A mine supervisor can now just ask out loud whether the excavator hit its shift target

MaxMine launched MAXI, a voice-based AI assistant for fleet management, tested at Australian coal mines — and the result changed how supervisors track the shift in the field.

MaxMine, an Australian mining technology company, launched MAXI, an artificial intelligence assistant built for fleet management systems (FMS). The tool lets supervisors, operators, mine managers and dispatchers ask natural-language questions about the shift in progress — by voice or text, straight from a phone or desktop — instead of waiting for consolidated reports after the shift ends. The technology has already been tested at two coal sites operated by Batchfire Resources.

What MAXI solves isn't a lack of data — miners already collect a huge volume of fleet information. The problem is the time between the data existing and someone in the field being able to use it to act. MAXI shortens that gap: a supervisor can ask out loud, without taking off a glove, how many times a water truck passed a refueling point during the shift, or whether an excavator is hitting its dig target — and get an immediate answer, without opening a spreadsheet or waiting for the end-of-day report. Technically, the assistant runs on a language model hosted inside MaxMine's own data center, accessing the mine's operational data through a proprietary context protocol — an architecture choice that keeps sensitive operational information out of third-party cloud services, a point that often blocks generative AI adoption in critical industrial environments. According to Stuart Schmidt, ESG and Development manager at Batchfire Resources, the clearest gain in initial testing was in supervisors' routines: the tool helped identify performance bottlenecks and equipment failures earlier, and made short-interval control — the practice of checking and correcting production deviations every few hours, instead of only at shift close — easier to sustain in practice, not just on paper.

Short-interval control is a widely recommended practice in mining operations management, but historically hard to sustain because it depends on supervisors stopping their field routine to consult management systems built for the office. A voice interface, paired with an assistant that already understands the mine's operational context, attacks exactly that friction. For managers evaluating generative AI tools, the case is also a practical example of how to balance access to language models with protecting sensitive operational data — internal hosting, rather than relying on an external provider, being the path chosen here.