Blog

<figure id="figure-5m2thean6mftsqatm"><img id="5m2thean6mftsqatm" src="https://cdn.durable.co/blocks/21jsej0gTgeQ6plkuIlEYUR5LXoSrKVHBf5jlgyHHs8t44SB0ABK8SVZi6XHYNzo.png" alt="" title="" class="object-cover bg-gray-100 rounded-2xl md:rounded-3xl lg:rounded-4xl" style="aspect-ratio: 16/9; object-position: center center;" data-corners="default" data-aspect="16:9" data-caption="" data-positionx="undefined" data-positiony="undefined" data-fullwidth="false" data-mce-src="https://cdn.durable.co/blocks/21jsej0gTgeQ6plkuIlEYUR5LXoSrKVHBf5jlgyHHs8t44SB0ABK8SVZi6XHYNzo.png" data-mce-style="aspect-ratio: 16/9; object-position: center center;"></figure><h3>TL;DR;</h3><p>For the second consecutive year, Aignosi had the privilege of participating in<a href="https://www.abmbrasil.com.br/evento/abm-week-9" target="_blank" rel="noopener" data-mce-href="https://www.abmbrasil.com.br/evento/abm-week-9"> ABM Week 2025</a>&nbsp;the largest technical event for the steel and mining value chain in Brazil.</p><p>This edition was particularly meaningful: together with our partner <a data-mce-href="https://www.samarco.com" href="https://www.samarco.com" target="_blank" rel="noopener" data-mce-selected="inline-boundary"><strong>Samarco</strong></a>, one of the biggest mining companies in Brazil, we presented a peer-reviewed technical paper at the conference — a milestone that highlights how Artificial Intelligence is creating measurable impact in complex operational environments.</p><h3>From Concept to Production: Virtual Sensors in Mining Operations<br></h3><p>The paper, “<strong>Scaling AI and Machine Learning in Samarco’s OT</strong>”, describes how two virtual sensors, developed and deployed in Samarco’s iron ore operations, are driving operational efficiency:</p><p>Real-time mass flow estimation on conveyor belts, eliminating the need for additional instrumentation.<br>Silica content prediction in rougher flotation, directly feeding advanced controllers to optimize recovery.<br></p><blockquote><p>These models are not experimental. They are in production today, automatically retrained, with end-to-end observability and running fully on-premise to meet OT cybersecurity and latency requirements.</p></blockquote><h3>Scaling AI in OT: A New Benchmark<br></h3><p>The deployment of these virtual sensors was made possible through our <strong>SIENTIA™ Industrial AIOps platform</strong>, which enabled:</p><ul><li>15 models deployed in less than 3 months</li><li>A governance framework ensuring security and reliability</li><li>A scalable foundation for future AI use cases</li></ul><p>Looking ahead, <strong>Samarco plans to scale to around 50 models by the end of 2025</strong>, consolidating one of the largest industrial AI initiatives in OT environments in Brazil.</p><p>This demonstrates that scaling AI in heavy industry is not just aspirational — it is achievable, when rooted in operational reality.</p><h3>Key Insight: Scaling Starts with What Works<br></h3><blockquote><p>A critical lesson from this journey: scaling AI in industry doesn’t start with deep learning. It starts with virtual sensors that work — and work well.</p></blockquote><p><strong>When organizations deploy 15 models in less than 90 days </strong>— and project 50 by year’s end — it signals more than technological advancement. It shows real impact in industrial operations: fewer unplanned stoppages, higher process stability, and improved productivity.</p><h3>Gratitude and Collaboration<br></h3><p>We thank Samarco for leading this work, especially Jeancarlo, whose vision and leadership were instrumental throughout this transformation.</p><p>It was also an honor to share this achievement with professionals and companies shaping the future of the industry, including colleagues from Vale, Stefanini IHM, Usiminas, and all who contributed to the technical discussions at ABM Week.</p><h3>&nbsp;Conclusion<br></h3><p><strong>Industrial AI is no longer a distant promise</strong>. As Samarco’s journey shows, it is scalable, sustainable, and operationally impactful — even in environments with the strict constraints of OT.</p><p>The path forward is ambitious, but the foundation is clear: governance, scalability, and trust in AI systems that deliver measurable value.</p>

