As autonomous AI begins to move out of experimentation and into real world operations, investors are being forced to develop sharper points of view on where the technology creates durable value and where it does not.

To explore that question, we sat down with Andrew McMahon, General Partner at Ridgeline, one of AMESA’s investors. Ridgeline focuses on companies modernizing foundational industries such as manufacturing, energy, logistics, and defense, making Andrew’s perspective particularly relevant as AI adoption accelerates in complex, high stakes environments.

In this conversation, Andrew shares what initially drew Ridgeline to AMESA, including the importance of founder market fit, the role of machine teaching in scaling institutional expertise, and why multi agent systems are better suited to industrial and physical domains than monolithic models. He also discusses the labor dynamics shaping modern industry, the realities of selling into regulated and dual use markets, and where he sees genuine pull for operational AI across commercial and defense sectors.

Q: What convinced you that AMESA was the right team and technology to back? Were there any signals that stood out in that decision?

Andrew: The conviction started with Kence Anderson. We saw clear founder market fit. He understands the technology required to build automation, AI, and agentic systems in industrial and physical environments and can set a clear vision for that technology and execute against it.

It’s one thing to know the market you are selling into. It’s another to execute on the technical side at a very high level. Kence brings both and that combination is rare.

For us, AMESA checked several important boxes. We saw it as a compelling platform to reach the large population of industrial engineers and to start decomposing their institutional knowledge in specific processes and sectors into systems that support those industries and those engineers.

We were also excited that AMESA is focused on physical environments, not only digital workflows. When you combine that with the macro trends we see in industrial markets that are increasingly ready to adopt technology more rapidly, the decision became straightforward. It’s a massive market, with a strong technical product vision, and an exceptional founder backed by a deep team.

Q: There is a significant skills gap in industry as experts with twenty or thirty years of experience retire. Did that dynamic factor into your decision to invest in AMESA?

Andrew: It definitely did. When you describe turning knowledge into agents and software models, some people immediately jump to the idea of replacing labor. I see something different. I see an enabling technology for the renewed industrialization of Western economies.

The reality is that many economies simply do not have the labor force required to meet their ambitions. In many cases the existing workforce is aging, retiring, or experiencing high attrition. That is the backdrop.

Many of AMESA’s use cases today involve running agents alongside industrial engineers or operators. The system augments their work, or it creates higher value work for those operators by removing some of the repetitive load.

Q: When you look across the AI landscape, especially for companies selling into complex industrial or defense environments, what stands out about AMESA’s multi-agent orchestration and machine teaching approach compared with other platforms you see?

Andrew: I’ll start with the machine teaching methodology. Existing machine learning approaches are usually data first and data driven. Sometimes there is a human in the loop. Sometimes there is reinforcement learning. And those approaches can be powerful.

What excites us about the machine teaching approach is that it codifies expert knowledge ahead of deployment into operational workflows. That expertise can be knowledge of the process itself, but it also includes knowledge of goals, constraints, and success criteria that are inherent in that process.

When you can embed those elements into models and agents, you are no longer just throwing data at a model and hoping the right answer appears on the other side. That has been an important success factor for AMESA as it works with very large, complex customers. These customers are deliberate and patient. They will not adopt a system simply because it’s in the zeitgeist. They operate in settings with large capital risks, safety risks, and human life at stake. They need to be confident that the technology will work in that context.

The way I think about AMESA’s Orchestration Studio is as an org chart for AI agents. It’s a system for operating, permissioning, and moving through a process. That kind of hierarchical design, where orchestrator agents manage specialized skill agents, is a real differentiator. The reusable element is also important. As different hierarchies, processes, and agents are built, reuse becomes valuable. Over time you get a library of patterns that can be applied across new problems.

We are also very excited about the Agent Cloud. It clearly differentiates AMESA from other multi-agent AI platforms. The proving ground concept, where agents learn, fail safely, and improve in a digital representation of the environment before they operate in the real one, is powerful.

You are validating complexity before production, rather than experimenting in production with small slices of data or process. In many digital workflow AI companies, the need for a digital twin or digital representation of processes is less central. For AMESA It’s foundational.

Q: Ridgeline invests in companies that modernize foundational industries such as supply chain, energy, and manufacturing. Within those domains, where do you see AMESA’s platform having the potential to unlock meaningful operational impact?

Andrew: When we look across ICPs, we keep coming back to a pattern. Complex operations with high value decisions. That is where AMESA is likely to have the most impact.

That does not always mean heavy industry in the traditional sense. It can be highly complex and massive scale in ecommerce, which is a digital environment but still tied directly to supply chain and distribution.

Another pattern is companies with deep expertise trapped in people’s heads. That often points you toward organizations with some history. It does not mean they must be fifty years old, but they have developed institutional knowledge that is not fully captured in systems. Even relatively young companies like Google have large stores of institutional knowledge.

There are also a lot of newer manufacturing companies that will benefit as they move from factory zero to factory n. They can use AMESA to standardize and scale what works.

Finally, I do think physical systems where AI interacts directly with real world processes will be a major portion of AMESA’s customer base. That aligns with Kence’s background. He is a mechanical engineer who spent a decade at Bonsai and Microsoft serving industrial customers. It’s a natural evolution in his story, and he speaks the language of those environments. All of that shapes where we see the common characteristics for AMESA customers.