Honeywell Process Solutions rolls out an AI-enabled digital suite that connects platforms, applications and advanced cybersecurity to make plant floor predictions and recommendations.
When artificial intelligence (AI) showed signs of potential in manufacturing, Chevron asked its automation partner, Honeywell Process Solutions, how it could effectively use the technology in its plant operations. This was not a shortsighted request to layer on a query-based AI assistant, but rather a plea to help fix some existing process-related problems that, when not handled correctly, negatively affect operations.
The call to action required Honeywell executives to figure out the areas AI would have the most impact. To do that, they did a little role-playing.
“We went into the personas to understand what a plant operator needs,” said Pramesh Maheshwari, president and CEO of Honeywell Process Solutions, in an interview with Control Engineering. “What are the challenges a plant operator in the control room is facing which we can solve? What are the challenges a field operator is facing which we can solve?”
At Chevron, the focus was on managing abnormal situations and understanding what happens in the plant before the alarm comes on, as well as what actions an operator should take.
“They provided a lot of insights and we started developing the solution,” Maheshwari said. That solution came in the form of AI-based alarms in the control room, which are currently being tested by Chevron.
But the use of AI doesn’t stop there. Honeywell recently rolled out its AI-enabled digital suite for connected platforms and applications.
Autonomous automation
In June, during the 49th annual Honeywell Users Group (HUG) in San Antonio, Maheshwari unveiled key innovations that developed from the Chevron conversation — as well as many other customer discussions.
Among the announcements:
- A new wave of AI-enabled digital technologies designed to accelerate the shift from automation to autonomy.
- A suite of sophisticated AI-enabled operational technologies (OT) cybersecurity offerings that not only detect and mitigate attacks but thwart them all together.
- An updated Honeywell Digital Prime Ecosystem that allows engineering teams to test projects before implementation to reduce plant downtime and increase throughput in production.
Digitalization and AI are topics that have evolved rapidly. At Honeywell, the focus is on OT and helping plants become more autonomous while keeping people in the loop. “AI should augment, not replace, the worker,” Maheshwari said during his HUG keynote address. “Our goal is to remove the dull, dirty and dangerous from daily operations.”
At Chevron, for example, the Experion Highly Augmented Lookahead Operations (HALO) AI-based alarm guidance helps operators access and resolve alarms quickly by combining dynamic contextual knowledge with a reference to historical responses and correlations across the plant. It even provides corrective actions based on cause and correlations.
In the past, operator response to alarms was based solely on their own experience or that of a senior operator. “If they haven’t seen the alarm in the past, they have to spend time analyzing what led to this alarm,” said Amit Jain, simulation advisor and senior process control engineer at Chevron. “So, the action they take is dependent upon their knowledge and experience. If they are not able to act in time with the right actions, that leads to a safety event. What if we can analyze five to 10 years of data to figure out what actions to take and the outcome [for] the most optimum response for the situation?”
Jain started by using machine learning data, but it was time-based data that is missing context because it doesn’t take into consideration other things happening in the plant. To gather that context, Chevron and Honeywell worked on a knowledge graph of everything in the plant.
The knowledge graph took processes, alarms, events, action relationships and lineage data and contextualized it into process information using standards and policies. It links events and alarms to past operator actions, identifies alarm impacts and provides actional recommendations via natural language.
“Now we have a map of all the data and every alarm and action to do a better association between alarms and actions,” Jain said. “But do we trust it? No. So we put a human in the loop to validate the optimum response library against the operating procedures and assumptions.”
This AI-enabled solution, which was developed in eight months and is currently in use as a pilot project at Chevron, digitizes the knowledge of operators. The expected benefits include enhanced operator response, reduced loss and safety risks, knowledge automation and accelerated operator development. It also opens the door to more predictive analytics associated with autonomous operations.
Emerging digital cognition capabilities
Generative AI has been a tool used to create new content. Agentic AI, on the other hand, is designed to autonomously make decisions. In the future, agentic AI will make predictions and recommendations with control room operators overseeing the system, like a pilot in a self-driving car.
Another way to look at this is through the lens of “digital cognition,” that combines humans and machines to build expertise into the process. “Harness the power of data, analytics and digital technology to provide guidance that helps the least experienced personnel act like the most experience personnel,” said Jason Urso, CTO of Honeywell Industrial Automation.
To make this leap, control systems need to transform, Urso said. Controls run effectively during steady state, but outside of that it is dependent on human expertise to make decisions when presented with alarms.
What’s needed is a new kind of distributed control system (DCS) with digital cognition capabilities that convert data into knowledge.
