AI reached 100 million users faster than any consumer product in history. Instagram took two and a half years to get there. TikTok took nine months. ChatGPT did it in two months. By every headline metric, this is the fastest technology shift ever recorded.

But the speed is concentrated. The industries adopting AI fastest are the ones that were already digital: finance, tech, professional services. Meanwhile, industries like construction, agriculture, and sanitation, work that holds everything else up, are largely still waiting.

Construction is one of the least digitised major industries. Productivity growth has lagged almost every other sector for decades. Firms spend less than 1 percent of revenue on technology, compared to 3.5 to 4.5 percent in automotive and aerospace. Billions have been invested in construction tech in recent years, and the productivity numbers have barely responded.

Agriculture tells a split story. Around 70 percent of large-scale farms in developed countries use at least one form of AI-driven technology: precision farming, crop monitoring, yield prediction. But that number doesn't reach most of the people actually farming. Small and mid-sized operations, especially outside developed economies, face the same barriers they always have: cost, connectivity, training, fragmented infrastructure. The technology exists. The access doesn't.

These are the industries most people never think about until they stop working. The food on the table, the building you're sitting in, the systems that keep a city functional. Office workers interact with the outputs of this labour every day without noticing. The work only becomes visible when it breaks.

Technology follows return. Builders and investors gravitate toward industries with wide margins, fast growth, and obvious ROI on a new tool. Agriculture and construction offer none of those. Margins are thin, environments are unpredictable, and the workforce is resistant for reasons that aren't irrational. Construction crews still coordinate entire projects on WhatsApp. No central task management, despite that technology existing for decades. The tools aren't missing. The incentive to adapt them is.

I grew up around agriculture. I've seen what the work costs and what even small improvements mean for the people doing it. In hours, in physical strain, in decisions made under uncertainty that compound across a season. Most of tech doesn't think this challenge is worth it. I do. Not for monetary reasons. These industries feed, house, and sustain everyone else. That should be reason enough.

The gap won't close on its own. And it won't close with tools designed for an office and repurposed for a field. It needs solutions built for the constraints these industries actually face, by people willing to start where the margins don't make it easy.