Six unsexy industries where AI is solving real problems

When venture capitalists discuss the next wave of AI opportunities, they’re increasingly looking past the obvious targets. As Kevin Novak, Managing Partner at Rackhouse Venture Capital, puts it, there’s immense opportunity in “dirty, dusty, unsexy” industries that previous technology waves left behind. These sectors didn’t participate in mobile transformation or big data adoption, creating what he calls “two decades worth of economic lag” now ready to be harvested. The timing for a major disruption could hardly be better. Labour shortages have reached crisis levels across multiple sectors: 62% of contractors are struggling to find skilled workers according to the Business Barometer 2024 survey; meanwhile, the US is facing thousands of unfilled HVAC technician positions, and industrial laundries are automating simply to keep operations running. 

Meanwhile, sustainability mandates are forcing companies to track and optimise resource use with laser precision. AI and robotics, arguably, are not theoretical improvements in these contexts, but operational necessities that directly impact whether businesses can fulfil orders, meet regulations, and remain profitable. What makes these industries particularly compelling is their inefficiency. When recycling rates have flatlined for years, even modest technological interventions create outsized impact. The companies pioneering solutions in these spaces are seeing an opportunity others appear to have neglected; so, who are these companies, and what sectors are they looking to serve?

Waste management: turning trash into treasure

AI-powered sorting and smart sensors are rescuing billions in recyclables, and cutting disposal costs by up to 20%.

The global waste management industry handles over US$400bn annually, yet it operates with stunning inefficiency. Nearly 40% of global food production ends up as waste, and for the global economy that’s roughly US$1tr in losses shouldered by everyone from farmers to restaurants. Meanwhile, the US recycling rate has plateaued around 35% for years, while an estimated US$2.6tr worth of valuable materials in fast-moving consumer goods goes unrecovered annually. Instead it just gets thrown out; put in landfill, serving no use to anybody at all. 

The core problems here are pretty straightforward: traditional sorting facilities struggle to keep pace with the complexity and volume of waste streams; collection routes operate on fixed schedules regardless of actual need; and there’s minimal data visibility across the supply chain. Manual sorting, on the other hand, is dangerous, inconsistent, and prohibitively expensive. As venture capital firm Capital Ventures noted, these inefficient, high human-capital operations have “ample opportunity for AI not just to incrementally improve existing processes, but to create entirely new possibilities”. Between 2024 and 2025, waste management has seen a surge in AI and robotics deployment. Automated sorting systems using computer vision are capable of outpacing even large teams of human workers. Simultaneously, smart collection systems with IoT sensors report fill levels and waste types in real-time, enabling dynamic routing that cuts fuel costs and prevents overflowing bins.

Rescuing billions from landfill

Among the market leaders in recycling AI is technology firm Amp Robotics, which raised US$91m in late 2024 to expand its AI-driven sortation systems. The company’s robots pluck recyclable materials (i.e. plastics, metals, paper and so on) off conveyor belts at remarkable speed, while their AI system continuously learns and adapts by analysing millions of waste images. In next-generation plants, these systems have been able to trim the amount of manual sorting down to virtually zero, dramatically improving reliability and throughput. The company has begun operating full recycling plants itself, since the technology so substantially improves efficiency that it justifies direct facility operation rather than just selling equipment. “Amp’s AI sortation systems enable consumers to recycle both with and without curbside separation,” said Amp investor Abe Yokell, adding that the technology is “prime example of how AI can deliver meaningful environmental and economic benefits.”

Meanwhile, smart waste collection is optimising the front-end. Re-Learn’s ‘Nando’ system retrofits existing bins with sensors and cameras that recognise what people discard and guide proper sorting in real-time. Deployed across seven countries, the company claims these smart bins have improved waste segregation quality from roughly 30% to 75% – all while cutting clients’ disposal costs up to 20% through better recycling and fill-level monitoring.

“[Back of house industries] didn’t participate in mobile; they didn’t participate in big data. If I can figure out what is the problem holding you back from doing your job 10 times better, I effectively get the opportunity to harvest two decades worth of economic lag.”

Kevin Novak, Managing Partner, Rackhouse Venture Capital

HVAC services: the AI inside every AC unit

HVAC units contribute massively to CO2 emissions and drag up building costs, but AI agents can slash costs by as much as one third and predict failures.

