Logistics Startup of the Month: Gather AI

6 min read

Every month we select one logistics startup which represents a positive example of innovation in Logistics and Supply Chain and has the potential to alter the way the industry operates. This month, Transmetrics selected Gather AI, The world’s first software-only autonomous inventory management platform for modern warehouses, as the “Logistics Startup of the Month” for its outstanding and innovative approach to the inventory management.

In order to learn more about the company and what makes Gather AI unique, we have talked with Sankalp Arora, Co-founder and Chief Robotics Engineer at Gather AI, about the journey and how drones can disrupt the inventory management.

First of all, congratulations on becoming the Logistics Startup of the Month! Can you please briefly introduce our readers to Gather AI?

Thanks for picking us, it is quite exciting to be selected. Gather AI provides a software-only automated inventory monitoring solution using a fleet of autonomous drones. We make it possible for businesses to monitor inventory in their warehouses faster and at a lower cost by providing the capability to do hourly physical counts, bringing warehouses up to speed with e-commerce.

The thesis around which we built the company is that we need to bring the power of data analytics to physical business processes. Businesses need large scale, autonomous data gathering tools to be able to harness the power of analytics. Our background in developing autonomous aerial vehicles helps us realize such tools in the form of autonomous drones, combined with computer vision capabilities.

Scanning Process. Video Credit: Gather

Which industry problem are you trying to tackle with your solution? What kind of other solutions currently exist to address the same problem? What is the unique aspect of Gather AI that makes it stand out among these solutions?

We are focusing on Inventory Monitoring and Asset tracking services for warehouses and factories. The current process of collecting physical inventory data and feeding it into analytics pipelines has been slow, dangerous, and costly. Warehouse team members must climb scissors or forklifts to reach inventory stored in high places. There are RFID based solutions to automate the inventory monitoring process, but they are infrastructure intensive.

Instead, we offer an automated way to capture inventory that requires no active infrastructure in the warehouse. Drones are agile — they can capture information at multiple resolutions and can reach vantage points which other vehicles or people cannot. Gather AI’s drones can provide physical, and cycle counts with temperature measurements. The person who would previously be up on a forklift can now work from the safety of a desk. 

Other than no reliance on a wifi network or active infrastructure, another significant difference is that we are not a hardware company. We take off-the-shelf drones and automate them using our autonomy engine, which means customers get access to hardware that has been tested for over 1000-flight hours at a price point that only large scale hardware manufacturing can provide.

Can you please explain the deployment process for your solution? Are there any prerequisites for the companies to start using Gather AI?

Our deployment process involves mapping the facility and tagging it with visual landmarks (stickers). Our drones are then ready to fly and provide data. We do not need any network and other infrastructure availability. But, we do require our clients to run a 1-month pilot, where we automate a 100 ft aisle to ensure that the data is useful for them. If the pilot is a success, the service extends to the full warehouse on subscription bases. 

The pilot usually takes two days to set up with no interruption of warehouse operations. 

There are some constraints; the aisle should be broader than 5 ft 8 inches and barcode’s x-dimension should be higher than 15 mils.

What are the common operational concerns that your clients express towards your solution, and how do you address them?

The main concern the clients have is the predictability of the drone. It is understandable with autonomous indoor flying machines being a new tool that they have to share the space with. It was a significant factor in our decision to field flight-tested off-the-shelf drones.

A 1-month Pilot also helps them get comfortable with the drones operating in the facility. Our drones have now done over 700 flights in 5 customer facilities across the US.

Drone Inventory Management from Gather AI

Could you please share any recent client’s success stories? How did Gather AI help them in particular? 

The solution’s main benefit is the increase in the speed of inventory monitoring and the visibility it provides the supply chain organization by collecting images of the inventory.

The most resounding success came in the Air Cargo sector where we reduced inventory monitoring time for a single-aisle from 3 working hours to 8 minutes, by using a single drone. For the full facility what used to take them 8-40 hours to take inventory with the help of 8 people, our drones could reduce the time down to 15 minutes. The projected increase in inventory monitoring speed meant that the planes would not go partially empty because a pallet was missing in the warehouse. 

In other successes have been able to reduce the need for physical inspection of pallets for food distributors to 0, for fashion distributors to reduce their projected cycle-counting time from 3 months to a week while reducing the cycle counting from a 3-person to a 1-person job. Most importantly, the inventory control team and buyers at the facilities do not need to leave their desks to take inventory.

With your solution in place, what are the other significant challenges preventing the warehouses from being fully autonomous?

Fully automating a warehouse is still an infrastructure intensive exercise. The infrastructure or redesigning cost is the primary deterrent. The reliance of robots on an environmental structure needs to be removed. Our solution is an initial step in that direction. However, Robots need to become better at operating in environments designed for people for broader adoption.

Your product is a combination of such technologies as computer vision, drones, IoT to mention a few. These technologies are already an essential part of some of the logistics operations. In your opinion, which are the other three technologies that will improve the logistics sector in the next three years?

We have seen early adoption of the technologies you mentioned, they are already generating tremendous value. But they will still take a few years to fully realize their potential. Mobile Autonomy, Blockchain, and automated manipulation will significantly improve the logistics sector, and we will start seeing these technologies percolating supply chain industry over the next three years.

There is a strong and talented team behind Gather AI. Can you please elaborate on the background of your team members and the outstanding projects from their past that helped them bring Gather AI to life?


Our founding team is a group of PhD and research engineers from Carnegie Mellon University (CMU) with over 25 years of combined experience in developing autonomous aerial vehicles. We have had the fortune of working on the world’s first full-scale autonomous helicopter, a project that won the Howard Hughes award and was nominated for the Collier Trophy.

We then went on to develop an aerial data gathering platform that can be carried on a backpack, as Qualcomm Innovation Fellows. Working on next-generation autonomous aerial vehicles at CMU, we realized the incredible potential of drones to act as data-gathering tools. We are fortunate to be able to scale that potential by minimizing human supervision through our autonomy engine.

What major developments should we expect from Gather AI in 2020?

We are focusing on expanding with our current customers; larger installations are in the pipeline.

On the product side:

  • We will be releasing our analytics engine this year.
  • Our web-dashboard and WMS integrations will be broader. The drones are becoming smarter with features like case counting and semantic obstacle avoidance to start customer testing.
  • Our data gathering is getting faster since last year we have reduced per pallet inventory taking time from 7s to 4.7s. It will continue to decrease this year. 
Transmetrics Demo