The Grantham firm putting AR on the shop floor
Strip an engine down to its block, measure every bore, replace what's worn, rebuild it to manufacturer tolerance, certify it roadworthy. That is what happens, day after day, on the factory floor at Autocraft Bridford — a Grantham engineering firm that has been remanufacturing automotive engines for decades. It is not a production line in any conventional sense. Each engine arrives in a different condition, carrying its own history of wear, neglect, or damage. Workers must read what they find and respond accordingly, drawing on hard-won practical knowledge rather than following a fixed sequence.
What makes Autocraft's recent trajectory unusual — and worth examining — is that this emphatically hands-on, high-judgment workplace became one of the more-cited provincial examples of AR and digital twin adoption in UK manufacturing. According to Made Smarter programme case study outputs, the firm introduced AR-assisted digital work instructions and created a virtual model of its production floor: tools more commonly associated with large automotive plants than a market-town engineering business with no software heritage.
The interesting question is not whether the technology works in controlled conditions — there is reasonable evidence that it can. The question is what the shift actually demands of the people doing the job.
What AR and digital twins do in a factory this size
Think of the AR overlay less as a screen and more as a heads-up display projected onto the work itself. A worker approaching a stripped engine block sees step-by-step reassembly guidance anchored visually to the component in front of them — torque values, sequence prompts, cautionary flags — without having to consult a separate paper manual or step away from the bench. The instruction travels with the eye rather than sitting beside it. In a 2023 controlled study of experienced workers on manual repair tasks, this shift produced a 21% reduction in task completion time and a 26% drop in perceived cognitive workload compared with paper-based methods.
The digital twin works differently: it is a virtual replica of the production floor, used for planning and diagnosis rather than live guidance. At Autocraft's scale, the practical applications include simulating layout changes before equipment is moved, identifying bottlenecks in the workflow, and trialling new process sequences without interrupting live production. Changes can be tested virtually first — a meaningful advantage for a firm where any unplanned downtime has a direct cost.
Neither tool operates autonomously. In remanufacturing, where every engine arrives in a different state, AR instructions must accommodate variable fault conditions and non-standard reassembly sequences; the system presents options and data, but the worker reads the actual component and makes the call. The technology does not override craft judgement — it structures and informs it.
What the research says about AR and repair work
The 2023 study behind those figures — published in the International Journal of Advanced Manufacturing Technology — deserves closer examination than the headline numbers suggest. Its participants were experienced metalworkers performing manual, object-specific repair tasks: not novices learning a fixed sequence, but skilled workers whose judgement was already well-developed. Every single participant preferred the AR system to paper instructions. In studies of technology adoption among experienced manufacturing workers, unanimous preference is unusual; resistance to new tools in this cohort is well documented.
The cognitive workload finding is the more significant result. A 26% reduction in perceived mental effort does not mean the task became easier — the physical and diagnostic demands of repair work remained unchanged. What decreased was the load associated with finding and holding information: locating the right page, cross-referencing tolerances, remembering where in the sequence a worker had reached. That freed attention for the parts of the task that genuinely require experienced judgement — reading a worn surface, deciding whether a measurement is within acceptable tolerance, or recognising an anomaly the manual does not anticipate.
The analogy to engine remanufacturing at Autocraft is close enough to be useful. Both contexts involve experienced workers, manual repair rather than repetitive assembly, and high variance between individual jobs. That does not guarantee identical outcomes — controlled study conditions differ from a live production environment — but it makes the evidence more transferable than AR studies conducted in highly scripted, high-volume assembly settings.
Why workers push back — and when they are right to
Unanimous preference among experienced workers — as recorded in that 2023 study — is not, however, the typical finding. A 2021 survey of 263 manufacturing workers identified substantial barriers to AR adoption operating at the worker level: technology resistance and low perceived utility were the two most commonly reported, and their magnitude varied in structured, predictable ways across groups defined by tenure, skill level, and age.
The pattern is not random, and it is not simply stubbornness. Workers who have spent years building fast, reliable mental models of their tasks have the most to lose from disruption and the least obvious immediate incentive to adopt a system that may, in its early iterations, be slower and less accurate than their own accumulated knowledge. A veteran engine remanufacturer in Grantham who can feel whether a bore is within tolerance has something the AR overlay does not yet possess. Resistance in this context can be a rational assessment: this system doesn't yet know what I know.
That reading deserves to be taken seriously rather than treated as an obstacle to route around. When experienced workers push back on AR instructions, they may well be identifying gaps — edge cases the system has not been trained on, prompt sequences that do not reflect actual component variance, or interface designs that obstruct rather than support judgement. Pushback from the most skilled quarter of a workforce is often the most accurate diagnostic of where an implementation remains incomplete.
