British manufacturers rarely have the luxury of choosing between speed and certainty. A food producer under retailer audit, a logistics operator running a sortation tunnel around the clock, an automotive supplier working to tight tolerances; each needs vision systems that are fast to develop and faster still to trust. That combination has historically been difficult to buy as a single package, which is why the arrival of Scorpion Vision AI at VISION 2026 in Stuttgart matters to UK engineering teams, even from several hundred miles away. Scorpion Vision Ltd, as Tordivel AS's UK company, will be bringing the platform (and what it means for application support closer to home) to customers across the country in the months that follow.
A familiar platform, extended
Scorpion Vision Ltd has supplied and supported the Scorpion Vision platform in the UK for years, built on the same deterministic engine that Tordivel AS has developed in Norway for over two decades. Scorpion Vision AI does not replace that engine or change how it behaves on a validated production line. What it adds is a structured, controlled interface (built on the Model Context Protocol) through which engineers, and increasingly AI agents working alongside them, can inspect a live application, check calibration, pull up results and compare test runs, all without taking control of the actual production decision. For UK integrators managing several live sites at once, that distinction between engineering assistance and production authority is the whole point.
What Scorpion Vision AI means on a UK production floor
The platform is built around three ways of working, and most UK sites will use a mix of all three depending on the shift and the task. An engineer can work interactively, prompting and testing while Scorpion Vision AI handles the repetitive groundwork of building and validating components. A set of agents can work in the background across multiple sites; reading logs, checking image sets, flagging anything unusual before it turns into a support call. Reports or alarms can generate themselves automatically from the same live context driving the vision process, arriving with the supporting evidence already attached rather than requiring someone to reconstruct what happened after the fact. Throughout, the real-time processing that makes the production decision stays at the edge, close to the camera, exactly where it has always needed to be for a UK line running to tight cycle times.
The camera behind the demonstration
Tordivel is showing Scorpion Vision AI running live at Stuttgart alongside a new camera, the Scorpion T-Rex 4K SH EVO, and Scorpion Vision Ltd expects strong interest from UK customers specifying new inspection or measurement cells over the coming year. It is a compact FPGA smart camera with a 9-megapixel global shutter sensor, capturing up to 290 FPS at full frame and up to 3600 FPS at reduced ROI; fast enough to freeze a can on a filling line or a parcel on a belt without the motion smear that ordinary sensors introduce. The range extends from 5 to 25 megapixels on a shared architecture, so the same camera family covers everything from high-speed code reading to detailed 3D surface inspection. For UK food and beverage lines in particular, an optional stainless-steel enclosure rated IP69K is available for equipment that has to survive daily washdown with hot, high-pressure water and cleaning chemicals, a requirement that rules out a good deal of standard industrial hardware.
Why the deterministic core stays put
It would be easy to read a story about AI and industrial vision as a story about handing judgement to a black box. That is precisely what Scorpion Vision AI is designed not to do. The pass or fail, the measurement, the coordinate sent to a robot arm, continues to come from a tested, versioned, deterministic Scorpion Vision profile, and nothing reaches a live line without an engineer reviewing and approving it first. AI is used to build and test the components around that profile (comparing performance, documenting changes, preparing evidence) which is precisely the kind of work that consumes UK engineering time without directly improving line performance. Freeing that time up, rather than removing human oversight from the process, is the practical benefit for teams already stretched across several sites.
Application support, not just supply
Scorpion Vision Ltd's role in this has always been about more than distribution. UK customers working with us typically start with a conversation about the actual application (the environment, the throughput, the compliance requirements) rather than a fixed specification handed over in isolation. That means selecting the right sensor, housing and enclosure for the environment in question, building and tuning the Python components the task actually requires, and fitting the resulting system around an existing line rather than asking a customer to redesign around it. The same underlying stack that inspects a washdown bottling line one month can be reconfigured for high-speed parcel sortation the next, because the flexibility is designed into the platform from the sensor upward, and our engineering support continues well past initial commissioning.
Seeing it in Stuttgart, or closer to home
Scorpion Vision AI has been running in live installations since Easter 2026, and VISION 2026 (6-8 October) is where it will be shown properly for the first time, running alongside the new Scorpion T-Rex camera on the Tordivel stand. UK manufacturers and integrators who cannot make it to Stuttgart do not need to wait until the next trade show to find out what this means for their own lines. Scorpion Vision Ltd's engineers are available to talk through specific applications directly, and for teams currently modernising an existing Scorpion Vision installation or specifying a new inspection cell, that conversation is often the more useful starting point than a stand visit in any case.