“What if we could automate decision making in the same way we automated the process control equipment 50 years ago?” Urso asked. “I think that’s really the next steps that we are headed towards.”
It’s moving beyond just the control of the equipment and giving advice and guidance and making that more autonomous. It’s trying to figure out what to make, when to make it, what to fix, when to fix it.
“We’ll see a giant step forward in the next release of Experion and things will stop looking like a traditional control system. It will look like a portal of recommendations and guided work rather than here’s a bunch of numbers, good luck figuring it out,” Urso said.
Honeywell is calling this Experion Cognition, which consists of three elements that allows sites to perform flawlessly during steady state and transient operations.
First, a new intelligent user interface and a new user experience that goes beyond data and alarms and presents recommendations. Second, there is a reasoning engine and analytics that combine the process with digital twins. And third, connectivity beyond the control system to planning, scheduling, maintenance, reliability and process to provide a closed loop response within plant operations.
With cognition and connectivity also comes cybersecurity.
“As we embrace connectivity and digitalization, cybersecurity becomes more critical than ever before,” said Maheshwari. “Without a strong defense, the benefits of AI and autonomy can be undone by single cyberattack.”
Cybersecurity
At HUG, Honeywell introduced a suite of AI-enabled cybersecurity solutions designed to bolster the defenses of OT environments against the evolving landscape of cyber threats. The solutions include:
Honeywell Cyber Proactive Defense, software that uses AI and behavioral-based analytics to detect anomalies by establishing a comprehensive baseline of system operations and provides actionable insights designed to strengthen OT cyber defenses.
Honeywell OT Security Operations Center, a vendor-agnostic and agentless service designed to provide industrials with advanced capabilities tailored to OT environments, providing a 24/7/365 view of the cyber threat landscape.
The Honeywell offering starts with asset discovery, including risks and vulnerability, pulling it together in a report to provide a view of what could be happening from a vulnerability and compliance standpoint. There are also cybersecurity services, including software, professional services for training and awareness and managed services including secure remote access for patching vulnerabilities.
“The OT Security Operations Center pulls everything together because it monitors the assets, detect threats and vulnerabilities, sends alerts out to the security operations center for analysts to take a look at, and then follows the proactive defense solution that offers next steps,” said Nav Sharma, cybersecurity product leader for Honeywell, during a press briefing.
Ransomware attacks are getting more sophisticated and more frequent. Ransomware attacks continue to rise in the industrial sector, jumping by 46% from Q4 2024 to Q1 2025, according to Honeywell’s recently-released Cybersecurity Threat Report. In the first quarter of 2025, an additional 2,472 reported ransomware victims were documented – on top of the 6,130 identified in 2024. These incidents, especially those that directly attack OT environments, can cause significant downtime, lost productivity and safety concerns.
With more bad actors demanding higher ransoms and breaching OT networks faster, there’s a need for ways to identify cybersecurity incidents faster with more clever solutions.
Honeywell Cyber Proactive Defense, using historical data and AI, can model the process, identify anomalies and predict threats. It also includes deception technology, a decoy-based method that lures the threat actors away from the real assets. While deception technology is not new, Honeywell has been able to put it deeper into the control environment to cover safety systems.
“If you’ve seen the Ocean movies where they go into a museum and steal the art, basically the deception technology is creating a fake gallery, and the attackers will try to go there,” said Dragos Rosioru, Honeywell’s senior global OT cybersecurity product leader. “But we also have sensors to detect when someone is trying to get there. If you are someone that works at the gallery, you’ll know it’s fake, so you won’t enter it. So, whenever someone enters it means there is an attacker there that is trying to compromise you.”
Automating system health
Also at HUG, the company announced the expansion of the Honeywell Digital Prime platform to encompass an enterprise-wide set of solutions that test and modify engineering projects before implementation, helping reduce plant downtime and increase throughput when deployed into production.
Launched two years ago as a digital twin for tracking, managing, and testing process control changes and system modifications, the Digital Prime platform, which sits on top of the Experion DCS, is now expanding with an analytics engine to connect to services that can easily gather information around lifecycle, inventory management and migration planning.
Looking to drive more value through service offerings, it can bring in data that would have historically been done manually by an engineer. “With Digital Prime, at that base level, what we’re doing is taking some of the technology we’ve developed and automating it,” said Christian Salazar, Honeywell Digital Prime’s director of portfolio management. If it’s a migration, “the customer can instantly register that migration, they know exactly what their support schedules are, and then that migration happens with that twin.”
Digital Prime enabled services go beyond migration and look at the whole system health, which will also include cybersecurity. The expanded version of the Honeywell Digital Prime ecosystem will be available to customers in Q4 2025.
