Suresh Venkatarayalu says Honeywell has been seen as a control system company for 140 years.
“if you really go back and look at Honeywell's history for 140 years, we've been seen as a control system company.”
Person
Suresh Venkatarayalu
1 episodes · 19 quoted claims
Suresh Venkatarayalu describes Honeywell's framework of autonomy as four layers: see, think, act and learn.
“There are four layers to autonomy, which is see, think, act, and learn.”
Venkatarayalu says sensor and field device inputs can contain 20 to 60% false positives that AI can help remove.
“Assume that you're actually taking twenty to 60% of false positive coming from sensor inputs or from field devices.”
Honeywell began training AI on legacy knowledge such as service tickets about three years before the interview, when early frontier models like GPT 3.5 launched.
“Maybe three years ago, when the frontier models were launched, like GPT 3.5 and Gemini initial version, we started by taking all our legacy knowledge from service tickets.”
Honeywell's first AI innovation in this effort was called Maintenance Assist, a tool for service technicians.
“I think the first innovation was called maintenance assist.”
Venkatarayalu says one of Honeywell's largest ethylene plant customers processes roughly 20,000 to 30,000 sensor data points every second.
“This is one of our largest ethylene plant, where we have close to twenty to 30,000 sensor data that is processed every second.”
Venkatarayalu says the ethylene plant's sensor processing produces about 15,000 to 20,000 alerts or alarms.
“you will end up seeing closer to fifteen to 20,000 alerts or alarms that keep popping up.”
Venkatarayalu says Honeywell uses a system called expedient cognition on top of the ethylene plant to separate false from real alarms.
“I think we use a word called expedient cognition, which sits on top of this ethylene plant”
Venkatarayalu says Honeywell's probabilistic AI models may be about 85% accurate but are overseen by deterministic control systems he says provide 100% assurance.
“we have the probabilistic model that could be 85% accurate, but we give the reins back to the control system that we have built, which always assured 100%.”
Honeywell's building management automation system is called Enterprise Buildings Integrator, abbreviated EBI.
“we call EBI, enterprise building integrator in our building management automation system.”
Venkatarayalu says connecting a million-square-foot, ten-floor building with about a thousand assets once took 120 to 150 hours.
“It would have taken anyway between 120 to 150 hours.”
Venkatarayalu says Honeywell can now connect a building in roughly ten to twenty hours.
“We can connect anywhere between ten to twenty hours within a building context.”
Honeywell created a program called Forge Academy to train and certify system integrators.
“We deployed a massive program called Forge Academy.”
Venkatarayalu says Honeywell Forge had about 700,000 connected assets two years earlier and 5.3 million as of July 2026.
“Two years ago, we had probably 700,000 assets connected. As of July 2026, we have 5.3 million assets connected to our forge platform.”
Venkatarayalu says Honeywell's installed base is estimated at 11 to 12 million assets.
“We believe that our install base is close to eleven to 12 million assets”
Venkatarayalu says connecting an industrial environment is about ten times more complex than connecting a commercial building.
“an industrial environment is ten x complex than a commercial building.”
Venkatarayalu says Honeywell can render a building's digital twin from CAD drawings and piping and instrumentation diagrams in one to two days.
“We were able to feed the CAD drawing of a building and a P-N-I-D documents that are available today within a day or two days.”
Venkatarayalu says rendering a building's three-dimensional model took about three months six months before the interview.
“the process itself would have taken a building like this three months to just render the three-dimensional artifact.”
Venkatarayalu says Honeywell's North Carolina campus uses face scanning to identify employees and send them to their floor without an access card.
“We implemented in our North Carolina campus with ID Mia, your ability to scan your face and then it determines which floor you belong to.”