The company leveraged AI algorithms to proactively divert volume ahead of severe winter weather at its Memphis, Tennessee hub earlier this year. Rival UPS has used it to match network capacity with volume fluctuations, helping it keep labor expenses in check.
But more visible benefits for shippers have emerged from AI technology. Parcel delivery companies can now synthesize and analyze the wide swaths of data tied to packages flowing through their networks, along with outside factors like weather and traffic, said Sri Sripada, managing director at West Monroe, a business and technology consultancy. This has accelerated the use of AI in their customer-facing applications.
"The majority of them are actually making [their] way all the way to the finish line, which really means that you can scale those AI innovations all the way through the various networks and ultimately to a large segment of the customers," Sripada said.
AI boosts FedEx, Veho's delivery precision
FedEx offers a two-hour estimated delivery time window for most shipments on the day it's slated to arrive via AI. By using its Global Delivery Prediction Platform, the company merges real-time shipment data with last-mile delivery information, which is then fed into "a sophisticated predictive model that leverages street level geography," Kreager said.
"We utilize AI to ensure we are giving our customers the most accurate targeted information for their FedEx shipments," Kreager said. "Customers want near real time visibility to their shipments and are refreshing fedex.com and the mobile app to know when exactly that important outfit or gift will arrive."
Alternative carriers are tapping into AI for more precise delivery time estimates as well. Veho, the fast-expanding carrier that uses gig drivers, leverages technology that accounts for a variety of factors beyond the standard driving time between stops, such as individual driver metrics and the size of a package.
"If someone's delivering three large boxes to an apartment, we know that would take longer than delivering a single poly bag to an apartment," said Fred Cook, Veho's co-founder and Chief Technology Officer, during a session at the Home Delivery World conference in June.
Veho is also trialing the use of generative AI to see how closely drivers followed customers' specific delivery requirements, Cook said. The technology compares the driver's picture proof of delivery with the given instructions and scores the delivery accordingly.
UPS limits porch piracy with AI
AI's presence can influence where a parcel carrier ends up dropping off a package.
For example, UPS' DeliveryDefense offering allows shippers and their customers to reroute shipments at risk of being stolen to alternate dropoff locations. Its software uses AI and machine learning technology to weigh factors like location, number of delivery attempts and frequency a package is lost to predict the likelihood of a successful delivery at a particular address.
DeliveryDefense will help shippers make safer and smarter delivery decisions prior to printing a shipping label, EVP and Chief Digital and Technology Officer Bala Subramanian said during a UPS Investor Day presentation in March.
"AI helps create simple actionable scores that identify 2% of addresses that drive more than 30% of shipping losses," Subramanian said. "We will be using unique data that only UPS can aggregate for our customers to help unlock value."
Internally, UPS leans on generative AI to automate responses to some of the roughly 52,000 customer emails it receives daily as well, the publication CIO reported in May. During pilot testing of the "Message Response Automation" project, UPS saw a 50% reduction in the time its agents spent resolving emailed inquiries, like requests to hold a package for pickup.
AI chatbot parses shipper data
Shippers are harnessing AI to improve their own decision-making around parcel delivery.
ShipScience, a software provider for parcel auditing, released its ParcelAI chatbot earlier this year to quickly respond to customer inquiries on shipping data, such as carrier performance comparisons and time-in-transit speeds. It can integrate with shippers' data from their FedEx and UPS accounts, in addition to other carriers. The more specific a query is, the better of a result it will provide, ShipScience founder and CEO Anthony Robinson said.
"The overall goal is to just give anybody who's a logistics professional for a company that has even slightly complex parcel data the ability to get to quick answers," Robinson said in an interview.
Generative AI's ability to interpret shipping data stands to benefit high-volume shippers moreso than smaller businesses, at least in the near term, given the tens of thousands of shipments they have to manage on a monthly basis, Robinson said.
"Their order volumes are changing, their customers are changing, the carriers are changing," he said. "There's so many things moving that it's helpful to have something in the middle that can kind of stitch it all together and help you make the right decisions in all these cases."
AI isn't a magic wand companies can easily wave to improve the shipper experience, however. They need to hire the right talent, develop the necessary infrastructure and figure out how to easily access customer shipping and order data in order to feed AI systems large quantities of clean and accurate information, West Monroe's Sripada said.
"I think there's a barrier there around access and availability of clean data, which allows you to integrate all those data elements and put those element models, or AI models, on top of that," Sripada said.
