Inside Greg Toroosian’s conversation with Foxglove CEO Adrian Macneil and why the next robotics breakthrough won’t come from a robot at all

If you have spent any time around the robotics industry, you have probably heard some version of the same complaint. The ideas are there. The talent is there. The funding, at least for the companies that can prove themselves, is there too. So why does it still feel like every robotics team is fighting the same uphill battle just to get their systems working reliably?

That question sits at the center of Episode 120 of Machine Minds, the podcast hosted by Greg Toroosian, founder and CEO of Samson Rose, a retained search firm that has become one of the go-to talent partners for robotics, AI, and hard tech companies. In this episode, Greg sat down with Adrian Macneil, co-founder and CEO of Foxglove, for a conversation that goes beyond the usual founder origin story. It is a grounded, detailed look at why robotics development is so much harder than it should be, and what it will take to fix it.

If you work in or around robotics, this episode is worth your full attention. Below is a breakdown of what Greg and Adrian covered, along with why the discussion matters for anyone building, hiring for, or investing in the physical AI space.

From Payments and Crypto To the Front Lines of Self-Driving Cars

Adrian’s path into robotics infrastructure was not a straight line, and that is part of what makes his perspective so useful. He started with an early curiosity for programming, then spent years building infrastructure in payments and crypto before landing at Coinbase during a formative stretch of his career. It was there that he developed a deep appreciation for what good developer tooling can do for a fast-moving team.

From Coinbase, Adrian moved into Cruise during the early rise of self-driving cars, a period he describes as eye-opening. According to Adrian, autonomous vehicles made the value of robotics tangible in away that few other applications could. You did not need to explain to someone why a self-driving car mattered. They could see it, ride in it, and understand the stakes immediately.

But what Adrian also saw at Cruise was something less visible from the outside: just how much bespoke infrastructure was required to build, debug, and scale an autonomous system. Every leading AV company was quietly reinventing the same internal tooling. Different teams across the industry were solving identical problems in isolation, wasting enormous amounts of time and engineering effort that could have gone toward the actual product.

That realization became the seed for Foxglove.

What Foxglove Is Actually Solving

Because of Adrian’s observations, Greg prodded him to explain in plain terms what Foxglove does and why it matters. The short version is that Foxglove is a data visualization platform built specifically for robotics and physical AI teams. It gives engineers the kind of off-the-shelf leverage that software startups have taken for granted for years, but that robotics companies have historically had to build themselves from scratch.

A few specific pain points Adrian and Greg discussed include:

Robotics teams are dominated by custom tooling and siloed data formats, which makes debugging painfully slow and inconsistent across companies.
Robotics data is fundamentally different from typical software data. It includes multimodal sensor inputs, massive data volumes, limited bandwidth in the field, and edge-first constraints that most data infrastructure was never designed to handle.
Without a shared platform, every incident, every simulation run, and every real-world deployment becomes its own isolated investigation instead of part of a connected picture.

Foxglove was built to be a single pane of glass for understanding robot behavior, whether that behavior occurs in simulation, during an unexpected incident, or in a live deployment. Adrian walked Greg through how the platform brings logging, visualization, debugging, and analysis together in one place, so teams are not stitching together five different internal tools just to answer a basic question about why a robot did what it did.

MCAP and the Case For Interoperability

One of the more technical, and honestly, more important parts of the conversation centers on MCAP, the open data format that Adrian and his team created. Greg asked why an open format mattered so much when Foxglove could have simply built a proprietary system and locked customers into it.

Before MCAP, robotics teams were largely stuck with ROS 1's .bag file format, which tightly coupled data storage to the ROS ecosystem. The files were hard to read or use outside of it, and, as Foxglove has written, working with them meant contending with formats that weren't self-describing or friendly to third-party tooling. Foxglove built MCAP to fix this: a self-contained, framework-agnostic container format that keeps message definitions inside the file itself, so any tool (not just ROS-based ones) can read and decode the data without needing access to the original workspace or schema files.

Adrian’s answer gets at something bigger than Foxglove itself. He argued that interoperability is a prerequisite for robotics to scale beyond a small group of well-funded teams. If every company has to build its own data format and its own tooling from the ground up, only the companies with the deepest pockets can compete. An open standard like MCAP lowers that barrier and lets smaller, scrappier teams build on top of shared infrastructure instead of reinventing it.

This is a theme that shows up again and again in the episode. Adrian is not just trying to build a successful company. Instead, he is trying to change the default starting point for what it takes to build a robotics company at all.

