Watch Now


Flexport to open platform for third-party app developers

(Photo credit: Flexport). A Flexport product engineer at his workstation.

James Chen, the chief technology officer (CTO) at Flexport, has a grand vision for global trade technology – Amazon-like ETA prediction and an open ecosystem of new tools built in a layer on top of Flexport’s data.

Flexport, the digital freight forwarder founded in San Francisco in 2013, has double-downed on its technology investments over the past year and says that the COVID-19 pandemic has deepened the company’s commitment to its long-term tech vision.

Chen spoke with FreightWaves about what Flexport’s engineering teams have been working on during the past year. He also shared new details about Flexport’s grand vision for technology’s role in global trade – a Shopify-like ecosystem for application developers housing “not just billions, but trillions dollars of value.”

Before joining Flexport as CTO 11 months ago, Chen was director of technology at Amazon Shipping and had the same role for Amazon Global Logistics. Prior to that, at Rakuten, Chen supervised everything from technology M&A to cloud computing and mobile strategy. 


In less than a year, Chen has grown Flexport’s technology team from about 190 employees to just over 400, building what he calls an “enterprise software development process” capable of shipping bug fixes, usability upgrades and new features dozens of times a day. 

“The vision is to build the technology platform to make global trade easier,” Chen said. “We will open up more and more capabilities for anybody in the world to use. When we get to the end, in all 163 countries, someone is using our tech to move freight and we’re the brand everyone will go to get help.”

Today, Flexport serves almost 10,000 suppliers and customers in 116 countries. The Journal of Commerce estimated Flexport’s 2019 gross revenues at $860 million.

(Photo credit: Flexport. A meeting of ‘makers’ — product managers, designers, engineers, and data scientists, at Flexport’s Shenzhen office.)

Chen said that the COVID-19 pandemic has made it obvious how vital it was to invest in building a great customer experience on a web-based application. Many of Flexport’s customers’ supply chain managers are working from home, or have limited staff in their offices, and Flexport has heard from more than one of its customers that it’s the only freight forwarder that it can currently use.


Chen explained that while Flexport has its own operations in 14 countries, it works with a number of freight forwarder partners who have access to the tools in Flexport’s platform in order to drive a unified customer experience.

“What I believe we can offer, more than other forwarders, is our focus on customer experience,” Chen said. “We’re standardizing the way our partners work with us and providing more capabilities for them to use. Whether your freight is managed by Flexport or a partner, you’re going to get that experience. We’re driving standardization for carriers, freight forwarding partners, and customs brokers to provide a unified experience.”

Flexport’s commitment to a superior user experience includes offering to perform software integrations with its own resources – like connecting Flexport to SAP or Oracle – which allows companies to de-risk integration projects and cut out expensive professional services consultants.

“We take care of interconnectivity on behalf of our customers and carriers without adding costs to them,” Chen explained. “It’s about lowering barriers to entry in order to have very efficient digital communications between systems.”

The policy responses to the COVID-19 pandemic have profoundly disrupted global supply chains, causing historical patterns to break down and emphasizing the importance of real-time data. On many lanes, air freight capacity is a fraction of what it was previously, and an extraordinary number of blanked sailings has led to the acute imbalance in container networks “reaching a tipping point,” in the words of the International Federation of Freight Forwarder Associations.

Those market conditions make Flexport’s work connecting global trade partners, cleaning data, and providing real-time multi-party views into visibility, documentation, capacity, transit times, and rates more important than ever.

But a great deal of work has to be done to make global trade data standardized and accessible. Visibility data is, of course, fundamental, Chen said, all the way from purchase order and cargo ready dates to the transfer to a warehouse, where freight is consolidated and prepared for the next leg in its journey.

“That visibility information definitely needs to get shared across the board,” Chen said.


Across the freight tech industry, visibility solution providers from project 44 to FourKites and Blume Global have collected massive amounts of data and then started building predictive analytics capabilities as an additional layer on top of the visibility data. The basic idea is that by tracking thousands of freight movements on a given lane, a company can become pretty good at predicting transit times and estimated times of arrival, based on market conditions, lead time, carrier selection, etc.

