A glasshouse near Boston that changed the question
Six miles south of Boston, on flat Lincolnshire fenland that has been farmed for centuries, a 26-acre glasshouse at Carrington is producing strawberries in January. That alone might sound like a curiosity. The detail that changes it into something more significant is that no human hand picks them.
Dyson Farming's facility at Carrington is not a research installation or a technology demonstration. It supplies M&S, Sainsbury's, Ocado, and Co-op on a year-round basis, drawing on more than 1.2 million plants to yield roughly 1,250 tonnes annually. By January 2024, BBC Lincolnshire reported that 20 robot pickers — trained on tens of millions of strawberry images to distinguish ripe from unripe fruit — had already produced around three million strawberries at the site. The infrastructure is comprehensive: 24-metre rotating plant wheels to track available light, AI-guided picking arms, autonomous UV-light robots for chemical-free pest control, an on-site anaerobic digester for power, and rainwater capture for irrigation.
Lincolnshire's existing concentration of soft-fruit production, its flat terrain, and its proximity to large distribution networks made it a practical choice for a shift of this scale. The question the Carrington glasshouse quietly poses is not whether robotic harvesting can work — it already does — but what that means for the people who have always done it.
What the robots can and cannot do
Cambridge-based Dogtooth Technologies, founded in 2016, supplies the fifth-generation picking system deployed at Dyson Farming and is offering 50 units commercially for the 2026 season — a signal of where the technology now sits on the journey from prototype to product. A £14 million funding round, closed in July 2026, supports that trajectory.
The picking economics are worth stating plainly. Each Dogtooth unit carries two arms that harvest around 12.5 kg of strawberries per hour — roughly half the rate a skilled human picker achieves. The operational advantage comes from continuity: with swappable 8-hour batteries and LED arrays for night-picking, a single robot produces 200–300 kg per day without a shift pattern. What changes the calculation is not speed but hours.
Where the system genuinely outperforms human labour is in consistency and grading. Five on-board cameras assess every berry for size, ripeness, rot, powdery mildew, and deformation; a patented 360-degree inspection system sorts fruit directly into retail-standard punnets, removing a separate repacking stage. The University of Lincoln's Lincoln Centre for Autonomous Systems contributed a cloud-based mapping tool, developed through the Agri-OpenCore project, that generates a full robot navigation roadmap for a strawberry farm in under one hour — substantially reducing the time and friction involved in deploying a new unit on site.
The limits are real, though. For all the precision in visual grading, robots still do not replicate the dexterity or situational judgement a picker brings to damaged plants, unusual growth patterns, or unexpected obstacles. Those tasks remain human work.
The labour pipeline that broke
For decades, the UK soft-fruit harvest ran on an informal but reliable arrangement: workers from Poland, Romania, and Bulgaria arrived each summer under EU freedom-of-movement rules, picked through the season, and left. Brexit dissolved that pipeline overnight.
The government revived the Seasonal Worker Visa scheme in 2019 to fill the gap. By 2025 it ran to a 45,000-visa annual cap, extended since to 2029 — but with a six-month stay limit that makes multi-season retention impossible. In 2024, 46% of all temporary worker visas granted in the UK were for seasonal agricultural work, a figure that shows how structurally dependent the sector remains on migrant labour even a decade on from the referendum.
The scheme has not run smoothly. In 2024, mid-season revocations of sponsor licences left farms without workers during peak ripening windows, resulting in unharvested fruit and direct crop losses. Rising compliance costs and wage pressures squeezed already thin margins. In July 2024, the Migration Advisory Committee described 'an overwhelming need for long-term certainty for food labour needs' — a conclusion the NFU had been arriving at independently for years.
Against that backdrop, the commercial logic of robotic investment becomes legible without requiring endorsement. A machine with swappable batteries does not need a visa, cannot be recalled mid-season, and does not age out of eligibility after six months. Farms were not choosing between robots and a stable workforce; they were choosing between robots and ongoing uncertainty.
