Robotic Revolution

Welcome Shareholders,

Every Sunday, We'll highlight one company to do an in-depth breakdown of their business, financial situation, and whether or not it's a worthy investment opportunity - large or small. In today's edition, we'll look at a startup looking to revolutionize e-commerce warehousing and logistics by developing advanced automation solutions.

In recent times, the venture landscape has been abuzz with discussions on generative AI, fueled in no small part by Microsoft's staggering $10 billion investment in OpenAI. A wave of generative AI startups, including Adept and Anthropic, have attracted hundreds of millions of dollars in funding, all in the hopes of transforming the knowledge industry and bringing about productivity leaps as significant as those ushered in by the personal computer and the internet.

While there is no denying the transformative potential of generative AI for white-collar jobs, a more pressing challenge looms on the horizon: the future of blue-collar industries. Factories and warehouses are bracing for unprecedented labor shortages, particularly in countries like the United States, where many of these roles have traditionally been filled by immigrants willing to work for low wages. As regulatory pressures to increase wages mount and supply-demand dynamics fluctuate, retail and e-commerce brands face the challenge of maintaining their historically high margins or being forced to choose between cost and delivery speed/convenience.

The true potential of AI and automation lies in their ability to address challenges in blue-collar industries, transforming labor-intensive tasks and paving the way for workers to contribute to higher-wage sectors more effectively. As we grapple with the future of factories and warehouses, the role of innovative solutions in mitigating labor shortages and maintaining business efficiency becomes increasingly crucial. Enter Covariant, an AI startup that's determined to be at the forefront of this transformation, revolutionizing e-commerce warehousing and logistics with cutting-edge automation technology.

Operational Uncertainty

The current state of e-commerce distribution centers presents a complex challenge for retailers. These sprawling warehouses, packed with products from floor to ceiling, store everything from consumer packaged goods and apparel to creams and pills. As online commerce continues to rise, the industry faces a growing need for robotics to automate warehouse operations and delivery. At present, distribution centers (DCs) are buzzing with activity. Customer orders arrive, crates and baskets move around, and orders are assembled. However, the work of packing and stacking boxes, moving crates, and picking items from bins is monotonous, tiring, and repetitive, making it prone to errors and mistakes. People may be performing these tasks now, but this scenario is likely to change within the next five years.

E-commerce growth has been fuelled by the COVID-19 pandemic, which has increased the demand for efficient warehousing and logistics solutions. With a shortage of workers, industry leaders are pivoting towards AI-powered robotics to enhance their operations, from package sorting to warehouse management. Retail executives are also keenly aware of the potential for AI robotics to alleviate operational uncertainty in an unpredictable market. According to a February 2023 Covariant-led research survey, over 80% of retail leaders see automation as a crucial strategy for navigating these uncertainties, and a resounding 98% plan to invest further in AI robotics in 2023, despite challenging economic conditions. As e-commerce's share of all retail sales is expected to reach over 40-50% (up from 15% today) over the next decade, it is increasingly clear that some or most of the tasks will be automated in the future.

Implementing Ai and Robotics to automate labor-intensive operations will not only lead to lower costs but also increase throughput, improve packaging efficiency, and help maintain the consistently high margins seen in the industry. This is precisely where Covariant aims to make a significant impact. The company recently raised a Series C extension of $75 million led by Radical Ventures and Index Ventures after posting a 6x growth in revenues following a strong 2022.

But how did Covariant get to where it is today? Let us take a look at the company's origins and the factors that fueled its ambitious foray into the realm of AI-driven robotics.

General AI for Robotics

Covariant, initially known as Embodied Intelligence, was co-founded in 2017 by renowned UC Berkeley professor Pieter Abbeel, along with Peter Chen, Rocky Duan, and Tianhao Zhang from UC Berkeley and OpenAI. This team of experts in artificial intelligence, deep learning, and robotics set out with an ambitious and somewhat ambiguous mandate: to revolutionize traditional robot programming, which previously required extensive time and expertise.

Covariant's aim was to create an AI layer that could be added to any existing robot, enabling robots to learn new skills without the need for explicit programming. The company took a unique approach, focusing on pure software and AI. In fact, their first hires were all AI researchers, with no mechanical engineers or robotics specialists on board. This decision allowed Covariant to delve deeper into AI than many other players in the field. In 2018, the team developed the Covariant Brain and launched their first AI robotic solutions designed to automate pick-and-place processes in warehouses and fulfillment centers.

Leveraging advanced research techniques and swiftly deploying practical warehouse solutions, Covariant demonstrated remarkable progress and success relatively quickly. In a 2019 global competition hosted by Swiss-based ABB Robotics, 20 AI startups were challenged to develop software for ABB robot arms, with their performance assessed on 26 real-world picking, packing, and sorting tasks. Covariant emerged victorious in the global contest and subsequently entered a partnership with ABB to develop AI solutions that enhance autonomous materials handling collaboratively.

Pragmatic Research

Covariant-powered robot arms have carved a successful niche for themselves in the logistics sector, but what sets them apart from the dozens of other automation focused startups? The company believes that its strategy is rooted in embracing the unknown and solving unsolved problems. By assembling a team capable of cutting-edge research, they've pushed the boundaries of what's possible in AI robotics. While much of the exceptional research talent in AI is concentrated in organizations like Google, Facebook, DeepMind, and OpenAI, these entities often focus on long-term, fundamental research. In contrast, Covariant adopts a more pragmatic approach, targeting specific real-world problems.

