Trax: Digitizing retail stores with Google Cloud Platform to take shelves from image to insight

About Trax

Trax is a world leader in computer vision solutions for retail, ranking in the Top 25 Fastest Growing Companies on Deloitte's Technology Fast 500 list. The Trax platform enables tighter execution controls in-store and the ability to leverage competitive insights to unlock revenue opportunities at all points of sale.

Industries: Retail & Consumer Goods, Technology
Location: Israel

Trax deploys Google Cloud Platform in a multi-cloud service-based infrastructure for increased stability, scalability, security, and commercial flexibility.

Google Cloud Results

  • Provides a flexible environment that accommodates multiple cloud vendor and marketplace integrations
  • Scales at speed, allowing AI recognition models to quickly adapt and respond to clients' changing needs
  • Opens up new markets with a multi-cloud infrastructure that can ease retailers' concerns about commercial rivalries

Recognizes almost 1B products from images monthly

In competitive markets with tight margins, every detail counts for brick-and-mortar stores. Decisions that put the right products and promotions in the right place on shelves can mean the difference between survival and success. That's where real-time data becomes invaluable. Retailers and consumer brands are turning to data analysis and AI to draw out as much insight as possible and gain an edge. They can look through inventories and sales receipts to see what items are coming in and how well they're selling, but it's much harder to find data on what happens in between.

That's where Trax comes in, with an image-recognition platform built on powerful AI technology, to scan individual products and check shelf display, product prices, and promotions. Clients can get real-time insights into how their products are performing in their exact locations on shelves, so they can react accordingly. "Our mission is to digitize the store," says Yair Adato, CTO at Trax. "We can collect data from multiple devices and turn that into a digital representation of shelves in the physical store."

"Working with big retailers meant that we had to be able to recognize tens of thousands of products from images. We needed a new solution with massive compute power and the flexibility to integrate with our existing cloud infrastructure. Google Cloud Platform meets those needs for us."

Yair Adato, CTO, Trax

When Trax launched in 2010, its first clients were consumer brands that wanted to know more about how their products were being displayed and promoted in retail stores. The company grew quickly, acquiring world-famous clients such as Coca-Cola. By 2017, Trax started to work with large retailers as well, which was a great opportunity that came with unique challenges. First, these new clients had security and commercial concerns, which required Trax to diversify its existing cloud infrastructure. Second, the sheer scale of retailers' product ranges meant that Trax's platform had to cope with new technical challenges of scale and complexity. For Trax, the solution has been Google Cloud Platform (GCP).

"Working with big retailers meant that we had to be able to recognize tens of thousands of products from images," says Yair. "We needed a new solution with massive compute power and the flexibility to integrate with our existing cloud infrastructure. Google Cloud Platform meets those needs for us."

Google Cloud Platform for power, scale, and flexibility

For its products to work, Trax needs to make sure that a number of complex operations go smoothly. Ingesting, processing, and analyzing hundreds of thousands of images a day requires huge resources, and from the start, Trax knew that only a cloud-based infrastructure could cope with its demands. With data streaming in from multiple services, Trax adopted a microservice-based architecture to handle the complexity of the tasks.

As well as technical challenges, retail stores had specific concerns about the security of their data and where it was being stored. Retail clients made it clear to Trax that they wanted its services to be free of any links to the retail market, in order to safeguard their data from competitors. Trax saw an opportunity to both reassure its new clients and build resilience in its platform with a multi-cloud infrastructure. The company embarked on a three-month trial of the leading cloud vendors, and decided on GCP for its cost-effectiveness and ease of use. "Our engineers found they just clicked most easily with GCP," says Yair. "Their opinion carried a lot of weight in the final decision."

For its new multi-cloud infrastructure, Trax planned in advance what the architecture would look like and how the clouds would relate to each other. Specifically, it considered whether there would be a master-slave model, or whether the two clouds could operate independently of each other. Given the commercial needs of its clients, Trax opted for the latter, building an infrastructure that switches between clouds if needed, with no loss of service. As the mainstay of Trax's GCP implementation, Compute Engine instances allow the company to replicate existing tasks and services with ease. With the ability to spin up and shut down servers as needed, Trax can now respond instantly to market changes that reveal trending merchandising display techniques or even provide valuable upsell opportunities.

