The challenge of sensing a world that machines still don’t fully understand

It’s easy to think autonomy has already solved perception. On the surface, everything looks like it’s accelerating: better models, more capable robots, stronger decision-making. But underneath that progress, something more fundamental hasn’t kept pace.

Most systems are still built on a narrow set of sensors that were never designed for the environments they’re now being pushed into.

GPS drops out underground, underwater, and in contested environments. Cameras need light and a line of sight. Inertial systems drift over time. As autonomy expands across defence, mining, logistics and infrastructure, these limitations stop being edge cases and start becoming real constraints.

What’s emerging is a disconnect. Machines can increasingly think and act, but they still can’t fully sense the world around them.

Seeing the problem differently

When we first met Jim Rabeau and the team at Deteqt, what stood out immediately was how they were framing the problem.

Instead of trying to improve existing sensors, they stepped back and asked a more fundamental question: what if the world is already full of signals we’re simply not using?

That line of thinking led them to magnetic fields.

Magnetic fields are everywhere. They don’t rely on infrastructure, connectivity, or ideal conditions. They are always present in the background of the physical world. But until now, we haven’t had a way to measure them with enough sensitivity and stability to make them useful at scale.

That is the gap Deteqt is going after.

Jim and the team combine deep technical expertise with a strong instinct for how this needs to work in the real world. They are building a new sensing category. And importantly, they understand what it takes to move from research into deployment.

Building from day one with Atmosphere

Deteqt was one of the first companies we partnered with through Atmosphere, our venture platform designed to help founders build momentum at the earliest part of the journey..

. At its core, Atmosphere exists to back exceptional deep tech founders and support them to translate  breakthrough science into world-class companies. That means getting involved early - partnering closely with founders to shape direction, validate with customers, connect them to the right people, and lay the foundations needed to scale. We worked closely with the team to deeply understand where the technology would create the most value, to refine go-to-market thinking by engaging directly with customers across aerospace, mining and healthcare, where sensing limitations are already becoming bottlenecks.

What stood out early was how consistent the signal was (pun intended).

Before the product was fully built, customers weren’t just interested; they were asking what this could unlock for them. In some cases, the conversation moved quickly from curiosity to urgency: what would this look like if we had it at scale?

That kind of demand rarely shows up unless something fundamental is missing.

Alongside that, the team continued to evolve quickly and hit key milestones — not just advancing the technology, but building the operational foundations, assembling an A-team, and putting in place the structure to transform into a venture-scale business. 

Turning an invisible signal into something usable

Deteqt is building a chip-scale quantum magnetometer using nitrogen-vacancy centres in diamond, integrated onto silicon with a path to scalable manufacturing.

In simple terms, they’re taking quantum sensing, traditionally confined to fragile lab environments and turning it into something that can be deployed in the real world.

What this unlocks is a new sensing layer.

Magnetic fields become something machines can actually use — for navigation, detection, and understanding environments where traditional approaches break down. It shifts sensing from something conditional to something always present in the background.

Expanding what machines can actually perceive

We often talk about autonomy as if the hardest part is intelligence. But more and more, the real limitation is much simpler: what the system can actually perceive.

Deteqt expands that.

It gives machines access to a signal that has always existed, but has never really been usable. And once you start to think about that, the implications become clear: more robust navigation where GPS fails, better visibility underground, stronger performance in complex environments, and new possibilities in areas like healthcare.

Across all of these, the same pattern shows up: existing sensing is reaching its limits.

What happens next

At Main Sequence, we often talk about the Next Intelligence Leap, the idea that the next phase of progress won’t just come from better models, but from better ways of interacting with the physical world.

Deteqt sits right in that shift. Before machines can reason better, they need to sense better and magnetic sensing is one of those missing pieces, but until now, it has not been accessible in a meaningful way.

The focus now is on turning early validation into real deployment. That means moving from pilots into production, building systems that can be manufactured at scale, and deploying into the environments where this matters most.

Breakthrough technologies often don’t introduce new physics. They turn existing physics into engineering capability. Magnetic fields were always there. The shift today is that we can finally use them, expanding what machines can sense and what systems can do.

‍