How RHI Magnesita Is Modernizing Planning for High-Pressure Industries

When RHI and Magnesita merged, the newly combined company needed to establish global processes across a complex supply chain.

Excel played an important role in those early years. It gave teams the flexibility to build new processes and helped the business operate successfully as a global organization.

Over time, however, the planning environment became harder to manage. Demand and supply planning were disconnected, forecasts were not being created at the required level of detail, and customer orders were still being promised manually.

Then COVID-19 and a period of sustained market volatility exposed the limits of the existing approach.

“We were not able anymore to be as reactive as we wanted,” said Ricardo Dominguez, Head of Innovation and Product Governance, Supply Chain and Procurement at RHI Magnesita. “And forget about being proactive. It was impossible with the tooling that we had before.”

That realization led to Project Everest, RHI Magnesita’s multi-year supply chain transformation with o9.

Watch

Building a connected planning environment

RHI Magnesita is the global leader in refractory products. These materials are used in industrial processes that require extremely high temperatures, including steel, cement, and glass production.

The company’s customers are operating in an unstable environment shaped by geopolitical conflict, tariffs, and excess supply from China. They need greater stability to plan investments and maintain production. That places further pressure on RHI Magnesita to respond quickly and reliably.

Its previous planning processes made that difficult.

Demand planning took place at a higher level of aggregation than the business needed and was not properly connected to supply planning. Order promising also created significant friction because sales orders were handled manually.

RHI Magnesita began looking for a platform that could connect these activities and support decisions at every level of the organization.

Dominguez said o9’s technology stood out during the selection process. The company could work with a single data model, move between aggregated and detailed views, and retain the context behind each decision.

The way o9 approached the selection process was equally important.

“We didn’t tell them, ‘This is my requirement, this is my requirement,’” he said. “We told them, ‘This is our problem. How do you fix that?’”

According to Dominguez, the o9 team took the time to understand the company’s challenges, ambitions, and current limitations.

“They really got us,” he said. “They understood the problems that we had, what we were trying to achieve, and what our shortcomings were.”

Maintaining detail across planning processes

The Enterprise Knowledge Graph has become one of the most important parts of the platform for RHI Magnesita.

It connects information and planning processes within a shared model. Teams can move from global or regional views into product-level detail without relying on separate handovers between systems.

“The Enterprise Knowledge Graph is the one aspect of o9 that surprises everybody,” Dominguez said. “Everything is connected in the tool. There is not a handover where you are losing information or losing granularity.”

This matters because handovers often create gaps between demand, supply, and execution. Data may be summarized, delayed, or interpreted differently as it moves between teams.

A connected model gives RHI Magnesita a more consistent view of the business and provides a stronger foundation for future AI capabilities.

Dominguez believes data-related investments will remain valuable even as the AI market develops quickly.

“Investment in data is always going to be a good investment”, he said.

Using segmentation to focus human attention

RHI Magnesita is beginning to see how connected planning can change the day-to-day role of planners and sales teams.

In parts of the North American business, the statistical forecasting engine is already producing strong results for high-volume, predictable products.

The company uses ABC-XYZ segmentation to classify products according to their importance and predictability. This helps identify where statistical forecasting can manage routine decisions and where human knowledge remains essential.

“Planners are already perceiving that o9 outperforms them in certain segments,” Dominguez said. “I really think that now we are going to start unlocking the power of o9.”

This is helping the organization move through the acceptance phase of the change. Planners can see where the system performs well and begin to adjust how they spend their time.

The same principle applies to sales teams.

Dominguez gave the example of a salesperson reviewing 100 planning lines. Analysis within o9 may show that only 12 require direct input from the salesperson. The remaining lines can be handled through the statistical process.

“If I only focus on 12 of those lines, that is where my value is,” he said. “I can just let o9 do the rest.”

This allows salespeople to concentrate on the products and customers where their market knowledge can improve the forecast. They no longer need to review every line with the same level of attention.

Dominguez sees this as an important next step for the company’s planning processes.

“I think our sales teams are going to love it because they don’t have to take care of everything,” he said. “They just have to take care, in this case, of 12% of everything.”

Looking beyond traditional change management

For Dominguez, the most difficult part of a transformation like Project Everest is changing the way people work.

Town halls, impact assessments, training sessions, and informal engagement activities all play a role. However, he believes transformation teams also need to address more difficult questions about skills, incentives, culture, and organizational design.

An advanced planning platform may require people to develop new capabilities. It can also change which tasks planners perform and where their judgment creates the most value.

“Are people ready to let go of old ways of working?” Dominguez asked. “Is our culture really ready also to do that?”

He believes organizations should examine the behaviors they reward. Teams may be encouraged to transform processes, while internal incentives continue to favor people who maintain the status quo.

“Are we rewarding the people who are challenging the status quo?” he asked. “Or are we rewarding the people who are just following the status quo?”

Dominguez describes these issues as the “hardcore” side of change management. They are more difficult to address than communications and training, but they often determine whether new ways of working take hold.

“People need to understand what the new capabilities are and sometimes even the new kind of organization you need to have,” he said.

Learning throughout the transformation

The name Project Everest reflects the scale of the journey.

For a large and complex organization, supply chain transformation is likely to take several years. During that time, the company will learn more about its processes, data, people, and technology.

Dominguez advises organizations to use those lessons rather than treating the original business case and timeline as fixed.

“Most companies make a business case to go on a journey like this, and they are very afraid to touch that business case or touch that initial timeline,” he said. “But you are going to learn so much during the journey.”

He also recommends keeping decisions grounded in data. Large transformation programs often involve internal politics, especially when they affect roles, processes, and responsibilities.

“Try to isolate the politics of projects such as this,” he said. “There will be politics involved, but try to stick to the data.”

Preparing for agentic AI

Dominguez is also exploring how agentic AI could change enterprise software and business processes.

His view of the technology changed after he built an AI agent himself. Experiencing what an agent could do made the potential more tangible.

“The moment that people build AI agents and they see what they can do, this is the aha moment,” he said.

He believes companies are moving toward an era in which software can be created for highly specific needs. This could be especially valuable for smaller organizations that lack the resources to commission large software projects for every process.

The speed of development makes it difficult to predict exactly where AI will go. Dominguez noted that his own answer would have been different only a few months earlier.

For that reason, he sees investments in data and connected knowledge as more dependable than bets on individual AI applications.

He also believes companies need to invest in the people who will identify and develop new use cases.

“The biggest gap to really unlock the power of agentic AI is the imagination of the people,” he said.

AI may allow teams to solve familiar problems in completely different ways. Reaching that point will require people to examine processes closely, question existing assumptions, and experiment with approaches that were previously unavailable.

Project Everest is already starting to change how planning decisions are made and where teams focus their attention. The next stage will depend on how successfully the organization turns those early lessons into new ways of working.

A Guide to the o9 Enterprise Knowledge Graph

The o9 Enterprise Knowledge Graph (EKG) is a four-layer, closed-loop system designed to transform how enterprises plan, decide, and execute.

Download now