The Supply Chain Is One Big Graph In Start-up Elementum’s Platform

By on

rsz_elementum_transport_appby Jennifer Zaino

Startup Elementum wants to take supply chains into the 21st century. Incubated at Flextronics, the second largest contract manufacturer in the world, and launching today with $44 million in Series B funding from that company and Lightspeed Ventures, its approach is to get supply chain participants – the OEMs that generate product ideas and designs, the contract manufacturers who build to those specs, the component makers who supply the ingredients to make the product, the various logistics hubs to move finished product to market, and the retail customer – to drop the one-off relational database integrations and instead see the supply chain fundamentally as a complex graph or web of connections.

“It’s no different thematically from how Facebook thinks of its social network or how LinkedIn thinks of what it calls the economic graph,” says Tyler Ziemann, head of growth at Elementum. Built on Amazon Web Services, Elementum’s “mobile-first” apps for real-time visibility, shipment tracking and carrier management, risk monitoring and mitigation, and order collaboration have a back-end built to consume and make sense of both structured and unstructured data on-the-fly, based on a real-time Java, MongoDB NoSQL document database to scale in a simple and less expensive way across a global supply chain that fundamentally involves many trillions of records, and flexible schema graph database to store and map the nodes and edges of the supply chain graph.

“Relational database systems can’t scale to support the types of data volumes we need and the flexibility that is required for modeling the supply chain as a graph,” Ziemann says.

From a database standpoint the goal is to build a single supply chain platform that leverages multiple data stores. A URI is used for identifying data resources and metadata, and a federated platform query language makes it possible to access multiple types of data using that URI, regardless of what type of database it is stored in. “That is the unique thing, the source of innovation,” he says, “from the standpoint of the nuts and bolts of what we are building.”

Elementum starts things off with an enormous amount of data from its relationship with Flextronics, and also can connect via Dell Boomi’s integration cloud to other data endpoints in a customer’s IT stack, such as ERP technologies from Oracle or SAP. Elementum says it can take the pipe of data from customers’ ERP or shop floor systems directly in without having to spend months on data mapping and configuration. With many supply chain participants still communicating by email spreadsheet attachments, it also has developed a way to ingest and insert that data into its system. “This ties things back to the graph — just attaching to the systems in the supply chain once and then being able to ingest that data for all the customers that leverage that node in the supply chain,” says Ziemann.

The front-end is comprised of HTML5, CSS3, JavaScript, and GIS visualization technologies for helping to track shipments. Integration is all via JSON and REST APIs, including data access by its native iOS and Android mobile apps. Indeed, while it’s a busy backend, simplicity is the byword for the front-end experience for customers. “We want to make the supply chain simple,” says Ziemann. “We present ultimately to the end user or customer in the form of a mobile app that is tailor-made to the specific use case for the business user.” With the Transport app, for example, users get logistics visibility that shows all their company’s inventory around the world that’s in movement so they can see quickly where shipments are and if they are late, drilling down by carrier or customer, to address problems as they arise.

Elementum_CEO_Nader_MikhailThe value there can be significant. “We give you blind spot awareness that, for example, one key part is missing to build your widget, so you can look for an alternate supplier,” says CEO Nader Mikhail. He points to a fire in a factory in China last year that took out 25 percent of the DRAM capacity in the world – and most OEMs weren’t aware for days that the event had taken place. Elementum picked it up for its early adopter customers within minutes. “They were notified and were able to lock down their DRAM supply in the channel before it hit the 24-hour mark.” After that point, other DRAM suppliers knew about the event and prices spiked amid the shortfall in supply. “This is tens of millions of dollars on the line,” he says.

Change Is In The Air

Supply chain parties have been late to moving beyond their accustomed technologies for accomplishing business to gain such realtime advantages. Elementum was born at Flextronics with the hiring of Mikhail to lead global innovative solutions by Flextronics CEO Mike McNamara, to unlock further value in its own enterprise. As Ziemann explains it, Mikhail quickly identified that Flextronics itself has tons of ERP systems running in the company and one of if not the largest supply chains in the world in terms of systems and interconnectedness.

“And at the end of the day, after spending $100 million in IT every year, what executives get is a spreadsheet that tells them what is going on,” he says. “That’s a phenomenal fact. Why spend all this money on these systems if you’re still getting your data via something that existed for 30 years. And that is endemic to the industry, not unique to Flextronics.”

Mikhail also points out that mobile apps had received only half-hearted attention at best from the sector, given the extensive queries one might have to write to understand what’s going on in a certain factory given backend systems based on relational database technologies and multiple one-to-one integrations across the supply chain. “A lot have shied away from building mobile applications because they can’t be responsive enough,” he says. “But you can solve the problem using graph databases,” he says, in seconds, “because you can traverse the edge of the graph and find different connections vs. having multiple layers of queries.” Structure data in a way to let it be crawled easily, Ziemann adds, and you can do all sorts of interesting things like this.

Adapting To The Next Wave

There’s a learning curve, Milhail notes, in the first few months of an engagement, where people appreciate the mobile apps and realtime data but still like to download a spreadsheet at day’s end. So, Elementum supports that, but in fairly short order people do stop downloading spreadsheets, he says. Elementum also had to confront the fact that with a NoSQL database and graph database, with a single customer’s data potentially stored in multiple locations in the cloud and replicated across systems, it is possible that realtime query results sometimes can be off by a hair’s breadth – and it had to explain that 100 percent fidelity in the supply chain isn’t required. In fact, Ziemann says, they discovered that business users weren’t concerned about that, though IT users were.

In the old way of doing things, the business user was getting information that had gone through multiple layers and been reconciled, so the accuracy was there. But by then that accurate data was 3 to 4 weeks old, Mikhail explains. “Then you go to the business user and say you get real-time information as of the last ten minutes, with a margin of error because the data comes from multiple systems so there’s a slight gap in synchronization – they get it,” he says. “They valued that much, much, much more because the data is so fresh.”

Another challenge is getting users used to the idea that “you don’t need 50,000 features in your supply chain system to run your supply chain effectively,” Mikhail says. “In every part of your supply chain, whether factory or transportation or inventory management, truly an 80/20 approach works. There are 10 to 15 things you really need to manage,” which is why Elementum app screens are limited and focused on specific problems, he notes.

Customers currently include Flextronics, Dyson and Enphase Energy. Elementum is using a node-based pricing model, so as manufacturing, supplier or distribution sites grow or shrink the cost of the system stays proportionate to that.

Leave a Reply