Death By A Thousand Paper Cuts

One of the more candid moments in the conversation comes when Greg and Adrian talk about why robotics startups struggle so much, even when the underlying technology is sound. Adrian describes it as “death by a thousand paper cuts.” It is rarely one catastrophic failure that sinks a robotics company. Instead, it is the accumulation of smaller problems across:

Hardware reliability and the physical constraints of operating in the real world.
Autonomy and the sheer difficulty of handling edge cases at scale.
Go-to-market challenges that are different from typical SaaS sales cycles.
Pricing models that have to account for hardware costs, service contracts, and long deployment timelines.
Reliability expectations that are often far higher than what customers demand from software alone.

Greg, drawing from his own experience of working closely with robotics founders through Samson Rose, connects this directly to hiring. Building a team that can absorb and solve these accumulated problems requires a very specific kind of talent, not just strong engineers but people who understand the operational realities of physical systems.

Fundraising In A Market That Does Not Yet Believe

Another candid stretch of the conversation deals with fundraising. Adrian talks openly about what it was like to raise capital for a company built around infrastructure for a market that many investors did not yet fully understand or believe in. His advice is refreshingly practical: finding investors who already believe your thesis matters more than trying to convince skeptics.

This is not a flashy insight, but it is an honest one, and it reflects a broader pattern Adrian sees in the current wave of humanoid robotics hype. He draws a direct parallel between where humanoid robotics is today and where self-driving cars were in the early days at Cruise. The excitement is real, but so is the long tail of real-world deployment challenges that will not resolve overnight. Adrian’s take is not pessimistic. It is simply grounded in what he has already lived through once.

What Foxglove Looks For When Hiring

Given that the episode is hosted by someone who runs a talent search firm built specifically for robotics and hard tech companies, it is no surprise that hiring becomes a natural thread throughout the conversation. Adrian shares what Foxglove looks for in candidates, and he keeps coming back to one particular trait: proactive ownership.

Adrian tells Greg that if he could clone one mindset across his entire company, it would be that sense of ownership, the instinct to notice a problem and take responsibility for solving it rather than waiting to be told. It is a simple idea, but one that becomes especially important in a company building infrastructure that other companies depend on.

That is exactly the kind of insight that makes Machine Minds valuable beyond entertainment. Greg has built the show into a resource for founders, operators, and talent leaders in robotics who want to understand not just the technology, but the people and decisions behind it.

A Ten-Year Vision For Robotics Infrastructure

Toward the end of the conversation, Greg asks Adrian to zoom out and share his longer-term vision. Adrian’s answer is ambitious but concrete. He imagines a future, roughly ten years out, where starting a robotics company feels a lot more like starting a SaaS company does today. Off-the-shelf infrastructure would handle the foundational, repetitive engineering work, freeing founders to focus almost entirely on solving real customer problems instead of rebuilding internal tooling from scratch.

It is a compelling vision, and one that ties the whole episode together. Foxglove is not just trying to build a good product; it is trying to shift the baseline expectations for what building a robotics company should require.

Why This Episode Matters

What makes this conversation stand out is the combination of technical depth and hard-won perspective. Adrian is not speculating about robotics infrastructure from the outside. He lived through the earliest, messiest version of these problems at Cruise, and he has spent years since then building the solution he wished had existed at the time.

Greg’s approach as a host deserves credit here, too. Rather than steering the conversation toward surface-level talking points, he consistently pushes for specifics: what the actual data challenges look like, what fundraising conversations actually sound like, and what hiring for a robotics infrastructure company actually requires. That approach reflects the same philosophy behind Samson Rose, where the goal has always been to understand the real operational needs of robotics companies rather than treating every search the same way.

Since founding Samson Rose in 2019, Greg has built the firm into a trusted partner for robotics, AI, and hard tech companies navigating hiring during a period of rapid change. Machine Minds grew naturally out of that work, giving Greg a platform to have exactly these kinds of in-depth conversations with the people building the infrastructure and technology shaping the industry’s future.

Deepen Your Understanding of Robotics Infrastructure:

Listen to the Full Conversation: Listen to this Machine Minds episode on Apple Podcasts or Spotify to hear the full discussion on the missing infrastructure holding robotics back.
Explore Foxglove’s Platform: Visit the official Foxglove website to learn more about their data and visualization tools built for robotics and physical AI teams.
Connect with the Host: Follow Greg Toroosian on LinkedIn for insights on robotics hiring, talent strategy, and the future of physical AI.
Build Your Robotics Team: If your company is scaling in robotics, AI, or hard tech, visit Samson Rose to learn how Greg and his team can help you find the talent you need to grow.

The Missing Infrastructure Holding Robotics Back