On a cold, gray day in February 2014, Sankalp Arora stood alongside a group of United States military personnel at the Marine Corps Base in Quantico, Virginia, watching the result of three years of his work. An unmanned Little Bird helicopter simulated a resupply drop, avoiding a wire and some trees before safely landing on a frosted patch of grass, kicking up a whirl of snow. It had just completed the first fully autonomous helicopter flight demonstration. “I have it recorded in my head,” Sankalp said recently after telling the story, a big smile on his face. “It was a childhood dream come true moment for me.”

Sankalp was part of a group of Carnegie Mellon students in the university’s AirLab that had been recruited by the U.S. Office of Naval Research to work on the Autonomous Aerial Cargo Utility System (AACUS), utilizing cutting-edge robotics technology to develop a way to resupply Marines without risk to a pilot. That AACUS demonstration brought the team accolades like the Howard Hughes Award, and provided the basis for technology the Navy would continue to improve and utilize. Riding the success of the project, Sankalp and his AirLab colleagues Geetesh Dubey and Daniel Maturana — the former a childhood friend made in Delhi, India, and the latter a friend made over many late-night discussions about the helicopter — began working on another Navy-sponsored project, this time a small, autonomous drone.

When large companies began inquiring about licensing the technology from CMU, Sankalp started to look at his work differently. “It felt to me that commercial hardware was at a point where we could make a much broader impact beyond academics by taking this technology to the end user.” As he was wrapping up his PhD, he began reaching out to a wide variety of executives in retail, cargo and defense, and around 150 calls later, he was convinced that the greatest opportunity to make that impact would be through warehouse inventory management. Geetesh and Daniel were onboard, and that’s when Gather AI was born.

Gather AI’s software enables drones to autonomously inspect and monitor inventory in warehouses, replacing the time-intensive, dangerous and often clunky work of riding lifts to heights that can reach up to five stories and manually scanning barcodes and recording inventory information.

The first time we heard about what the team was building, we were admittedly skeptical, given the many venture-backed companies over the past 10 years that have tried to build custom drones to solve the warehouse visibility problem, only to disappear or plod along with just a handful of pilot sites. But after spending time with the team in their Pittsburgh headquarters, we were blown away by the elegance of their solution and their strong traction, with many customers who are expanding to multiple facilities after seeing success in one site.

In a field littered with solutions that have not worked or scaled, the Gather AI team has made a series of commercially oriented technical decisions, including the decision to use off-the-shelf hardware and radically simplify the upfront costs to deploy the solution.

We became convinced that this combination of world-class technical founders with strong commercial instincts would be the team to achieve what others could not, and that’s why we led Gather AI’s $17 million Series A-1, with participation from Tribeca Venture Partners, Dundee Venture Capital, Expa and Bling Capital.

At BCV we’ve built an incredible set of portfolio companies focused on the supply chain and warehouse space — ShipBob, FourKites, Kiva Systems (now Amazon Robotics), Vention, Cofactr and Pallet — and we couldn’t be happier to have Gather AI join them.     

Meet the Founders: Bringing robotics research to the real world

Their work at AirLab showed Sankalp, Geetesh and Daniel that their skill sets complemented each other perfectly.

Sankalp is an expert in robotics curiosity (imagine a robot assessing your face through a camera and, through its training, it’s 60% certain it’s you — if it’s curious, it will try viewing your face from another angle); Geetesh developed his robotics specialty, obstacle avoidance, as the project lead behind India’s first autonomous tank; and Daniel was the first researcher to bring three-dimensions to computer vision applications powered by deep learning.

The founders then got to work building out the wider Gather AI team. As of late March 2024, the company now has around 50 employees, and its leadership also includes COO Charlie Reverte and CTO Andrew Hoffman, an alumnus of BCV portfolio company Kiva Systems. This group’s capabilities have resulted in incredible customer love and accelerating commercial traction.

How It Works: Unmatched software for off-the-shelf hardware

The world of supply chains continues to be messy, complex and still very much managed with analog tools. As we all experienced during the pandemic, most companies have no idea where their inventory is, how much they have and when it might be available to deliver to their customers. A number of software startups have stepped in over the years to solve this. Most notably, our portfolio company FourKites pioneered the supply chain visibility space, providing manufacturers with real-time data outside their four walls — on third-party trucks, ocean vessels, railcars and airplanes. Despite the progress made in the space by FourKites and others, visibility remains a huge problem within warehouses.  

The physical nature of moving boxes, cases and pallets, combined with the sheer size of most warehouses, creates complexity. Pallets regularly get placed in incorrect locations and many of them end up sitting somewhere on the warehouse floor. Labels on cases and pallets regularly get damaged, are hard to read or have the bar code crossed out. While these issues exist within warehouse storage, they also exist during inbound receiving and outbound delivery, where warehouse managers are wondering if the right packages came in, in the right quantity.  

In the majority of even the best-run warehouses today, you’ll find humans doing visual inspection of inventory. These operations engineers walk through the warehouse with clipboards and are moved up and down a five-story warehouses in lifts. Given the cost and time associated with this manual process, these inventory checks are performed periodically, resulting in significant gaps in visibility. Not knowing what you have and where it’s located ultimately impacts revenue (you have the inventory but you can’t sell it because your system incorrectly says it’s out of stock), operational cost (warehouse management system says inventory is on shelf 120, but someone incorrectly put the item on shelf 220) and working capital (we have to purchase additional inventory because some percentage of our inventory can’t be found).  

Gather AI leverages off-the-shelf commercial drones with standard cameras to visually inspect and check the inventory in the warehouse. This solution does not require any changes or add-ons to the off-the-shelf hardware — no changes to the camera, no addition of any specialized location beacons or wireless transmitters to the drone or the warehouse and no need for other types of imaging devices. 

The magic of Gather AI’s solution is the strength of its computer vision and autonomy stacks.  

Gather’s computer vision AI has been trained on millions of cases and pallets in the real world and thus can recognize even the messiest and damaged barcodes and labels on cases, delivering best-in-class precision and recall. The autonomy stack provides instructions to the drone on its path and determines the location of the drone simply through computer vision, with no location hardware needed. 

And Gather AI’s workflow engine allows warehouse engineers to take action and trigger other systems when the inventory counts have been completed, or when inventory has been misplaced and needs to be relocated.  

The result is that Gather AI’s customers can realize a massive ROI inside their warehouses with a very low total cost of ownership. The deployment is fast and simple, and the solution simply works.  

What’s Next: Fulfilling the vision built into the name

Gather AI’s early growth has largely been built on organic customer demand. With this next step in the company’s journey, the team will be deliberate about developing and scaling its GTM strategy.

And, a bit further out, the other core focus will be, as Sankalp put it, fulfilling his and his co-founders’ full vision. “Our name is Gather AI, not Drone AI,” he said. That is, Gather AI is a software company that is using artificial intelligence to process inventory data. The team wants to eventually expand its footprint in a warehouse, which could entail things like static cameras or cameras on forklifts.

Sankalp pointed to the rise of ubiquitous cameras and the information they can gather as a turning point for supply chain logistics on par with the popularization of bar codes in the 1980s. In this early stage of a new era, the companies that utilize technology like Gather AI’s will be able to stop being reactive when it comes to their supply chains and start being proactive, saving millions of dollars annually on shrink per facility.

AI computer vision is the key to unlocking visibility across the supply chain.  A decade from now, the notion of a warehouse worker scanning a barcode will seem like a distant relic from the past. We are excited to partner with the team at Gather AI to make this a reality.