Hamid Montazeri, SVP of Software & AI at Locus Robotics – Interview Series

As Senior Vice President (SVP) of Software and Artificial Intelligence at Locus Robotics, Hamid Montazeri brings over 30 years of experience leading and scaling globally distributed teams. He specializes in modular and scalable software architecture and the application of transformative technologies such as cloud, IoT, big data, and AI/ML. Throughout his career, Hamid has delivered differentiated solutions and products for companies ranging from startups to multinationals, powering applications in autonomous robotics, intelligent warehouse automation, and supply chain systems serving industries worldwide.

Locus Robotics designs and delivers AI-powered warehouse automation solutions by combining autonomous mobile robots with intelligent software. Their platform, called LocusONE, orchestrates fleets of robots to handle tasks like picking, putaway, transport, and mezzanine operations—boosting productivity by two to three times while cutting labor costs. The system integrates seamlessly into existing warehouse environments and scales flexibly, enabling operations to deploy a few robots or thousands without major disruption.

You’ve held leadership roles at organizations ranging from CNN to Stanley Black & Decker, Dematic, KION Group, and now Locus Robotics. How has that diverse journey shaped your perspective on where AI and robotics can deliver the most impact?

Holding technical roles that work closely with AI and software for over three decades, my perspective has been heavily shaped by how I’ve witnessed the industry evolve.

When I first started out, the industry was at a time when software development efforts were embracing transition from structured to the object-oriented paradigm on the programming languages front and from a single process monolith software to software components that could run on different operating system processes and leverage inter-process communication to achieve goals in a distributed manner. We were on the cusp of a major shift, moving software away from being run on a particular machine to instead becoming distributed leveraging inter-process and/or network communication technologies. Over time, this focus shifted to different areas, driving new developments from programming underlying systems to leverage emerging networking and Internet capabilities to increase the scalability of systems, as well as deployments, data storage, leading to eventual cloud development.

These changes were instrumental for the industry, as they incrementally brought a level of compute and storage elasticity,  and introduced new opportunities for AI and robotics. These advancements continued as I progressed throughout my career, and I got a firsthand look at the impact that AI and robotics can bring, especially within the supply chain and logistics industries. We’re now at a point where robots possess the ability, in terms of compute, storage, and AI,  to accurately navigate and operate in massive high-density environments, with complex geometries, such as warehouses and deliver major business impact, including reduced costs, enhanced throughput,  improved flexibility, and enhanced labor performance.

What does “physical AI” mean in the context of warehouse automation, and how does it differ from more traditional robotics or generic AI models?

Physical AI is the future of logistics and warehouse automation. It’s the backbone powering autonomous operations through a combination of real-time perception, decision-making, and continuous learning, allowing robots to instantly optimize every decision.

With physical AI, robots aren’t just powered to only move goods in highly structured and permanently fixed configuration environments. They’re equipped with intelligence that goes beyond memorizing a specific warehouse layout and bringing something from point A to point B. If things within their environment change, they can automatically adapt and reroute the planning to make the best decisions based on current conditions.

Robotics that doesn’t utilize physical AI will be increasingly limited in its applicability. Interestingly, even the application of modern general foundation models are not highly effective in creating the type of physical AI that addresses application domain needs. When you look at warehouse automation, the goal is to have processes be as efficient as possible and generic models can’t truly deliver this. The reality with generic models is that they’re not designed to efficiently handle domain specific concerns such as efficient navigation and interaction with associates in warehouse environments. Physical AI, when equipped with a specially developed warehouse foundation mode,  provides a purpose-built approach that ensures robots are performing as efficiently as possible with the ability to automatically adapt and adjust to deliver the best outcome at all times.

How do AI-driven robotics systems adapt to constantly changing environments, such as new SKUs, shifting layouts, or sudden spikes in demand?

