Portuguese 
English 
Spanish 
4 min of reading 
0 comments 
Tractian, a startup founded in 2019 by three Brazilian engineers, became a unicorn by creating sensors with artificial intelligence that anticipate failures in equipment. It currently monitors 250 thousand machines live in factories of P&G, Unilever, Nissan, and Danone. The Manual of the World, Iberê Thenório’s channel, visited the company’s AI Center in São Paulo for the Bora ver segment.
In a factory, replacing parts too early becomes waste. Waiting to replace them after a breakdown leads to losses. A Brazilian startup solved this million-dollar dilemma of the industry using smart sensors and artificial intelligence, and the result placed it in the club of billion-dollar companies in the country. Tractian, founded in 2019 by engineers Igor Marinelli, Gabriel Lameirinhas, and Leonardo Vieira, raised R$ 700 million in its Series C round in November 2024, reached a valuation of R$ 4 billion, and today monitors over 250 thousand machines live in 1,000 industrial plants worldwide.
The report from Manual do Mundo, Iberê Thenório’s channel, visited the company’s Artificial Intelligence Center in São Paulo for the Bora ver segment. It is a 10,000 square meter space where Tractian intentionally breaks machines to generate data. Each bearing that bursts, each shaft that breaks, each belt that snaps feeds an artificial intelligence trained to recognize the first signs of failure before the problem reaches the clients’ factory floor.
How does a sensor predict a machine failure?
The Tractian sensor measures four variables simultaneously: vibration, ultrasound, temperature, and magnetic field. The big breakthrough lies in ultrasound, sound of such high frequency that the human ear cannot detect it.
-
Urea reaches R$ 272 per sack on April 14, 2026, already rising nearly 75% at the ports and could soar even higher if the conflict in the Middle East escalates by May.
-
Labor crisis in Japan leads the country to a record of 2.57 million foreign workers, but 85% still pay agencies to find jobs and face fees of up to 6,000 euros.
-
China suspends meat imports from a slaughterhouse in Várzea Grande, and the market tries to panic: an analyst says that the ban is common, may affect only one batch, and does not lower cattle prices.
-
With 93 companies selling armored vehicles, aircraft, and monitoring systems to 148 countries, Brazil’s defense industry is doubling exports and is on track to break all records in 2026.
When the vibration of a machine changes noticeably, the problem is usually already advanced. Ultrasound detects the problem earlier, while a simple solution like tightening a screw or enhancing lubrication is still possible.
At the AI Center, the team runs tests with eight bearings mounted in different ways. In a few minutes, the thermal camera shows the defective bearing reaching 97°C. Soon smoke begins to emerge. In a real factory, this would cause a fire.
But the sensor had already detected the failure long before. 80% of failures in industrial equipment occur due to lack of lubrication, and this is precisely the type of anomaly that ultrasound identifies at an early stage.
The sensor still undergoes brutal testing before going to market. It is exposed to water at 80°C, saline atmosphere simulating years of salt spray, and agitation at 85°C with 97% humidity. It operates in temperatures from minus 40°C to 120°C positive, is sealed against water, dust, and chemicals.
How much money does predictive maintenance really save?
The numbers from Tractian explain the billion-dollar valuation. Clients report a reduction of up to 40% in downtime, an average increase of 15% in productivity, and an average return on investment of 383%, with payback in three months.
The company’s NRR, a metric that measures revenue expansion within already acquired clients, is 196%. The gross margin is 72%, extremely high for a company that combines hardware and software.
Revenue doubles year over year and closed 2024 between R$ 230 million and R$ 250 million. Series C was led by Sapphire Ventures, an American manager whose largest investor is SAP, with participation from General Catalyst, Next47, and NGP Capital.
Tractian’s total funding reached R$ 1.05 billion in four rounds, and the company was the only Brazilian on Forbes’ global list of the best AI startups.
The data collected by more than 100,000 installed sensors is processed in infrastructure built with NVIDIA H100, H200, T4, and L40 GPUs, in the CUDA ecosystem hosted on Oracle Cloud. The artificial intelligence compares each reading with millions of others taken over the years, discovers patterns that human engineers have never detected, and prescribes corrective action.
Why does this matter for the Brazilian industry?
Tractian operates in strategic sectors such as mining, food, automotive, agribusiness, and chemicals. The multinational Ingredion, a global giant in food ingredients present in over 120 countries, adopted predictive maintenance to address failures in plants with critical assets that are difficult to access. P&G, Unilever, Nissan Brazil, and Danone are also part of the portfolio.
The story of the founders explains part of the success. Igor Marinelli, Gabriel Lameirinhas, and Leonardo Vieira grew up seeing their parents and siblings work as maintenance engineers. The problem was real, familiar, and known.
Marinelli graduated in engineering from Berkeley. The company is based in Atlanta but has a Brazilian soul, and today exports industrial maintenance technology from Brazil to the world. Finep, the federal agency for innovation funding, provided resources via Inovacred in the early stages.
When a production line stops, the loss does not only appear on the factory’s balance sheet. It shows up on the empty supermarket shelf, in rising prices, and in delayed cars at the dealership. The invisible work of maintenance is what keeps the economy standing. Tractian transformed this into data, data into AI, and AI into billions.
Have you ever had contact with predictive maintenance at work or know of any factory that uses this type of technology? Share your experience in the comments; we want to hear from those on the factory floor.
Seja o primeiro a reagir!