"We've been using as much of the Google Cloud ecosystem as we can. The products integrate very well, and the platform as a whole works well in a multi-cloud solution. And if we need to, we can run everything from Google Cloud Platform itself."

Yair Adato, CTO, Trax

With so many different sources of data, Trax deployed a range of managed solutions from GCP, such as Cloud Storage, Cloud SQL for Postgres databases, and third-party integrations available on the Google Cloud Marketplace. Meanwhile, Cloud Pub/Sub gives the company a clear and easy-to-use messaging solution that helps all the components in its infrastructure run smoothly.

Trax then began migrating VMs using Velostrata, which enables an easier process and frees up developers' time so they can continue unlocking and building on data. "It reduces the time to migrate instances to GCP from two weeks to just a few hours. Our DevOps guys love it," says Paulina Fogel, VP of Strategic Alliances at Trax.

The company has also begun working with Kubernetes, a container orchestration system originally designed by Google. With GCP, it can be managed with Google Kubernetes Engine but Kubernetes itself is open source and can be deployed in different ways across different platforms. Using Kubernetes allows Trax to automate and speed up its microservices, and at the same time make these processes available to its other cloud instances.

"We've been using as much of the Google Cloud ecosystem as we can," says Yair. "The products integrate very well, and the platform as a whole works well in a multi-cloud solution. And if we need to, we can run everything from Google Cloud Platform itself."

Multi-cloud resilience, rock-solid stability

With the adoption of GCP, Trax has built a resilient multi-cloud infrastructure capable of meeting the demands of the company's growing success without disrupting the stability of the platform. For Yair, a multi-cloud solution means having the best of both worlds. "We had to think carefully about the implementation, but it means we can now get the right tools for the right job and have the flexibility to integrate with multiple vendors and marketplaces," he says.

Working with retailers means that Trax's platform has to perform at a much higher capacity than before. A typical consumer brand might have 2,000 products, which Trax's platform would have to learn and recognize from images. A retailer, by comparison, might have around 60,000 products, all of which have to be captured, processed, and learned by the recognition model. Today, Trax technology easily handles more than a million images every day and recognizes almost a billion products every month.

"From a business point of view, Google Cloud Platform has been vital for achieving our strategic goals of reaching more retailers. We have a stable, robust multi-cloud infrastructure that also meets the commercial needs of our customers. We can address the market in a much better way than before."

Paulina Fogel, VP of Strategic Alliances, Trax

As well as stability, GCP also gives Trax the flexibility and ease of use to scale quickly if it needs to. For instance, if a brand starts a new promotion that involves repackaging its products, Trax has to gather new images, retrain its AI recognition models, and then redeploy those models back into its products. With GCP, Trax can now add resources such as compute power and memory on demand to respond quickly and effectively to its clients' needs.

"From a business point of view, Google Cloud Platform has been vital for achieving our strategic goals of reaching more retailers," says Paulina. "We have a stable, robust multi-cloud infrastructure that also meets the commercial needs of our customers. We can address the market in a much better way than before."

Trax continues to look for new improvements and products for its clients. The company has been incorporating IoT devices into its processes and is also exploring augmented reality. Trax's close relationship with Google Cloud has been key to navigating the way forward as it experiments with new products such as Google Cloud IoT and Google ARCore, as Yair explains:

"The partnership with Google Cloud has been very important for us, in terms of access to new products and working with the product team," he says. "As well as the technical partnership, Google is a great company to work with in general. Our cultures fit very well with each other."

About Trax

Trax is a world leader in computer vision solutions for retail, ranking in the Top 25 Fastest Growing Companies on Deloitte's Technology Fast 500 list. The Trax platform enables tighter execution controls in-store and the ability to leverage competitive insights to unlock revenue opportunities at all points of sale.

Industries: Retail & Consumer Goods, Technology
Location: Israel
Google Cloud Platform logo

12 Months FREE TRIAL

Try Kubernetes Engine, BigQuery, and other Cloud Platform products with $300 in free credit and 12 months.

TRY IT FREE
Google Cloud Platform logo

12 Months FREE TRIAL

Try Kubernetes Engine, BigQuery, and other Cloud Platform products with $300 in free credit and 12 months.

TRY IT FREE