These areas are all ones where physical AI, equipped with an underlying domain foundation model, excels. As demand spikes, layouts change or new SKUs get introduced, robots powered with physical AI are equipped to seamlessly navigate constant change.

This is why robots powered by physical AI are optimal for logistics. This purpose-built approach can actually keep pace with the constant flux that the industry typically faces.

Why do you believe domain-specific AI is more effective than chasing broad foundation models when it comes to supply chain and logistics?

Domain-specific AI is most effective for supply chain and logistics in general, but especially within warehouse automation.

When it comes to warehouse automation, having domain-specific models is what ultimately takes automation to the next level. A general foundational model isn’t designed to handle the challenges that warehouse environments typically face—like safety and navigating complex layouts—which means as these challenges arise, operators will continue to face the burden of these obstacles.

In contrast, domain-specific models are powered with the necessary industry expertise to understand how to actually solve these challenges. Leveraging domain-specific models lifts pressure off operators and automatically pulls domain expertise to offer and apply solutions to challenges as they arise.

What measurable outcomes have you seen from deploying physical AI in warehouses, whether in throughput, downtime reduction, or error rates? 

At Locus Robotics, our solutions powered with purpose-built physical AI have delivered impactful results across our customers’ warehouses, including:

• Improving order accuracy to 99%
• Decreasing error rates 04% to 0.01%
• Reducing downtime by taking sites live in just weeks vs. months
• Enhancing throughput allowing customers to double or triple their operational throughput

How do you approach safety, reliability, and human oversight when deploying autonomous systems in high-volume operations?

When automating high-volume operations like warehouses, approaching safety, reliability and human oversight all start with the solutions you select.

These areas are critical within the design process and underscore why those looking to automate should prioritize solutions that are purpose-built during the decision-making process.

At Locus Robotics, our autonomous mobile robots (AMRs) are designed to meet and exceed industry safety standards. Our LocusBots utilize multi-sensor safety systems with cameras and light detection and ranging (LiDAR) that help them avoid collisions and obstacles, keeping warehouse operations and workers safe.

Reliability is also at the core of what we do. For warehouse operators, consistently negotiating demand surges is a reality of the business they’re in, but we believe this doesn’t need to be a pain point. At Locus Robotics, our robotics-as-a-service (RaaS) model allows operators to automate their warehouse environments without the upfront cost or time investment that automation typically requires. Instead, our solutions enable operators to automatically scale up or down to meet current demands, ensuring they’re always equipped to adapt to surges whenever they arise.

In high-volume operations, human oversight will always be required, but the key to successful automation is leveraging solutions capable of taking on greater responsibilities. A key differentiator of Locus’ AMRs is our combination of Discrete Event Simulation (DES) techniques with detailed robot autonomy models, which allows operators to design concepts of operations and accurately simulate the most efficient use of bots within their environments, helping to streamline pick time and ensure orders are shipped on time—two crucial components for all operators.

What challenges arise in integrating AI-powered robotics with existing warehouse management and ERP systems, and how do you address them?

Integration implementation time is typically viewed as the biggest challenge that comes with automating. Operators will need to examine existing technology stacks and consider moving away from outdated legacy systems. Given AI uses vast amounts of data, they’ll also want to examine their existing infrastructure to ensure it’s able to withstand and support AI models.

Depending on the solutions they select, operators may end up in a position where their transition requires a big time commitment, as they may need to migrate systems and undergo extensive training for their workforce.

Locus Robotics aims to remove time as a challenge for operators. Our AMRs are designed to seamlessly integrate with these systems, helping customers avoid the costly upfront time investment required for transition and training by other solutions on the market.

How scalable are these solutions across different warehouses and geographies, and how much customization is typically required?

The easy scalability is what makes flexible automation so ideal for warehouse environments. Traditional automation systems require significant upfront costs and long-term time commitment.

Automation that use a RaaS model, like Locus Robotics, allow warehouses to deploy and seamlessly adjust their fleet size based on demand. Meaning, as demand ebbs and flows, operators can scale their operations accordingly across their warehouses.