Covariant's said that its decision to tackle the challenges of logistics was no accident. The company found that despite numerous attempts to address picking problems in logistics and provide automated solutions, large-scale AI systems were still scarce. Covariant has emphasized that the new era of AI robotics differs from traditional industrial automation, as robots now need to handle variability, randomness, and environmental changes. The founders at Covariant spent almost a year conversing with hundreds of companies across diverse industries to understand how smarter robots could make a difference for them. Ultimately, it became evident that manufacturing and logistics were the sectors with the most pressing demand for automation, with logistics experiencing a particularly strong need.

Picking Up The Pace

Today's industrial robots excel at tasks that require speed and precision, but only when the objects they handle are consistent in shape and easy to grasp. This works well in manufacturing, where a machine repeatedly picks up the same item but poses significant challenges in retail logistics, where the objects being packed for shipping come in various sizes and shapes. Traditional programming, which relies on hardcoding a robot's every move, is highly effective in the first scenario but falls short in the second. However, when machine learning is employed to provide a system with data and allow it to generate its own rules for picking, the results are vastly improved.

Covariant employs a range of AI techniques to train its robots, including reinforcement learning—a trial and error process in which the robot has a specific goal (e.g., "move an object to a set location ") and must figure out how to achieve it independently. Much of this training takes place in simulations, where the robots can work at their own pace, often accumulating thousands of hours of experience. The result is what "the Covariant Brain" — a nickname for the neural network shared by the company's robots.

The Covariant Brain

Launched in 2018, Covariant Brain is a "universal AI for robots" designed by Covariant to be applicable to any use case or customer environment, offering a unique solution for various industries. The system is trained on data collected from millions of merchandise processing operations, allowing robots to handle a wide range of items out of the box and reducing the need for manual configuration. Covariant Brain equips robots with general abilities, such as robust 3D perception, physical affordances of objects, few-shot learning, and real-time motion planning. These capabilities enable the robots to quickly learn to manipulate objects without explicit instructions while also ensuring that the robots are trained to surpass traditional automation hardware.

Today, the Covariant Brain is powering a diverse array of industrial robots, managing tasks like order picking, putwall, and sorter induction across various industries with drastically different types of products. Aligned with its mission to create a universal AI platform that empowers AI robotics to operate autonomously in real-world scenarios, Covariant offers the Covariant Brain as a foundation for partners to build upon. This development partner model encourages collaboration and supports the integration of the AI system with the robot itself and surrounding mechatronics equipment.

As Covariant continues to expand its reach, the company targets industries where robots are needed for repetitive tasks, such as food, healthcare, retail, parcels, and manufacturing. Having already deployed its technology in numerous markets across North America, Europe, and Asia, Covariant caters to a wide range of sectors that rely on pick-and-place operations—from grocery to fashion to pharmaceuticals.

Reinventing Robotics

Covariant is reinventing robotics by focusing on versatility, advanced technology, and targeting the right customer problem. Its technology is highly adaptable and applicable across various industries, such as eCommerce, healthcare, parcel delivery, and manufacturing. Initially designed for picking and packing goods in distribution centers, the core skills modules can be used in a wide range of applications involving object recognition and manipulation.

Covariant leverages state-of-the-art deep learning techniques like reinforcement learning and domain randomization to create a highly advanced robot brain. This Brain can recognize even the most challenging objects and manipulate them with speed and incredibly low error rates, which is crucial for ensuring that human intervention is minimized.

Moreover, Covariant has identified a pressing issue that needs to be addressed in distribution centers worldwide: the struggle to find and retain labor for picking and placing tasks. These jobs are often ungratifying, tiring, and low-paying, leading to an acute need for a robotic solution. By focusing on this specific problem, Covariant's deep learning technology is perfectly positioned to meet the demand for automation in this area.

Looking Ahead

Covariant has made significant strides since its initial Series C funding in 2021. The company has successfully applied its industry-leading AI platform, the Covariant Brain, to a wide range of piece-picking and case-picking applications that encompass today's manual warehouse operations.

In less than two years, Covariant's product portfolio has expanded to include order sortation, item induction, good-to-person order picking, kitting, and depalletization solutions, all powered by a unified AI platform.

Covariant plans to use the latest investment to further develop the Covariant Brain as an even more capable foundation model. It will also enable the company to apply the Covariant Brain to an even broader range of use cases across various sectors.

Bottom Line

The rise of AI-driven robotics, particularly in blue-collar industries, is transforming the way we approach labor-intensive tasks and paving the way for workers to contribute to higher-wage sectors. As the e-commerce industry faces labor shortages and increasing costs, it becomes clear that AI and automation solutions, like those developed by Covariant, are essential in addressing these challenges and maintaining business efficiency.

Covariant's pragmatic research approach, focused on solving real-world problems, sets them apart from competitors in the AI robotics space. The development of the Covariant Brain, a universal AI for robots, allows for adaptability and versatility across a wide range of industries and applications. As a result, Covariant's advanced technology is well-positioned to meet the growing demand for automation in distribution centers and other sectors relying on pick-and-place operations.

AI and automation technologies like those developed by Covariant have the potential to increase safety in blue-collar industries. By automating tasks that can be hazardous for human workers, companies can reduce the risk of accidents and injuries, leading to a safer and more productive work environment. The integration of AI-driven robotics into industries will also lead to more sustainable business practices, as automation can increase resource efficiency and reduce waste. This shift towards sustainability is not only beneficial for the environment but also contributes to a company's long-term viability and success in an increasingly competitive market.

As e-commerce continues to grow, the industry's reliance on automation will only increase, leading to improved efficiency, reduced errors, and more sustainable growth in an ever-changing market landscape. Covariant's cutting-edge AI robotic solutions are an essential step towards this transformation, enabling businesses to maintain high margins and adapt to market fluctuations.