May 14, 2025
The enterprise XR space has come pretty far in the last 5-10 years. In more and more cases, pilot programs turn into wider-scale rollouts rather than end in purgatory. Check out three examples of XR pilots that made it to the rollout phase:
AMAZON
In 2024, Amazon incorporated virtual reality and hazard simulation into driver training as part of a $1.5 billion investment in safety.
Pilots of VR-enhanced Delivery Service Partner (DSP) training took place in Rochester, New York and other select Amazon delivery stations with dual goals of 1) making driver training more interactive and realistic and 2) getting new drivers quickly and safely on the road. The three-day training experience began in the classroom, where every week nearly 140 prospective drivers learned delivery basics, getting to preview the on-the-job experience, in VR. In addition to VR, a “trip, slip, and fall” simulator allowed trainees to practice driving on slippery and elevated surfaces.
Now in 2025, Amazon continues to use VR and simulation technology for delivery driver training. Instead of three days, the training takes place over the course of two days, with drivers “going through the basics” on day one in VR.
This includes situational simulations in which, for instance, trainees learn what to do in the case of an unleashed dog. On the second and final day, a dedicated simulator provides hands-on training, teaching drivers to move through unstable conditions with packages in hand.
Nicknamed the “last mile driver academy,” the training program is officially intended for Amazon’s delivery service partners (DSPs) or third-party delivery businesses that partner with Amazon. The logistics giant plans to roll out in more markets later this year.
Amazon has also explored AR to assist delivery drivers with on-foot navigation to customer delivery points. In 2024, the company announced plans to deploy custom smart glasses that display turn-by-turn directions in the “last 100 yards” of package delivery. Voice-operated and thus hands-free, the glasses allow delivery workers to carry packages more efficiently and feature an outward-facing camera to confirm delivery and identify obstacles in the wearer’s path.
Most recently, Amazon invested in IXI, a Helsinki-based AR glasses developer. The move may or may not be related to the above project “Amelia.” The ultimate goal? Faster deliveries, of course.
BANK OF AMERICA
In 2019, Bank of America worked with Strivr on a 400-person Oculus Go pilot in which participants were guided through a variety of scenarios in a virtual financial center. The pilot took place at The Academy, the bank’s internal L&D division, allowing users to practice difficult conversations and other interpersonal skills traditionally taught via instructor or eLearning.
In October 2021, Bank of America officially launched VR training in nearly 4,300 financial centers to teach a range of soft skills to approximately 50,000 client-facing employees. In addition to VR lessons for complex tasks like fraud detection, Strivr developed virtual simulations of client interactions, including one requiring the user to practice empathy with a customer following the death of a relative. Real-time analytics help identify skill gaps and provide personalized follow-up coaching where needed.
At this time, BOA was growing its library of VR modules and investing in AI to enhance role-playing opportunities in VR. Some of the reported results from the initial pilot and wider-scale rollout included faster learning, higher retention, and brand buzz. Feedback was overwhelmingly positive: Employees found the VR training content enjoyable and memorable and felt more confident after using VR, leading to increases in both employee and customer satisfaction.
In 2023, Bank of America announced that more than 200,000 of its employees had used VR to prepare for real-world situations like talking down an angry customer and remaining calm during a robbery. The bank hoped to have over 40 VR training experiences by the end of the year.
More recently, the bank created a metaverse in which 2,000 people from all over gathered for an unforgettable virtual onboarding. New hires experienced “a day in the life” at a virtual bank branch, interacting with avatars of real Bank of America employees located across the country. Portals launched additional immersive experiences, including a museum showcasing the bank’s history, a tropical island where employee benefits were explained, and a unicorn ride for “de-stressing.”
In 2024, Bank of America announced plans to add 10-20 modules to its VR training program every year. Among the 40+ lessons already available is iCoach which uses AI for role playing with clients, managers, and direct reports. In one simulation, you’re a manager conversing with a distressed associate who’s threatening to quit. VR provides a safe space to play the scenario over and over without real-world consequences and the option to listen to yourself at the end to hear how you come across.
LOWE’S
Lowe’s has been exploring customer-facing AR apps like in-store navigation and at-home measurement since 2016. In 2022, the home improvement giant began testing store digital twins using Nvidia’s Omniverse platform and Magic Leap 2 as a means of visualizing and interacting with store data to optimize operations and empower retail associates. Initial tests took place in Mill Creek, Washington and Charlotte, North Carolina.
For Lowe’s, a store digital twin opens up a variety of augmented reality applications like restocking support and X-ray vision. Headset-clad associates on the ground could refer to the twin to stock products in the right configurations and place AR “sticky notes” to collaborate with centralized store planners in real time. Instead of climbing a ladder to read labels in the stock room, an associate could simply look up and, with the help of computer vision and Lowe’s inventory APIs, see what’s inside any box as an AR overlay.
A store digital twin integrating product location, sales performance, customer traffic data, historical order information, etc. would enable Lowe’s to analyze and test different layouts and store displays to optimize the in-store experience and boost item sales. You could, for instance, view items frequently purchased together and direct associates to place them closer together to reduce friction for customers; or even simulate different store setups using AI avatars created in Lowe’s Innovation Labs. With Omniverse, it’s possible to run hundreds of sims in a fraction of the time it would take to build a physical prototype.
That same year, Lowe’s also announced it would make over 600 photorealistic 3D product assets from its library free for other Omniverse creators to use in their virtual worlds.
Fast forward to 2025: Lowe’s has created 3D digital twins of all of its stores, each of which is updated multiple times per day and uses physics AI to account for size, weight, etc. of every product. Merchandisers use the twins to make store-level decisions including optimal product placement and analyze sales performance.
Consider a product that’s not selling well. It may be there’s a similar, cheaper product nearby or the product isn’t close enough to eye level, or both. Optimizing layouts in the physical world can be expensive and disruptive. Store digital twins allow merchandisers to simulate hundreds of different (virtual) configurations, out of which three may be chosen for physical testing at the Lowe’s Design Center.
Other possibilities include using store digital twins to train robots for tasks like lifting large boxes, and store walkthroughs at any time from Lowe’s headquarters. Lowe’s is also using generative AI to provide on-demand information to associates via a chatbot available on Zebra handheld devices. The retailer plans to roll out a similar chatbot for customers in the spring to answer questions, make suggestions, etc.
LESSONS LEARNED
We can glean several lessons from the three cases above:
- START SMALL: Starting small - 100 volunteers, one or two select locations (a manageable user group) - allows you to test the waters and easily identify positives and negatives.
- DEFINE GOALS: In each use case, there were clear goals: To get new drivers on the road quickly and safely (Amazon,) make soft skills training more engaging and memorable (Bank of America), and optimize store layouts and empower retail associates (Lowe’s). Goals help determine KPIs.
- COLLECT FEEDBACK: User feedback is key to measuring value because ROI can be both quantitative and qualitative.
- DON’T ONLY LOOK AT STATISTICS: It cannot be stressed enough: ROI isn’t only quantifiable. In the case of Bank of America, for example, it was possible to measure accelerated learning but the decision to continue using VR was more about providing engaging learning and the ability to simulate complex, nuanced conversations not possible in a classroom setting.
- SPENDING LESS VS. SPENDING RIGHT: ROI isn’t a straightforward cost comparison. Your aim should not be to spend less but rather to spend in the right way to get the right result.
Image source: Bloomberg