Augmenting or replacing — what the industry says, and what it doesn't
The industry's preferred framing is clear. Marion Regan, Managing Director of Hugh Lowe Farms — one of Dogtooth's commercial partners — has stated that the robots 'will form an integral part of our harvesting team, augmenting our valued seasonal workforce.' That is a reasonable and probably sincere position: fruit farms still need humans for plant management, machinery oversight, and the kind of tactile judgement that cameras cannot yet replicate.
But 'augment not replace' is also self-serving, and in the absence of transparent workforce data from individual sites, it cannot be verified. Dyson Farming has not published figures on how many seasonal worker positions at Carrington have been reduced since automation arrived. Without that baseline, the claim is an assertion, not a finding.
Public commentary has gone the other direction — some media and AI-generated summaries have framed robotic harvesting as potentially ending Britain's dependence on foreign fruit pickers entirely. That overstates what the evidence shows. One commercial site, however well-equipped, does not redefine the sector.
What the debate mostly sidesteps is an uncomfortable finding from 2026. A paper published in the Industrial Law Journal and submitted in evidence to the House of Lords Horticulture Sector Committee documented illegal recruitment fees charged to seasonal horticultural workers — people driven into debt to loan sharks to secure a UK picking job. That evidence sits awkwardly beside both the optimist and pessimist framings on automation. If robotic harvesting is expanding, it is doing so alongside a labour pipeline that, in documented cases, generates serious exploitation. Whether automation eventually relieves that pressure — or simply reduces the number of people exposed to it — is a question the industry has not yet answered plainly.
What we genuinely do not know
Two gaps in the public record are worth naming precisely because they shape how confident any claim about automation's impact can be.
The full cost economics — unit price, financing, maintenance, set against seasonal wage costs per harvest — remain commercially confidential. No grower has published a head-to-head comparison. Dogtooth's £14 million funding round, closed in July 2026, signals investor confidence; it does not settle whether robotic harvesting is yet cheaper than seasonal labour at current scale.
More consequential is the worker-transition question. When a farm reduces its seasonal headcount, the public record says nothing about what those workers do next — whether they find equivalent agricultural work, return abroad, or exit the sector entirely. The seasonal visa framework tracks grants, not post-departure outcomes. The 2026 Industrial Law Journal paper documents conditions workers face entering the system; conditions on leaving it are entirely undocumented.
The 2026 harvest season is the first with 50 Dogtooth units offered commercially — the point at which meaningful comparative data could, in principle, begin to accumulate. Whether any of it reaches the public record is a separate question: farm economics are rarely disclosed in detail, and nothing in current UK policy requires growers using robotic systems to report on workforce changes. The next twelve months will be the first real test of whether the industry's claims about augmentation can be checked against evidence.
What this means for Lincolnshire — and the people still in these fields
Lincolnshire is not watching this shift from a distance — it is living the early draft of it. The county is home to the UK's most complete working example of automated strawberry production, and the questions that will eventually face much of rural England are already pressing here: not whether automation is coming, but how a community manages a transition that is already under way.
The reality on the ground is messier than either optimist or pessimist framings allow. Robots, seasonal workers on six-month visas, and a policy framework designed for neither are operating simultaneously at the same sites, with no coordinated account of what the combination is producing. The workers most directly affected — seasonal migrants with the least security in the local economy — have little presence in any regional conversation about the future of Lincolnshire agriculture. Their working conditions, and what happens to them when a farm reduces its headcount, are not tracked by any local body.
The practical obligation this creates is specific. The Greater Lincolnshire Combined Authority, which assumed its devolved economic development responsibilities in 2024, is the body best placed to monitor workforce change at sites adopting automation and to require transparency as a condition of any public support for adoption. So far, no such monitoring is in place. That absence — rather than the question of whether robots will eventually outpace human pickers — is the thing most worth pressing on. The data needed to have an honest conversation about this transition exists somewhere on these farms. It has simply not been asked for.
- [1] Agricultural robot – Wikipedia. https://en.wikipedia.org/?curid=11005995 https://en.wikipedia.org/?curid=11005995
- [2] Automation and robotics in the cultivation of pome fruit: Where do we stand today?. (2020). https://doi.org/10.1002/rob.22000 https://doi.org/10.1002/rob.22000