Scaling Industrial AI in Samarco's OT: Lessons from Samarco's experience at ABM Week 2025

<figure id="figure-1ynjcmr7cqmftwa0sz"><img id="1ynjcmr7cqmftwa0sz" src="https://cdn.durable.co/blocks/14YH3Bmwf5y4DUDjJ3MeSlECCB1i6dUazyeyJXuzZOKzMzkrFX6UrIRtgTeZlqFL.jpg" alt="" title="" class="object-cover bg-gray-100 rounded-2xl md:rounded-3xl lg:rounded-4xl" style="aspect-ratio: 16/9; object-position: center center;" data-corners="default" data-aspect="16:9" data-caption="Valmet's Directors and Aignosi's co-founders" data-positionx="undefined" data-positiony="undefined" data-fullwidth="false" data-mce-src="https://cdn.durable.co/blocks/14YH3Bmwf5y4DUDjJ3MeSlECCB1i6dUazyeyJXuzZOKzMzkrFX6UrIRtgTeZlqFL.jpg" data-mce-style="aspect-ratio: 16/9; object-position: center center;"><figcaption class="body-normal text-gray-500 mt-2">Valmet's Directors and Aignosi's co-founders</figcaption></figure><h3>TL;DR;</h3><p>We are proud to announce a new milestone for Aignosi and the Brazilian industrial sector: our partnership with Valmet, one of the world’s leading providers of industrial solutions and equipment.</p><h3>Who Is Valmet?<br></h3><p>Headquartered in Finland, <a data-mce-href="https://www.valmet.com/pt/" href="https://www.valmet.com/pt/" target="_blank" rel="noopener">Valmet </a>is a global reference in process automation systems for multiple industries, with a strong focus on pulp &amp; paper and energy generation. Its portfolio includes:</p><ul><li>Production equipment and machinery</li><li>Automation systems (DCS)</li><li>Information management solutions</li><li>Advanced control platforms</li><li>Lifecycle services</li></ul><p>This breadth of expertise consolidates <strong>Valmet as a global leader in the industrial sector</strong>, trusted worldwide for its engineering excellence and innovation.</p><h3>PlantTriage: World-Leading Control Loop Optimization<br></h3><p>Through this agreement, Aignosi becomes the exclusive <strong>Value Added Reseller (VAR) in Brazil for <a data-mce-href="https://www.valmet.com/automation/services/process-performance/pid-tuning-process-control/planttriage/" href="https://www.valmet.com/automation/services/process-performance/pid-tuning-process-control/planttriage/" target="_blank" rel="noopener" data-mce-selected="inline-boundary">PlantTriage™</a></strong>, the world’s leading platform for control loop optimization and performance monitoring.</p><p>The initial target market for this agreement is mining, but the potential extends to oil &amp; gas, metals, chemicals, and petrochemicals, which will be addressed in due course.</p><h3>Why This Matters<br></h3><blockquote><p>The combination of PlantTriage™ and our SIENTIA™ AI Ops platform unlocks a new level of efficiency, reliability, and sustainability for Brazilian industry. Together, they enable:</p></blockquote><ul><li>Interoperable data and processes across IT and OT</li><li>Virtual sensors and advanced anomaly detection in real time</li><li>Intelligent alignment with control loop performance</li><li>Integration with APC (Advanced Process Control) and regulatory control</li><li>Greater productivity, reliability, and sustainability on the plant floor</li></ul><h3>More Than a Partnership: A Transformation<br></h3><p>This partnership is more than a milestone for Aignosi. It is an opportunity to transform the Brazilian industrial landscape, connecting the best of process control engineering with the power of Artificial Intelligence.</p><p>We would like to recognize André Luiz Kakehasi, Geovane de Souza, and Cristian Moraes, MBA, whose leadership and commitment were instrumental in making this strategic initiative possible.</p><h3>Looking Ahead<br></h3><p>The future of industry is collaborative, intelligent, and data-driven. Together with Valmet, we are proud to bring this vision to life — setting a new standard for industrial performance in Brazil.</p>

Aignosi and Valmet: A Strategic Partnership

Chat with AI or Artificial Intelligence. Young businessman chatting with a smart AI or artificial intelligence using an artificial intelligence chatbot developed by OpenAI.