Basic ETA prediction is widespread in logistics, Chen said, but the estimates aren’t precise and service providers offering those predictions typically only commit to the highest time value out of a fairly wide range for fear of over-promising to their customers. That limits how useful the predictions can be – the narrower the range of ETAs provided with a high degree of confidence, the tighter transportation schedules can be run, eliminating wasted time and freeing up capacity.

“On a shipment from Shanghai to Los Angeles, most freight forwarders will quote a pretty wide range to manage expectations,” Chen explained. “We want to use technology to shrink that range and still provide a high level of confidence.”

Chen worked on similar initiatives – predicting tighter ETA ranges with progressively higher degrees of confidence – at Amazon Global Logistics, but the crucial difference at Amazon was that in many cases, the e-commerce retailer served as its own shipper. At a fundamental level, Amazon could control the schedule of its aircraft, the location of its trailer assets, and gain first-party visibility into the movement of delivery vans. 

That isn’t the case at Flexport, which relies on carriers in a variety of modes – ocean, air, drayage, intermodal and truckload – to physically transport freight. Multiple sources of data in multiple formats, from EDI to API, PDF, unstructured text, and physical documents, make the fundamental work of establishing an accurate version of reality quite difficult. Before Flexport can predict when a container shipment is going to arrive at a port, for example, it needs to know exactly when it was picked up by the drayman. 

“But what’s the right data to count on?” Chen asked. “The first data point is potentially from the consignee – they tell us the cargo ready date and  pick up at 1 o’clock, but then the supplier changes it to 2:00 p.m. But then the truck driver’s system says they picked it up at 1:30, and then you have their GPS device that shows another time, so which data source do you trust? For which event is which data set more accurate?”

(Photo credit: Flexport. James Chen, chief technology officer at Flexport.)

That’s why Chen’s team is currently focused on building internal data quality indicators for all of the events in Flexport’s platform. When multiple data sources about a single event are collected and are being integrated, if they conflict, the data quality indicators will score each data point and make predictions about which points are likely to be most accurate and most correlated to a specific outcome. The goal is to make confident predictions that are precise enough to drive efficiencies in Flexport’s customers’ operational planning.

“With better predictions, operational planning can be better and more agile,” Chen said. “For example, for an air shipment arriving at a hub in Chicago, we may know that if it’s on a certain schedule, it will definitely be available for the driver to pick up in a two-and-a-half hour window. If we have high confidence in the data we can provide that.”

The ultimate goal doesn’t stop at highly accurate predictions. By creating a vast data lake of scored information on its platform, Flexport is building the foundation for an ecosystem of third-party application developers who can build their own businesses and tools in a layer on top of Flexport. 

Chen said that supply chain consultants could build tools that analyze certain opportunities for their clients based on commodity flows in and out of markets, companies could build custom analytics providing real-time views into the efficiency of their facilities, and it would even be possible to drive new sustainability efforts by mapping imbalances in freight networks. A backhaul destination market where pallets tend to accumulate might be a candidate for an initiative around biodegradable shipping materials, Chen explained.

Flexport is currently working with partners on the vision for the app platform and making sure that the best technology leaders in the industry are part of the platform’s go-to-market. The public release is likely more than a year away. First, Chen said, Flexport is opening up more tools and data to its freight forwarder partners, working in other countries that currently have access to less than 5% of Flexport’s functionality, greatly expanding the amount of data the platform houses.

“Our measure of success, as we continue to open it up more and more, is to help other businesses run on top of our technology platform, so we have not just billions but trillions of dollars of value built on the platform.”

2 Comments

    1. Andersen

      They are not. However, neither was Amazon for the past 10 years. Flexport has top investors so they probably won’t have issues raising money until their model is profitable.

Comments are closed.

John Paul Hampstead

John Paul conducts research on multimodal freight markets and holds a Ph.D. in English literature from the University of Michigan. Prior to building a research team at FreightWaves, JP spent two years on the editorial side covering trucking markets, freight brokerage, and M&A.