As a global leader in warehouse automation, our solutions are scalable across all geographies for our customers. Our dashboards offer real-time insights into key warehouse performance metrics—like units and picks per hour as well as worker productivity. This unified visibility empowers operators to easily scale solutions across environments, meeting operational needs based on specific warehouse requirements.

With Locus Robotics solutions, customization is easily accommodated with no effort on the part of customers; things are designed to help each individual customer scale based on their own unique needs. 

How do these technologies change the role of human workers in warehouses, and what kind of upskilling or change management is needed?

AMRs are completely redefining standard warehouse positions for human workers by creating safer working environments and opening up new opportunities for the human workforce.

Locus Robotics provides a uniquely intuitive approach to AMR-associate interaction and makes employee training/change management for deploying and operating robotics automation a very light lift for operators. For example, when deploying LocusBots on-site, training can empower associates to start working with the robots in 10 minutes or less—meaning operators and employees don’t need to worry about being held back by stressful and time-consuming and costly training.

AMRs can reduce the burden of physical labor-intensive tasks, minimizing the risk of injury and human worker fatigue. By overtaking responsibilities like lifting heavy objects and long-distance travel within the warehouse, AMRs improve conditions for human workers by taking on the brunt of the tasks that require strenuous effort. They can also alleviate human workers from more repetitive responsibilities, like picking, providing an opportunity for focus to shift to more complex tasks.

Introducing AMRs into the warehouse is a great opportunity to enhance roles for human workers. Those looking to automate should also focus on identifying the upskilling opportunities this brings for their human workforce to explore higher responsibilities and identify areas where entirely new roles can be introduced, including positions that work directly with the bots, like analysts that oversee bot performance data.

Can you share what kinds of AI models and approaches power these systems, and whether they operate more at the edge or in the cloud? 

At Locus Robotics, our AI solutions focus on three traits: physical (AI embedded in the warehouse and designed to adapt to its environment), trustworthy (AI capable of explaining its decisions) and holistic (AI that orchestrates the warehouse as a system vs. just powering a single bot or task). For us, it’s about building AI that truly understands the warehouse and can deliver real results for our customers’ environments.

Data is the foundation of any model, and Locus’ deep industry expertise, paired with our treasure trove of real-world data—including nearly 6 billion units picked—allows us to build domain-specific models and develop systems designed to be warehouse-first.

At Locus Robotics we leverage AI both at the edge and in the cloud: our AMRs leverage edge AI to accomplish what physical and trustworthy AI demands while our “warehouse wide system of records-to system of actions” strategies and warehouse foundation model work draws upon the scalable computational power that cloud provides.

Looking ahead five years, what major advances or shifts do you expect in AI-driven robotics for logistics and supply chains?

The biggest shift we’ll see is that physical AI will dominate. As businesses continue to examine the return on investment (ROI) they’re getting from their AI investments, operators will closely assess the results they’re seeing from automating.

We can expect that those who’ve placed budget behind automating with solutions that aren’t purpose-built won’t be seeing the ROI in their warehouses that they’re hoping for. Solutions not powered by physical AI lack the necessary understanding to excel in these environments. This will spur operators to prioritize putting budget behind physical AI solutions, which can optimize every decision in real-time and deliver the results they’re seeking.

Alongside the rise in physical AI, we’ll also see robotics in logistics and supply chain move away from generic/general foundational models to focus on developing domain-specific ones. As mentioned, operators will be seeking ways to increase their ROI, and solutions that utilize domain-specific models are a critical part of this.

For AI to succeed, we’ll see the industry gain a better understanding of why we must invest in AI that benefits from and incorporates genuine domain expertise. Accordingly, we will focus on putting development and resources behind AI designed to thrive within supply chain and logistics environments.

Thank you for the great interview, readers who wish to learn more should visit Locus Robotics.