The Sensor Was Built to Find Explosives, But What Else Is in the Air?

What the sensor does today

Most readers here know that Intellistake has signed a definitive agreement to acquire NanoAi Technologies, subject to closing conditions. NanoAi's detection process was peer-reviewed and published in Scientific Reports, part of the Nature Portfolio.

The way it works, in plain terms: a lightweight nanotechnology sensor reads trace compounds in the air at parts-per-billion concentrations, with no physical contact, and can identify several distinct compounds at once.²

Parts per billion is hard to picture. One way to think about it: a single drop of something in a swimming pool that holds about fifty thousand gallons. That is the scale the sensor is designed to work at.

The applications we have discussed so far follow directly from that. Explosives. Narcotics, including fentanyl. Disease markers on exhaled breath. These are the use cases with data behind them, and the ones where security and defense customers already have budgets.

But a sensor that reads the air at that level of detail is not really an explosives sensor or a fentanyl sensor. It is an olfactory sensor. A nose, in other words. And once you have a very good nose, the interesting question is what else it might be trained to smell.

Why the 53,000 fires are an olfactory problem, not a visual one

Almost none of those fires started big. A fire begins as a hot spot in dry grass, a smoldering stump after a lightning strike, a spark from a chain dragging behind a truck. For a window of minutes, sometimes an hour, it is small enough for two people with hand tools to put out.

The problem is that nobody knows it is there.

For most of the last century, the answer was a person in a tower on a mountaintop, watching for smoke. At the peak, the United States had thousands of these lookouts, staffed through the summer by people whose whole job was to notice a thin gray line on the horizon and call it in with a bearing. It was a good system for its time.

The issue was that it also had a built-in delay, because a fire has to be producing visible smoke before anyone in a tower can see it.

Cameras replaced many of the towers. Satellites added another layer. All of them share the same limitation. They see the fire after it has become a fire, but only once there is a plume or a heat signature big enough to notice.

Smoldering wood and brush release a specific mix of chemicals long before there is anything to see. So here is the question I have been turning over…

If you placed sensors that read the air at parts per billion along a ridgeline, or a transmission corridor, or the edge of a town where the houses meet the forest, could they catch that chemical signature early enough for someone to act?

Truthfully, I don’t know for sure, but what I can say is why the question is worth asking.

Where the budgets already are

The 2026 season pushed the national preparedness level to its maximum for most of August. Federal firefighting resources were fully committed for weeks, and state suppression budgets were spent at a pace that would have looked impossible a few years ago.¹

Utilities are the other buyer. Power line ignitions have produced some of the most expensive wildfire liabilities in American history. A single failed line in the wrong conditions can bankrupt a company, and the utilities in the western states now spend heavily on anything that reduces that risk: hardening lines, cutting power on windy days, and monitoring corridors more closely than they ever have.³

And think about this, a transmission corridor is a long, narrow strip of land that a utility already owns, already inspects, and already has power running through. If sensors could sit along that strip and flag the first chemical trace of something burning underneath the lines, the utility would not need to build a new business case. It would simply be buying a better version of something it is already paying for.

As someone who watches how companies get paid, that matters. Wildfire detection is not a market someone has to invent. It is an existing market with existing budgets and an obvious gap between when a fire starts and when anyone knows.

The same question, elsewhere

The same question, elsewhere

Wildfires are just one example that grabbed me this month because of the numbers. But it’s not the only one. Other potential use cases could be:

• Gas leaks in a refinery or a pipeline corridor are a chemistry problem before they are a safety problem.

• Spoilage in a cold-chain warehouse announces itself in the air days before anyone opens the wrong box.

• Even things like mold behind a wall, corrosion inside a tank, or a battery beginning to fail in a storage facility.

Each one releases something measurable before it becomes something expensive.

What these have in common is timing. In each case there is a period where the problem exists, is chemically detectable, and is still cheap to fix. The cost curve is flat for a while and then it goes vertical. Almost every expensive failure in an industrial setting has a near invisible early stage that nobody’s equipped to notice.

All I am describing is the shape of a question that a sensor this sensitive naturally raises, and the reason I think the answer might end up being bigger than the applications we started with.

What stays constant

Whatever the sensor turns out to be able to detect, the second half of the system does not change. A detection is only useful if it reaches the right person quickly, with a location, a confidence level, and a record that can be trusted later when someone asks why the alarm went off, or why it did not.

That last point is easy to skip past. In a courtroom, an insurance claim, or a regulatory review, the question is rarely whether a sensor detected something. It is whether anyone can prove what the sensor saw, when it saw it, and that the record was not changed afterward.

Think about how that plays out in practice. A utility is facing a lawsuit over a fire that started near its lines. A border agency has a seizure that is now in front of a judge. A fire chief has to explain to a county board why an evacuation order went out at 2 a.m. In every one of these cases, the sensor reading is only the beginning of the argument. The rest of the argument is the paper trail. Who received the alert, how long it took to arrive, what the reading was at that moment, and whether anyone has touched the data since.

Proof travels with the reading

Most sensor systems are not built with that second half in mind. The data lands in a dashboard, maybe a spreadsheet, and the question of whether it would survive a hostile lawyer is left for later. Later is usually too late.

Intellistake's software layer is being built to fuse readings from many sensors into one picture, and to record those readings on tamper-proof data pathways designed to give a verifiable chain of custody of detection.⁴ The goal is a record that a court, an insurer, or a regulator can rely on without taking anyone's word for it.

That is the data infrastructure work Intellistake was doing before we met NanoAi.⁴ And it's the part that stays the same whether the compound in the air is an explosive, a narcotic, or the first chemical trace of a fire that is still small enough to stop.