AI Use Increases Mineral Processing Production — Interview with BrainiAll CEO
Fabio Suizu, founder and CEO of BrainiAll and former Microsoft executive, explains how artificial intelligence is transforming ore processing — starting with a verified +6% productivity gain in a Brazilian gold mine.
Published in Brasil Mineral Magazine
This interview with Fabio Suizu was originally published in Brasil Mineral Magazine, Issue 441 (July 2024), in the feature "As Maiores Empresas do Setor Mineral". The 2-page spread below is the original article layout as it ran in the magazine.
Why mining, why now
Using artificial intelligence to optimize industrial processes is the focus of BrainiAll, a company initially focused on the mining segment, where any productivity gain translates into significant financial results. According to founder and CEO Fabio Suizu — a former Microsoft executive with a strong leadership background in data technology, AI, and industrial processes — AI in ore processing can generate very significant productivity gains.
He cites the case of a gold mining company where the application of BrainiAll technology enabled a 6% increase in grinding production rates using the same equipment. Using data through AI to generate value is BrainiAll's core expertise.
The industry context
The decision to focus on mining came from contacts with mining companies while Suizu was still at Microsoft, where they faced the challenge of dealing with increasingly lower-grade and harder ores. "The easy, high-grade, and abundant ore has already been mined over the last 200 years," he says. "Ore processing today is increasingly challenging, especially in grinding, because it is very difficult to grind hard, low-grade ore while maintaining process quality."
Mining companies are already well-served in terms of equipment and have good levels of automation and data generation. What they need is greater agility in analyzing process parameter data — and this is where AI helps.
The data paradox
"The data is there, but they don't use it. There are operators and engineers sitting there, looking at screens, using outdated software. They generate billions — or tens, or hundreds of billions — of data points in a year, but hardly use any of it because it's a human being looking at the data, observing what's happening, with thousands of data points coming in every second."
Typically, the operator looks at one or two data points and makes an adjustment, a machine setpoint. And sometimes they rely on super-simple software supporting those people. BrainiAll's approach is different: the AI reads everything, finds where the problems and failures are, and corrects everything automatically, adapting to situations without human intervention.
Results at the gold mine
In the case of the gold mining company cited, production increased by 6% without a drop in particle size quality and using 3% less energy. A positive aspect of BrainiAll's operation is that in just two weeks the software is installed and running.
Additionally, based on data analysis — in the gold mining company, two years of operational data were analyzed — the system can predict production performance degradation and make micro-adjustments, for example in mill pressure, power, and speed, to maximize production and reduce equipment wear. The process is stabilized, variance drops, and the equipment is much less stressed.
+6% production · 3% less energy · 25-day controlled pilot · Brasil Mineral #441 (Jul 2024)
The future
"Within a few hours, AI starts adjusting, and the equipment begins to operate in cruise mode, generating the production increase. In the cited example, the increase was 6%, but we plan to soon exceed 10%. The future is using AI to analyze data point by data point, meticulously, making adjustments across millions of data points and performing fine-tuning."
"Humans weren't made to think about thousands of variables in seconds and make decisions like increasing a SAG mill's power by 3% or accelerating a ball mill's speed by 2%. But computers and AI were made for this."