HVAC services – that is, installing and maintaining furnaces, air conditioning units, and chillers – are a ubiquitous presence in modern life, found everywhere from homes and offices to data centres. Puzzlingly, however, the industry has been remarkably slow to modernise. Despite massive investments in building automation, 67% of commercial buildings still operate with reactive maintenance strategies that result in 25-40% energy waste and unexpected equipment failures that erode profits and leave people feeling either very hot and sticky, or rather cold. As it stands, the sector faces converging pressures. Experienced HVAC technicians are retiring while fewer and fewer young workers enter the field. According to The Access Group, the shortage currently sits at 110,000 workers in the US, and could increase to 225,000 by the decade’s end. 

These shortages could prove highly detrimental for the sector; after all it is no secret that few things upset a customer more than being stuck in an uncomfortably hot room during a heatwave. Meanwhile, energy costs and sustainability mandates are pushing building owners to optimise their consumption, but traditional systems lack the intelligence to adapt dynamically. Many HVAC contractors still schedule maintenance by phone, dispatch technicians with clipboards, and fix equipment only after it breaks. This reactive approach wastes energy, shortens equipment life, and creates expensive emergency repairs. The opportunity, then, lies in making HVAC systems self-aware and predictive through AI that learns building patterns, anticipates failures, and continuously optimises performance.

Cutting energy use by a third

Tech players are slowly cottoning on to this opportunity. Now, companies like Johnson Controls and Siemens are deploying agentic AI tools that learn building patterns – for example, occupancy, weather, thermal characteristics – and continuously adjust HVAC settings for optimal efficiency. In one rollout, Johnson Controls demonstrated a 35% reduction in HVAC energy use across more than 500 buildings by using AI to predict and fine-tune temperatures dynamically. Meanwhile, Siemens reported 40% lower equipment maintenance costs after implementing predictive analytics to catch issues before they escalate.

On the other side of the equation, HVAC manufacturers are also embracing software integration. Trane Technologies acquired BrainBox AI in late 2024, a startup whose deep-learning algorithms automate HVAC systems and reportedly cut energy use up to 25% and emissions 40%. BrainBox’s technology now operates in over 14,000 buildings worldwide, turning ordinary commercial HVAC setups into self-managing systems that anticipate needs and continuously optimise. With Trane’s distribution network, such AI will likely be embedded into new units from the factory, making energy-inefficient ‘dumb’ systems as outdated as a landline telephone.

Food distribution: from farm to fork, faster

The food waste problem is out of control, but can AI systems better match production to consumption?

Before your dinner hits the plate, it must first navigate a labyrinthine distribution system encompassing farms, processors, distributors, cold storage, and countless logistics touchpoints. The modern food distribution chain is startlingly fragmented; thousands of growers, distributors, and millions of foodservice outlets spanning a global footprint. Some food is caught in one country, sent to another to be processed, then a third for packaging – and that’s before it even makes it into packaging. It should be of little surprise that these elaborate networks often suffer from little or no end-to-end transparency or accountability. This fragmentation leads to overproduction as a hedge against uncertainty, which means surplus food is routinely discarded.

The statistics around this are sobering. Over 10% of global greenhouse gas emissions come from the food system, with up to 10% of that purely from food waste. We’re looking at a situation where around 1% of emissions come from producing food nobody even eats. Financially, food waste represents US$1tr in annual losses borne by farmers, distributors, retailers, and restaurants. Among the culprits: inaccurate demand forecasts, paper-based ordering systems, inefficient routing, and cold chain failures where temperature-sensitive cargo spoils due to equipment malfunctions. It’s no secret that this sector desperately needs visibility and coordination. When restaurants call in orders via voicemail, distributors manually transcribe them into spreadsheets, and trucks operate on static routes regardless of actual demand, inefficiency compounds at every node. 

Interconnectedness is the answer

The solutions to this staggering problem are straightforward – even if the underlying tech isn’t. Berlin-based Choco, for example, has built an ordering platform and AI assistant that acts as a digital conduit between tens of thousands of restaurants and their distributors. Instead of voicemails and scribbled orders, restaurants use Choco’s app to place orders in seconds, while distributors get a real-time dashboard of incoming orders. It goes without saying that this appreciably reduces errors and saves everyone a great deal of time. Choco’s mission explicitly targets food waste, noting that nearly 40% of production is wasted largely due to the fragmented, analogue structure of the industry. By 2024, Choco had expanded to six countries with 15,000 restaurants and 10,000 suppliers on its platform, having raised over €300m (US$347m) in funding. 

