AI is forever in the news with many commentators predicting an impact as disruptive as mechanisation during the industrial revolution. Some views are downright dystopian with a focus on US tech companies and their ‘obsolescence’ projects (an aim to completely replace staff with robots). It was therefore refreshing to catch up with Doug Brion, CEO of technology company Matta, who expressed his respect for the manufacturing sector and the skills that its workers have honed over decades
A University of Cambridge spin out, Matta is currently London based with plans to expand into Germany and the US in 2027. The company has grown rapidly since its inception a couple of years ago, with its current team of 21 up from 10 at the beginning of the year. “We’re basically in full scale-up mode,” Doug explained.
UK SMEs don’t typically find funding easy to secure, but Matta has faced no problem in this respect. The company raised seed funding of US$14m in December last year; has several European investors; funding from Silicon Valley; and money from business interests including Jaguar Land Rover and the Peugeot family.
Wide range of clients
The company’s clients are varied and colourful, including but not limited to a chocolate factory, wellie makers, and fighter jet engineers. Matta works with SMEs and global enterprises alike offering solutions of varying complexity, even offering a cloud-based solution for companies that don’t want on-premises AI servers.
What services does Matta provide
Matta’s services include the provision of industrial AI models that understand manufacturing, and a central platform designed to give factories real-time self-awareness or ‘factory sentience’. Practically, this means that after working out what a client is looking for, Matta will place sensors and cameras around a factory floor to watch the production line, then feed data regarding products to a central platform. This platform proceeds to analyse the information to find faulty produce or batches and ascertain where the fault originated, or to understand what is causing a bottleneck.
By watching parts on a production line these sensor cameras quickly learn what good looks like. They can also measure, count, and read barcodes. As Doug explained: “If there is a scratch on a part, the cameras will know the serial number and it can easily be retrieved.”
The complexity of Matta’s solution will depend on the client’s size and requirements, but every installation helps speed up processes and reduce downtime. Doug says: “Managers can easily get a snapshot of their throughput using this technology. It quickly becomes clear where there are bottlenecks.”
Factory managers will also be able to liaise with a chat-bot interface equipped to provide almost immediate answers to business-critical questions such as: “how does this quarter compare with the last quarter?”, or “what are the biggest issues faced by a particular production line?”.
How does the system learn?
Doug explained that the system learns by watching. “It’s like showing a toddler something it hasn’t seen before. We show the sensors what to look at and they learn patterns over time. If a toddler is shown 10 examples of well painted mechanical parts and there’s one with a scratch, it might say, “Oh, that’s wrong” without having seen it before.
Where do the people come in?
The system is a tool like a microscope or hammer and is not designed to replace staff currently employed, Doug explained: “I get pretty angry with all the AI tech bros saying, ‘we want humanoid robots everywhere’. To be honest, I think it comes from them never really visiting factories. They don’t understand the amount of skill involved in manufacturing, particularly with dextrous jobs like leather stitching, linishing, or maybe even icing a cake where each cake has a different shape and design.” He continued: “There are some jobs in manufacturing where staff have been training for 20 years. They use sight, sound, smell, touch and have done this work for eight hours a day every day. The idea that you’re going to replace these people overnight is crazy.”
What does the future look like for Matta?
Matta has concrete plans to grow and expand its services, with a move into predictive automation and even correction and fixing of faulty parts. Doug explained that the company has some more eccentric plans too: “The team has decided they’d quite like our headquarters to resemble Bletchley Park; a kind of Bletchley part 2.0 if you like. We imagine a stately home surrounded by boffins in sheds, maybe a cricket pitch outside. The Americans will have their huge buildings in San Franscisco and we’ll have our sheds. We’ve given ourselves eight years to make it happen.”
For more information visit: www.matta.ai