<h3>TL;DR</h3><p>When most people think about the complexity of Artificial Intelligence in industry, they immediately picture deep neural networks, advanced architectures, or models with dozens of layers.</p><p>But the real challenge often starts much earlier — with something as “simple” as a <strong>linear regression</strong>. Let's dig in.</p><h3>The Hidden Complexity of Simplicity<br></h3><p>Imagine a model with just two parameters, running in real time, every second, 24/7 on the plant floor.</p><p>That’s:</p><ul><li>1 model</li><li>Running every 1 second</li><li>For 30 days → 2,592,000 inferences per month</li></ul><p>And this is before you even validate whether the prediction can be trusted.</p><h3>Validating Reality: Millions of Operations<br></h3><p>Every inference should be checked against at least five basic validations:</p><p>✔️ Is the input data present?<br>✔️ Is it within the expected scale?<br>✔️ Is the feature within range?<br>✔️ Is there an outlier?<br>✔️ Is the model still calibrated?</p><blockquote><p><strong>Even with this conservative set of 5 checks</strong>, that single regression means 13+ million operations per month — just to ensure reliability.</p></blockquote><p>Now scale this up:</p><ul><li>Medium-sized industrial company → ~30 active models = ~400 million operations/month</li><li><span style="font-family: inherit; font-size: calc(18px * var(--body-fontSize, 1)); font-style: var(--body-fontStyle, normal);" data-mce-style="font-family: inherit; font-size: calc(18px * var(--body-fontSize, 1)); font-style: var(--body-fontStyle, normal);">Large industrial company → ~150 models = ~1.9 billion operations/month</span></li></ul><p>And this is with linear regression, one of the simplest ML techniques available.</p><h3>Why Talk About Linear Regression?<br></h3><p>Because despite its simplicity, linear regression remains one of the most powerful tools in industry.We’ve seen high-impact use cases such as:</p><ul><li>Virtual sensors with just one or two input variables</li><li>Simple and robust anomaly detection</li><li>Estimations of hard-to-measure variables in real time (e.g., NOx, CO, O₂)</li></ul><p>It works — reliably and efficiently — especially in industries such as mining, chemicals, and pulp &amp; paper.</p><p><strong>But “simple” doesn’t mean “trivial.”</strong> </p><blockquote><p>In OT, even a basic regression requires robust infrastructure, automation, and observability to remain trustworthy at scale.</p></blockquote><h3>The Real Lesson<br></h3><p>Never underestimate the power of linear regression — nor the complexity of keeping it alive in production.</p><p>The question every industrial organization must ask is:</p><p>👉 Is your data architecture and MLOps stack ready for this?<br>👉 Can you guarantee performance, reliability, and security at this level of scale?</p><h3>Final Thoughts<br></h3><p>At Aignosi, this is the daily reality we face with our industrial clients. There’s no glamour in it — but making AI work in OT means ensuring that every inference, every second, is dependable.</p><p>And that requires more than algorithms. It requires industrial-grade AI infrastructure.</p>