Cold chain management has simultaneously taken a giant leap forward, in no small part thanks to IoT sensors that live temperature and location tracking for shipments. If a cooler in a truck fails, the system can quickly reroute to nearby cold storage before food spoils. The IoT cold-chain logistics market hit US$114bn in 2024 and is projected to reach approximately US$204bn by 2030, highlighting massive investment flows. These technologies not only reduce waste but ensure food safety; no more uncertainty about whether fish stayed below proper temperatures during transit.

Pest control and facilities hygiene: AI against the ick factor

Smart traps and autonomous cleaning robots are taking undesirable work out of the hands of underpaid workers.

Pest control and janitorial services epitomise ‘dirty work’: exterminators crawling under floorboards to set rat traps, cleaning crews working graveyard shifts scrubbing toilets, polishing floors, . You name the unenviable chore, rest assured they do it. These services are vital for public health and business operations, but traditionally very low-tech and labour-intensive. The sector faces what industry observers describe as cleaning’s ‘triple threat’: high demand, fewer available workers, and pressure to control costs. Traditional pest control relies on scheduled visits where technicians manually inspect every trap and bait station regardless of activity. However, this routine approach wastes time checking empty traps while potentially missing actual infestations. Janitorial work is physically demanding, often poorly compensated, and difficult to staff consistently – particularly night shifts. 

Little surprise, then, that the result is inconsistent service quality, frequent turnover, and facilities that can’t maintain desired cleanliness standards. Both subsectors need solutions that reduce the need for constant human presence and improve detection and response. The solution again comes in the form of IoT and automation. Indeed, IoT-enabled monitoring can alert technicians only when intervention is needed, while autonomous cleaning robots can handle repetitive floor care tasks that consume most janitorial labour hours. This would then (ideally) allow these workers to redirect their efforts toward higher-value activities that require actual human judgment.

Smart traps and cleaning bots

In pest control, IoT-enabled ‘smart traps’ can detect when rodents are caught or when insect activity spikes and instantly alert technicians via mobile apps. Companies like Anticimex have deployed thousands of digital mousetraps that send signals when they snap, enabling on-demand response rather than wasteful routine rounds. Advanced detection systems analyse environmental factors like temperature, humidity, and even sounds to spot infestations early and manage them more effectively. On the robotics side, drones can inspect hard-to-reach areas such as rooftops or tall warehouses for pest nests, while some systems use climate data and AI to predict pest surges; for instance, forecasting mosquito breeding spikes after heavy rain combined with heat. This intelligence-driven model focuses efforts where and when needed, saving labour and reducing indiscriminate chemical use—beneficial for both safety and environmental compliance.

On the facilities hygiene side, autonomous cleaning robots have proliferated in malls, airports, hospitals, and supermarkets worldwide. Companies like Brain Corp have already deployed over 20,000 robotic floor cleaners. These machines work alongside human janitors, handling dull floor care tasks as humans focus on detail work like disinfecting high-touch surfaces or restocking supplies. An unnamed major grocery chain recently deployed Brain Corp’s robotic scrubbers across more than 100 stores saved thousands of labour hours and improved cleaning consistency. With labour shortages, one robot can offset hiring an extra night-shift cleaner, and machines reliably meet elevated hygiene standards without complaint.

Maritime shipping: automating the high seas

The world depends on shipping routes, so why is this pivotal industry still trapped in 20th century thinking?

Global trade floats on maritime shipping: a centuries-old and highly conservative sector that is responsible for moving 90% of world trade by volume. For decades, the industry operated with mountains of paperwork, legacy IT systems that don’t communicate, and labour-intensive port operations where longshoremen manually coordinate crane operations via radios and spreadsheets. Ports, the crucial nodes where ships load and unload, have often been bottlenecks with hundreds of trucks idling whilst workers orchestrate cargo movements. Finally, however, the sector is ready to embrace more cutting edge solutions. A 2025 survey by Flagship Founders found that 45% of maritime tech startups are leveraging AI, up from 27.5% the year before – a pretty rapid integration jump. 

Over the past twelve months, maritime tech startups raised US$234m in disclosed funding rounds, a 73% increase from the previous year’s US$135m. Key pain points include inefficient scheduling, safety risks in congested waters, lack of real-time coordination between stakeholders, and administrative bottlenecks from paper-based documentation. ‘Smart ports’ – the ostensible next generation of maritime shipping – use networks of IoT sensors, 5G connectivity, and AI algorithms to coordinate logistics. Automated cranes and self-driving trucks move containers with minimal human oversight, whilst AI systems optimise berth assignments and equipment allocation. The benefits extend beyond efficiency: better scheduling means ships spend less time waiting at anchor burning fuel, and precise tracking improves security and reduces theft.

