Vay Transforms City Driving in Las Vegas with Snowflake’s AI Data Cloud

New York has the Subway. Paris has the Metro. Las Vegas has…cars. The metropolitan area of Las Vegas is one of the most car dependent metros in the United States. It’s also one of the most expensive places to have a vehicle, which makes automobile ownership both a necessity and a burden.

Founded in 2018, German remote driving company Vay is transforming the face of transport for the Vegas residents and its 40 million yearly visitors. Bridging the gap between a car share app and ride-hailing service, the company uses remote, human drivers to transport electric vehicles on demand to customers, wherever they are in the city. The service provides all the advantages of having a car, and none of the hassles of owning one — which makes navigating Vegas less of a gamble. 

“Our mission is to have fewer vehicles just standing around in cities,” says Philipp Leufke, AI Principal Data Analyst at Vay. “We want better space utilization, and for people to have less need to own a car without that being restrictive.”

Data plays a vital role in delivering Vay’s flexible car share experience. Since 2023, the company has been using Snowflake’s AI Data Cloud to power the insights that drive the business.

To provide its new approach to driverless mobility, Vay collects and works with huge amounts of data. The majority of this comes from its “Remote Driving Stations,” where Vay’s professionally trained Remote Drivers sit and control the vehicles to deliver them to customers or park them at the end of a trip. The Remote Driving Stations are equipped with automotive-grade steering wheels, pedals, brakes, and other vehicle controls, as well as monitors. 

Data from the Remote Driving Stations includes telemetry data and live video streams, but the company also collects data from its mobile application, customer interactions, CRM systems and more.

“We need data to tell us where to position our vehicles to provide the best service and what routes we should take to stay within network coverage,” says Leufke. “All of our decisions as a company are made with data.” 

In processing this data, the company’s classic data lake architecture often encountered out-of-resource errors, which meant Vay had to split its data models into smaller and smaller chunks. Long query times also led to high costs, and the engineering workload required to overcome these issues was burdensome.

Due to these query performance issues, fewer stakeholders engaged with data. “If people open dashboards and then have to go make a coffee while they refresh, they don’t think about experimenting with data or extending use cases,” says Leufke. “So really it was restricting innovation.”

In 2023, Vay decided it was time for a change. Having previously worked with Snowflake in a different role, Leufke realized the platform would meet the company’s needs in terms of compute power, scalability and more. 

“We ran a proof of concept to compare it with other options, and Snowflake was orders of magnitude better,” Leufke says.

With the decision made, Vay’s migration to Snowflake was fast and simple. The company migrated all of its data workloads into the AI Data Cloud in just three months. 

Today, the company uses Snowpipe to ingest data into Snowflake from an Amazon S3 storage environment. This workflow enables fully schema-agnostic ingestion to accommodate Vay’s diverse and evolving data sources — something that’s vital for the speed the company requires to work with its large amounts of data.

“This is one of Snowflake’s biggest strengths,” says Leufke. “The variant data types are fantastic, because whatever we get from upstream we can just dump into a single variant column.” 

Vay’s data comes from close to 80 different sources, and serves every department in the organization, from product and software engineering to marketing and operations. Users can then access the information they need through a range of tools including Grafana for time series analysis and Apache Superset for exploration and visualization. The company also extensively uses Dekart — a mapping tool that Vay’s Director of Software Engineering, Volodymyr Bilonenko, developed.  

With regards to data governance, Vay uses Snowflake’s Dynamic Data Masking to protect personally identifiable information related to both employees and customers — something that was challenging with the previous data lake architecture.

“Dynamic data masking simplifies a lot of things,” says Leufke. “ Before, we’d have to create duplicate tables, which was resource-heavy. But with Snowflake, we can protect data without having to hide specific tables from anyone.”

For a company on the move, being able to tie data to locations is a vital part of Vay’s data strategy. Vay uses Snowflake’s geospatial features and a dashboard created within the AI Data Cloud to provide a near-real-time view of how customers are using its service — and fuel long-term strategic decisions. The platform allows Vay’s decision makers to see where chargers are, where parking is and the optimal routes Remote Drivers should take to deliver vehicles.

“Snowflake keeps adding geospatial functionality, which is really important to us,” says Leufke. “We use it for planning which areas we want to support with our remote driving.” 

Vay also brings data into Dekart to reveal long-term trends through its mapping functionality. 

“With Dekart, you can put properties on maps to visualize user demand or people’s journeys,” says Leufke. “We can also see how demand hotspots evolve and adjust our strategy accordingly.”

This combination of insights provides Vay with a comprehensive view of all its user and vehicle activities, both in the short- and long-term. This empowers the company to improve efficiency, customer experiences, vehicle maintenance and operational strategy.

When state-of-the-art technology acts as the basis for the business, bugs and issues will inevitably pop up. It’s important to Vay that its customer support efforts are fast, helpful and, where possible, preemptive to minimize disruption.

To achieve this, Vay uses Snowflake Cortex AI to categorize customer calls and better understand where problems occur. 

Using Snowflake’s AI_TRANSCRIBE feature, Vay can automatically transcribe audio files of its customer calls. Cortex then cleans up the transcription, before applying a SQL prompt that summarizes each issue.

These summaries enable Vay to identify and solve individual issues with its app or vehicles, as well as spot emerging trends, so it can make proactive and preventative improvements. Previously that would have been done manually, but with Cortex AI automating the process, there’s been a huge internal boost to efficiency. Leufke estimates this has already saved customer service agents over 300 hours.  

Vay can also use Cortex to automatically change and refine the categories it uses to organize these calls and improve the accuracy of its insights — a process that would simply be too time consuming to undertake manually. 

“Before switching to Snowflake, so many things took too long or were too hard to achieve. Snowflake enables us to unlock new and more efficient ways of working,” says Leufke.

After migrating to Snowflake, Vay saw instant results in the performance of its data infrastructure. Almost immediately, it was able to see that its BI and user-facing queries were 12x faster than before — with far more queries running than previously. As Leufke explains, the situation is significantly different from the frustratingly slow dashboards users were dealing with before.

Faster dashboards with Snowflake mean users start filtering and drilling down into data. The shorter feedback cycles have fueled innovation and increased user adoption. It’s not just performance that Snowflake provides Vay with though — it’s also simplicity. As a managed service, its minimal engineering requirements mean Vay’s data team has been able to redirect its resources to other areas of the business.

“With Snowflake, we’re hardly ever babysitting our data pipelines or compute clusters,” says Leufke. “The resource issues that we had to work around are all gone, and that makes maintaining the service very, very easy. When it comes to the things that are central to our company’s success, I want to rely on something that just works. That’s Snowflake.” 

With less time spent firefighting, Vay’s data engineers can now focus on driving the future, informing company strategy and providing Vegas residents with convenient, safe and efficient transport. In a city where anything can happen, the company has complete peace of mind thanks to a data platform it can trust.