Why Keeping a Simple Linear Regression Alive in OT Is Harder Than You Think

<blockquote><p><strong>The story of AI/ML at Samarco didn’t begin with a massive data science team or an ambitious enterprise-wide program</strong>. </p></blockquote><p>It began in 2022, when a handful of automation engineers developed a simple virtual sensor to estimate mass flow on a conveyor belt.</p><p>That model worked. It helped operators detect anomalies earlier, avoided blockages in crushers and screens, and kept the process flowing. </p><p>What seemed like a small step revealed a huge opportunity: if one conveyor could benefit from AI/ML, why not 30?</p><p><strong>But scaling wasn’t just about copying code</strong>. It required a holistic approach around People, Processes, and Technology.</p><h3>People: Building Internal Capability<br>&nbsp;</h3><p>One of the most remarkable aspects of this journey is that Samarco itself developed 13 of the 15 models currently in production.</p><p>This was only possible because the company created a dedicated Process Optimization team inside OT. Multidisciplinary by design, the team combined automation, process engineering, and data science skills. With Aignosi’s support, Samarco conducted an analytics maturity diagnosis to identify gaps, set priorities, and build internal skills.</p><p>This organizational move was crucial: <strong>instead of outsourcing model development, Samarco empowered its own people to own the models</strong> — designing, deploying, and maintaining them. AI became part of the operational DNA, not a black-box service.</p><h3>Processes: Governance and Repeatability<br>&nbsp;</h3><p><strong>The next step was to standardize. Without clear governance, models risk becoming isolated “one-off experiments</strong>.” Samarco adopted AIOps for OT, incorporating best practices from MLOps, DataOps, and DevOps:</p><ul><li>Version control for models and datasets</li><li>Automated retraining pipelines</li><li>Continuous monitoring for drift and anomalies</li><li>End-to-end observability of the models in production.</li><li>Standardized workflows and templates for rapid replication<p><br></p><p>This turned model deployment into a repeatable, reliable process. What once took months could now be done in days.</p>&nbsp;</li></ul><p>These processes turned AI from a one-time success into a repeatable system, cutting deployment lead-time from months to days.</p><h3>Technology: SIENTIA™ as the Enabler<br>&nbsp;</h3><p>The backbone of this transformation is SIENTIA™, an on-premise AIOps platform designed for OT environments. Unlike generic IT tools, it was built to meet industrial realities:</p><ul><li>Native OPC UA connectors for integration with SCADA, PLCs, and PIMS</li><li>100% on-premise deployment, addressing cybersecurity and latency requirements</li><li>Real-time inferencing, ensuring AI insights are available when operations need them</li><li>No-code templates, allowing engineers and operators to configure models without Python expertise<br>&nbsp;</li></ul><p>This meant that Samarco’s own team could focus on process knowledge, while SIENTIA™ took care of governance, automation, and observability.</p><h3>&nbsp;The Result<br></h3><blockquote><p>In less than 90 days, Samarco went from 1 model to 15 models in production — and 13 of those were developed internally by its own Process Optimization team.</p></blockquote><p><strong>This is more than scale. It’s organizational autonomy</strong>, where AI is not just a project but a capability owned by the operation itself.</p><p>And this is only the beginning. The roadmap now points to 50 models by the end of 2025, positioning Samarco as one of the leaders in applying Industrial AI to mining operations in Brazil.</p><p>We will bring more stories about how Samarco is continuing to scale their AI/ML in the shop floor.</p>

AI/ML Scalability at Samarco's OT

<blockquote><p>Conveyor belts are the arteries of a mining operation. </p></blockquote><p><strong>Yet, unplanned stoppages and failures in belt scales are costly and frequent</strong>. <br></p><p>Samarco solved this challenge not with deep learning or complex architectures, but with something deceptively simple: <strong>a linear regression model.</strong></p><p>This virtual balance sensor estimates conveyor mass flow in real time, delivering the same function as a physical belt scale — but with higher reliability and lower maintenance.</p><h3>Simplicity That Works<br>&nbsp;</h3><p>While often overlooked, linear regression remains one of the most powerful tools in industrial AI/ML. In Samarco’s case, it proved accurate enough to feed APCs directly, with retraining needed only once every 4 months.</p><p>Its success shows an important lesson: </p><blockquote><p>AI in OT doesn’t start with deep learning — it starts with models that work and keep working.</p></blockquote><h3>The Auto-ML Template: A Samarco Idea Scaled Through SIENTIA™<br>&nbsp;</h3><p>One of the key breakthroughs came not from the algorithm itself, but from how the model was operationalized.</p><p><strong>The idea of creating a no-code template for conveyor models originated within Samarco’s own team</strong>. By incorporating this into SIENTIA™, the approach became part of the platform’s DNA:</p><ul><li>Drag-and-drop simplicity, so models could be built without coding.</li><li>Standardized workflows, ensuring governance and observability by default.</li><li>Accessibility for all skill levels: from senior process engineers to plant operators, anyone could configure and replicate the model in minutes instead of months &nbsp;</li></ul><h3>&nbsp;Impact on Operations<br>&nbsp;</h3><ul><li>Reduced unplanned stoppages from conveyor overloads.</li><li>Reliable mass flow estimation, even when physical scales fail.</li><li>Scalability across 30+ conveyors, multiplying the impact.</li><li>Cultural empowerment, as more people across the organization can now build and deploy ML models safely.<br>&nbsp;</li></ul>