The skills that experience built — and what AR now needs from them
A 2025 study drawing on 87 stakeholder interviews across European manufacturing firms — including SMEs — offers a useful way to map what is actually being asked of workers at a site like Autocraft. Its authors identify three distinct skill demands, each operating at a different level of a worker's relationship to the production process.
The first is reskilling: adapting to changed task-level procedures. For an Autocraft engine builder, this means learning to follow an AR overlay step-by-step rather than the memorised sequence that experience has made automatic. The second is upskilling: engaging with real-time system data at the process level — reading dashboard information about machine state, flagging anomalies in engine history, interpreting what the system is reporting and deciding what to do about it. Neither of these is trivial, but both can be addressed through structured training.
The third mode — what the research calls craftsmanship — is different in kind. It describes the embodied, product-level knowledge that experienced workers carry: the ability to sense that something is wrong before any sensor flags it, or to recognise that a component's actual condition does not match the condition the AR prompt was written to address. This dimension cannot be trained into a new hire in a few weeks, and no AR system yet replicates it.
The practical challenge that follows is knowing when to trust the overlay and when to override it — a judgement that requires both layers working together. The study explicitly pushes back against the idea that workers are passive executors in these environments; tacit knowledge is not a legacy liability to be replaced but the capability that makes the whole system function when conditions diverge from the expected. Workers arriving at this technology are not starting from zero. They are translating one kind of expertise into a setting that now demands an additional register.
How Autocraft got here — and what it means for other firms in South Kesteven
None of this happened because Autocraft simply decided one morning to become a digital manufacturer. The firm accessed the Made Smarter programme — a UK government–industry initiative designed specifically to lower the financial and expertise barriers that prevent manufacturing SMEs from engaging with Industry 4.0 tools. Made Smarter provided both matched funding and dedicated digital transformation advisers: in a firm without an internal technology function, those advisers were the gateway, not an optional extra.
The structural logic here extends beyond Grantham. A 2024 study of manufacturing SMEs found that frontline worker insights are routinely underutilised in smaller firms — not through malice but because of limited resources, management capacity, and a tendency to focus narrowly on technical upskilling rather than whole-process change. External intermediary institutions are, in that research, identified as the primary mechanism for bridging this gap. Made Smarter is precisely such an intermediary, and Autocraft's experience is a working illustration of what that bridging looks like in a Lincolnshire market-town setting.
South Kesteven has other manufacturers with comparable profiles — experienced workforces, manual precision processes, no dedicated digital capability — who face the same barriers Autocraft faced before it engaged with the programme. Whether they know equivalent support exists is a different question.
The Autocraft story is ultimately a people story dressed in technology language. The headsets are visible; the harder work — persuading a veteran engine builder that the overlay is worth trusting, and knowing when their judgement should override it — is not.
- [1] Working with technology: The case for worker-centered innovation. (2024). https://doi.org/10.1177/02690942251359179 https://doi.org/10.1177/02690942251359179
- [2] Evaluating digital work instructions with augmented reality versus paper-based documents for manual, object-specific repair tasks in a case study with experienced workers. (2023). https://doi.org/10.1007/s00170-023-11313-4 https://doi.org/10.1007/s00170-023-11313-4
- [3] Augmented Reality in Manufacturing: Exploring Workers' Perceptions of Barriers. (2021). https://doi.org/10.1109/TEM.2021.3093833 https://doi.org/10.1109/TEM.2021.3093833
- [4] A Study of Commercial Augmented Reality Applications in Manufacturing: A Subject Matter Expert Analysis. (2024). https://doi.org/10.1016/j.procir.2024.04.003 https://doi.org/10.1016/j.procir.2024.04.003
- [5] Fourth Industrial Revolution. https://en.wikipedia.org/?curid=39773873 https://en.wikipedia.org/?curid=39773873
- [6] Reframing the narrative of workers' agency in Industry 5.0 manufacturing through reskilling, upskilling and craftsmanship. (2025). https://doi.org/10.1108/jwl-05-2025-0154 https://doi.org/10.1108/jwl-05-2025-0154
- [7] Digital twin. https://en.wikipedia.org/?curid=47896295 https://en.wikipedia.org/?curid=47896295
- [8] Towards Smart Manufacturing Metaverse via Digital Twinning in Extended Reality. (2025). https://doi.org/10.1115/1.4070437 https://doi.org/10.1115/1.4070437