By the age of eight, Conor Temple was already showing an interest in engineering. His grandfather had taught him how to use tools, and on one occasion his mother came home to find that he had cut his bike in half to make a pedal-powered paddlewheel boat. Temple went on to study mechanical engineering at university and worked in hands-on engineering roles before moving into automation equipment sales. It was there he spotted the gap in the market for his Athlone-based start-up, Rheba Intelligence Systems. While he was training two project managers on a €200,000 machine, he saw how difficult it was to quickly diagnose faults. “The machine was six weeks late being commissioned and they were under serious pressure. Then a fault flashed up on the screen, and both of them looked at me for an answer,” says Temple. “The answer to the problem was in the manual but the manual had several thousand pages, so I tried ChatGPT. “The answer came back instantly and looked completely plausible. But something felt off. “I went through the manual and found the ‘real’ answer. I made an adjustment and in two minutes the machine was running. “Later, I looked at the ChatGPT answer properly. Had I followed its advice, it would have destroyed a component with a six-week lead time and a €30,000 price tag.”“The answer wasn’t wrong in an obvious way,” Temple says. “It was wrong in a confident, well-presented, entirely believable way. That’s the specific kind of ‘wrong’ that gets expensive equipment damaged and careers shortened.” Temple’s has built Rheba to solve that problem. The company is aiming to transform industrial maintenance by providing engineers with a fast, accurate way to get the information they need to solve machinery glitches. Rheba takes a company’s suite of manuals and centralises them with a comprehensive indexing feature that finds all the related information quickly. “In a typical company there are manuals for everything but they’re on USB sticks, in folders, in files and on desktops and no one knows exactly where everything is,” says Temple, who adds that with 20 per cent of the engineering workforce due to retire within the next five years, a lot of what they intuitively ‘know’ about where to find things will be lost. Unlike generic AI tools, Rheba displays diagrams and schematics directly from source manuals.“Rheba doesn’t replace a company’s documentation. It ingests the plant’s manuals, drawings and procedures and makes answers findable, in seconds, by anyone, on any device, with the source page shown so an engineer can trust and verify the answer and can query the system in plain English,” says Temple. “Some AI tools give a page number as a reference, but they don’t show the page, and if an engineer has to leave the machine to find a manual to verify the AI’s answer, then it defeats the purpose. “Industrial manuals are full of diagrams and schematics, so text alone doesn’t cut it. Rheba shows the actual manual page with the diagrams right there in the interface. Developing the system to render those source pages inline was a challenge, but it’s what makes the tool safe to use on a factory floor and will result in a dramatic improvement in downtime.“Maintenance management systems have the relevant documentation but treat it as something to be stored, not knowledge to be queried.”Temple has built the company’s system from scratch with virtually no financial outlay, but an unquantifiable amount of founder time and mentor support from the Westmeath Local Enterprise Office and business development guidance from investor and businessman David Slowey.Rheba was officially launched last December, but Temple had been working on building the platform for months, and it wasn’t all plain sailing. “I had a few attempts that I didn’t think were good enough. For example, I wasn’t happy with how the platform was scaling, so I rebuilt it from the ground up to handle vast amounts of data,” he says. “It’s disheartening when this happens, but I’d let the problem go around in my head for a few days and eventually I’d figure out a solution.” The Rheba system is currently in live pilot with a logistics company in Ireland and a manufacturing company in the United States, with five potential customers waiting in the wings. The company will ultimately use a SaaS revenue model, but as of now it is charging on a case-by-case basis based on the number of users and frequency of use. The platform will work for any industry and in multiple languages, and Temple expects to be employing 10 people within three years. “The route to building my business was deliberate,” he says. “I figured the best way to find an opportunity nobody else had spotted was to go deep into a specialised industry, get close to the people doing the actual work, and pay attention. The deeper into a specialised area you go, the more likely you are to see something that hasn’t been capitalised on yet. “I liked sales because it was the best place I’d found to test unconventional things,” Temple adds. “So, if your competitors are all sending the same AI automated mailshot, then a handwritten letter is more likely to be what gets opened. “I also learned a lot from standing outside in the smoking area, as that’s where real conversations happen. I’d get chatting to people and inside 15 or 20 minutes I usually had the name, email and sometimes the direct number of exactly the person I needed to speak to.”