Soft sensors for conveyor belts at Samarco

<h3>TL;DR;</h3><p>Using the same template created for <a data-mce-href="https://www.aignosi.com.br/blog/soft-sensors-for-conveyor-belts-at-samarco" href="https://www.aignosi.com.br/blog/soft-sensors-for-conveyor-belts-at-samarco" target="_blank" rel="noopener">virtual belt scales</a>, Samarco built a silica prediction model for flotation. With SIENTIA™’s automated retraining, even a simple regression delivers reliable, real-time insights without costly custom development.</p><h3>The case</h3><p>In iron ore flotation, silica content is a critical quality parameter. Traditionally, plants rely on laboratory analyses every two hours, leaving Advanced Process Controllers (APCs) operating with delayed data. <strong>This slows decision-making, increases reagent waste, and risks product quality.</strong></p><p><strong>Samarco addressed this challenge with a virtual quality sensor, designed to predict silica grade every 5 minutes</strong>. </p><p>Surprisingly, the model wasn’t built on complex deep learning, but on a multivariate linear regression.</p><h3>&nbsp;Reusing What Already Worked<br>&nbsp;</h3><p>Instead of reinventing the wheel, <a data-mce-href="https://www.aignosi.com.br/blog/soft-sensors-for-conveyor-belts-at-samarco" href="https://www.aignosi.com.br/blog/soft-sensors-for-conveyor-belts-at-samarco" target="_blank" rel="noopener">Samarco reused the same no-code template originally created for conveyor belt mass flow models</a>. This was possible thanks to the flexibility of SIENTIA™, which allowed the Process Optimization team to adapt an existing template to a new use case.</p><p>But flotation is a nonlinear process, and a simple regression alone wouldn’t be robust enough over time. <strong>This is where another key feature of SIENTIA™ came into play.</strong></p><h3>Automated Retraining: The Game-Changer<br>&nbsp;</h3><p>SIENTIA™ provided automated retraining every 4 hours, which compensated for process variability and data drift. This capability transformed what could have been an unstable model into a reliable, production-ready sensor.</p><p>The combination of a simple template + automated retraining meant that Samarco could:</p><ul><li>Avoid exorbitant costs of custom model development.</li><li>Leverage its own team’s process knowledge.</li><li>Deploy a working solution much faster than traditional R&amp;D cycles.<br>&nbsp;</li></ul><h3>Impact on Operations<br></h3><ul><li>Silica predictions delivered every 5 minutes instead of waiting 2 hours.</li><li>Direct integration with the flotation APC, enabling real-time reagent dosing.</li><li>Significant reduction in reagent waste and higher product quality stability.<br data-mce-bogus="1"></li></ul><blockquote><p>This case is a good example that even “simple” models, when combined with robust MLOps features, can solve complex nonlinear problems.<br>&nbsp;</p></blockquote>

Blog

Scaling Industrial AI in Samarco's OT: Lessons from Samar...

Aignosi and Valmet: A Strategic Partnership

Why Keeping a Simple Linear Regression Alive in OT Is Har...

AI/ML Scalability at Samarco's OT

Soft sensors for conveyor belts at Samarco

Soft Sensors for quality prediction at Samarco