Transformative automation at Shanghai’s biggest seaport

Shanghai’s Yangshan Deep-Water Port stands as a striking example. One of the world’s busiest facilities, Yangshan embraced extreme automation and now operates with only 40% of the workforce a traditional port of similar volume would require, yet achieved 213% increased efficiency in cargo handling. Most container moves are performed by self-driving trucks and automated cranes guided by a central AI that schedules tasks to the second. These machines run around the clock without breaks, vastly increasing throughput. At the time of writing China has automated 52 port terminals in this vein, with facilities like Qingdao Port boosting throughput 15% and efficiency 6% by adopting smarter cranes and systems. Automation isn’t limited to China – although it generally leads there – Rotterdam and Hamburg in Europe have semi-automated terminals, while the UK’s Port of Tyne deployed a private 5G network in 2022 to enable real-time control of assets, becoming the country’s first 5G smart port.

The US has been a particular laggard in terms of port automation, but there are signs of movement there too. California’s state government awarded US$27m to major ports in 2024 specifically to develop data standards and digital infrastructure for better coordination. Meanwhile, startup Orca AI raised US$72.5m in 2025 to enhance ship navigation safety using computer vision, helping crews avoid collisions in congested waters. Meanwhile, digital maritime firm Marcura acquired Shipster, which uses AI to automate document processing like charter party contracts and bills of lading, addressing the administrative bottlenecks that have plagued the industry for generations.

“I don’t want someone to think that HVAC has not changed in the last hundred years. It’s changed in the last five to seven years, pretty dramatically. And we all need to be able to take notice of it.”

Dave Regnery, Chief Executive, Trane Technologies

Textiles and industrial laundry: robots that sew and fold

Autonomous sewing robots and folding systems promise to reshape fashion supply chains and eliminate exploitative labour.

The textiles industry and industrial laundry sector are massive yet rarely discussed. Currently, literally 100% of clothing items are produced by hand, with high labour costs driving production to Southeast Asia under often exploitative conditions – a widely-maligned practice known as the sweatshop. A single t-shirt might require as many as a dozen people, all working lengthy shifts. The clothing industry is responsible for a staggering 10% of global CO2 emissions when the supply chain is accounted for. Meanwhile, over 90% of manufacturing workers earn woefully insufficient wages, and nearly a third of produced clothing cannot be sold and ends up in landfills. From top to bottom (no pun intended) it’s a shoddy situation.

On the other side of the clothing spectrum, industrial laundries face parallel challenges: giant facilities consuming massive amounts of water and energy, with labour-intensive sorting, feeding machines, and folding operations. Both sectors struggle with workforce availability, sustainability pressures, and economic models built on exploitation, inefficiency or both. Consumers and regulators increasingly demand ethical, sustainable production, which current systems can’t deliver at scale. Fortunately, technology could offer a path forward. Robotic sewing was long considered too difficult due to unpredictable fabric behaviour, but startups have cracked it. Automated folding systems using computer vision can now handle soft textiles in industrial laundries. These innovations promise to reshape global supply chains, potentially moving production closer to consumers while eliminating exploitative labour practices.

Sewing robots, reshaped supply chains

Vienna-based Silana has developed what it claims is the world’s first autonomous sewing robot: robotic cells that take fabric from roll to finished garment with minimal human input. In 2024, the company raised €1.5m to accelerate the technology, explicitly aiming to “counter precarious working conditions” in garment manufacturing. The founders argue that by automating, local production becomes economically feasible again, eliminating the need to exploit labour abroad or ship clothes across oceans. Silana envisions overproduction being “effectively reduced to zero” through on-demand manufacturing. Their robot cells perform each step – cutting, sewing seams, adding prints – consistently within minutes. The company’s goal is for their ‘SiBot’ machines to produce one billion garments annually regionally, and they already had nearly 200 machines pre-ordered by major producers worldwide in 2024. 

Over in industrial laundry, German startup Sewts created Velum, a robot that automates folding towels and linens. One large Dutch commercial laundry, Rentex, installed robotic arms to lift and fold healthcare linens, reporting it “moved into robotics, big time” to cope with labour shortages. The Velum system handles 500 to 600 textiles per hour, matching human speed, and with software updates the system is quickly advancing well beyond human capability. While it comes with an undeniably steeper upfront cost than simply hiring someone, the system reportedly pays for itself in roughly two years and can run multiple shifts without breaks, filling gaps where companies simply can’t find willing workers.