The Best AI Opportunities Are in Markets Nobody Is Glamorizing

01 /The Concept

Cloneable built a platform that "automates and scales work that previously required experienced experts."

The distinctive angle: this is physical work – specifically, inspecting whether communications towers or power lines have been installed correctly. Visual inspection only, but inspection that's genuinely complex and requires significant hands-on expertise to get right.

The platform has two components: a mobile app for field workers, and the AI agent that powers it.

The app itself is straightforward – point the phone’s camera at a cell tower, and it returns an assessment across all required parameters, of which there are, surprisingly, quite a few. Beyond convenience, the app cuts meaningful time and cost from both inspections and training. One customer previously sent two-person teams with specialized equipment, averaging 6–7 towers per hour. With Cloneable, one person trained on the app in a single hour can inspect 8–10 towers per hour.

The underlying AI agent is what makes this possible.

To train the agent on a new type of inspection, you feed it video recordings of experts performing that procedure – multiple recordings, multiple experts, varying conditions. The more input data, the better the agent gets at identifying recurring patterns and adapting to edge cases.

After analyzing the videos, the AI agent produces a documented procedure. Once approved, it gets loaded into the app and goes live for field workers. The underlying model is purpose-trained on industrial operations analysis and doesn't require documentation – it learns entirely from video. Results can also be exported to the specialized process-documentation platforms that enterprises use for this type of work.

Cloneable launched the first version of its platform in February of last year. By the end of this year, expected annual revenue will be 100x what it was at launch. On the back of that growth, the company recently raised $4.6M.

02 /Why It Matters

We spend so much time inside computers and the internet that we forget our digital world rests on a physical one – millions of cell towers carrying mobile connectivity, millions of kilometers of fiber carrying broadband, miles upon miles of power lines, pipelines, and enormous quantities of everything else that keeps modern infrastructure running.

As one of Cloneable's investors puts it: "$30 trillion of the US economy depends on physical infrastructure that is aging and failing faster than it can be maintained. And the bottleneck isn't money – it's the decades of expertise that live inside the heads of the people doing the maintenance."

The problem is structural: those experts are aging out of the workforce, and younger workers are reluctant to fill their shoes. Transferring tacit knowledge through apprenticeship and mentorship is slow, expensive, and error-prone. Meanwhile, the infrastructure keeps growing – making the maintenance gap increasingly critical.

Physical infrastructure maintenance is a massive market with massive problems and massive money flows. And most startups simply don't go there – because their founders never look in that direction.

As the founder of Squint ([covered here](/review/zdes-u-tebja-100-shansov-pobedit-glavnogo-konkurenta)) put it: "For some reason, manufacturing and industrial operations are abandoned topics in the tech world. Even now, it's nearly impossible to gather founders building in manufacturing, because almost none are in Silicon Valley. So when we talk about our main competitors, we're not talking about other startups – we're talking about binders full of printed instructions."

Squint built a conceptually similar platform to Cloneable – but for factory workers who can point their phone at a machine and get step-by-step maintenance or troubleshooting instructions. Squint has raised $59M, including $40M last August. DeepHow ([covered here](/review/programmirovat-nuzhno-ne-kompjutery-a-ljudej)) raised $37.1M on a comparable concept. Zaptic raised $19M in the same space.

A more recent example: Circuit ([covered here](/review/ii-umejushhij-otvechat-na-voprosy-bolshe-ne-nuzhen)) raised $30M in February and has attracted more than $70M total, building an expert knowledge transfer platform for manufacturers and service companies.

XOi ([covered here](/review/tema-v-kotoroj-mozhno-i-horosho-zarabatyvat-i-horosho-prodatsja)) raised $200M last February for an app that lets a technician photograph the nameplate on equipment and receive both a step-by-step diagnostic procedure and a step-by-step repair guide.

Staying closer to Cloneable's specific domain of infrastructure inspection: Dutch startup Perry ([covered here](/review/peresadka-golovy-kak-vostrebovannaja-biznes-model)) raised its first €1.6M last summer for a simpler platform that nonetheless helps companies address the shortage of qualified inspectors and maintenance specialists.

03 /Opportunities

Building AI solutions is obviously the hottest area in technology right now – no argument there. But the question isn't whether to build AI; it's what to build it for.

Want a large market and minimal competition? Go into the physical world. Manufacturing, infrastructure maintenance, equipment repair and servicing. That's the general direction, with plenty of specific niches to choose from based on what feels most natural.

The money is there. The problems are there. Many of them are solvable with AI today. But most founders simply don't look that way. So maybe you should.