Honeywell launches its building AI with autonomous mode switched off
Honeywell's CTO says Forge Cognition will run with people in the loop for probably two to three years, autonomous mode off. The evidence so far is one vendor's pace, tested in Nvidia's own headquarters; how fast buyers move is still a forecast.
The Investor · Invest desk
Key figures from Voyager: up to a 90% reduction in unnecessary Building Management System alerts, per Honeywell; close to 90% to 95% of security and access-control alarms were false, per Honeywell's CTO; roughly 10% energy efficiency identified, per Honeywell, as potential savings, not achieved yet.
| Figure | Value | Note | Claim |
|---|---|---|---|
| Cut in unnecessary BMS alerts, up to, per Honeywell | 90% | 11 | |
| Access-control alarms that were false, close to | 90% | close to 90% to 95%, per Honeywell CTO | 10 |
| Energy efficiency identified, roughly, per Honeywell | 10% | potential savings; not achieved yet | 15 |
What happened
- The platform is built with technology from Nvidia and Dell and layers AI agents and digital twins over the heating, cooling, security and power systems in buildings.
- The project began at CES in January, when Nvidia asked Honeywell for a lighthouse customer and Honeywell CEO Vimal Kapur asked why Nvidia could not be it.
- The two companies set a 90-day target at Voyager, Nvidia's roughly 750,000-square-foot headquarters, with five agreed targets tracked across about 900 pieces of equipment.
- False alarms made up nearly 90% to 95% of the total in Voyager's security and access-control system, and Nvidia kept eight or nine people on each shift to clear them.
Why it matters
- What it limits constraint A buyer cannot get software acting without sign-off at launch, so any savings case that depends on autonomy waits out the window while the advisory case has to pay for itself now.
- Who's exposed exposure The only deployment described is at a company that also supplied technology for the product, so the buyer-side evidence comes from a partner, not an independent owner.
- Who has to decide decision Owners will have to choose when to ask for autonomous mode, using a vendor's two-to-three-year estimate and no published customer data.
Next story · Invest · yesterday
NatWest and Nvidia tune speech recognition to UK accents and cut misheard words by 20%Confirmed2 publishers
Venkatarayalu hedged the timeline twice in one sentence. "Maybe we will start running in this mode for probably next two to three years," he said [7]. After that, customers decide when to make the switch [8]. The two-to-three-year figure is Honeywell's pacing of its own rollout, and Fortune's account does not include a customer's timetable.
Honeywell's launch announcement says the agents can "autonomously coordinate actions across existing systems" [3]. Capability and permission are separate switches here, since the platform has three modes [2] and the CTO set the permission himself: "For me, I wouldn't deploy these agents without my control system in place," he said [9].
The Voyager results are, in our view, advisory ones. If 90% to 95% of access-control alarms were false [10], real alarms were 5% to 10% of volume [16], and that is the share a person still has to find. Honeywell says the agents cut unnecessary alerts by up to 90% [11], but Fortune attaches that figure to the Building Management System, not to the access-control stream behind the false-alarm rate, so we have not multiplied the two. Operators can open every alarm the system classified and mark whether they agree [12]. "That's all about bringing in an operator in the loop," Venkatarayalu said [13]. The energy finding is a recommendation too: "We have identified close to 10% of energy efficiency," he said [14], and Fortune notes that this is potential savings, not achieved ones [15].
Our view is that operator trust sets this schedule, and that building owners will move at about the pace Honeywell has chosen. The counter-thesis is that a vendor could pick two to three years for its own reasons whatever buyers feel. The only site Fortune describes is Nvidia's headquarters [5], and Nvidia supplied technology for the product [1] and issued the lighthouse challenge that started the project [4]. We would be wrong if owners ask for the autonomous mode before the two to three years are up.
What to watch
- Whether Honeywell reports achieved energy savings at Voyager against the roughly 10% it has only identified so far.
- Whether a customer outside the Nvidia relationship goes public with Forge Cognition results.
- Any Honeywell customer asking for or receiving the autonomous mode before the two-to-three-year window ends.
Information, not investment advice. This story is general information about companies and markets, drawn from cited reporting. It is not a recommendation to buy, sell or hold any security; check the cited sources before acting.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence35
- Adoption12
- Hype gap+30
- Incentives65
- Confidence60
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
Forge Cognition is built with technology from Nvidia and Dell and puts AI agents and digital twins on top of the systems that heat, cool, secure and power buildings.
- [2]
Forge Cognition can run in three modes: manual, semi-autonomous and autonomous.
- [3]
Honeywell's launch announcement says the agents can "autonomously coordinate actions across existing systems."
- [4]
At CES in January, Nvidia's leadership heard Honeywell's pitch for autonomous buildings and asked who its lighthouse customer was; Honeywell CEO Vimal Kapur asked why Nvidia could not be that customer, and Nvidia took the challenge.
- [5]
Honeywell and Nvidia set a 90-day target to show results at Voyager, Nvidia's roughly 750,000-square-foot headquarters building, agreeing five performance targets and tracking about 900 pieces of equipment week by week.
- [6]
"We have not opened up an autonomy right now," Suresh Venkatarayalu, Honeywell Technologies' CTO, told Fortune ahead of the Forge Cognition launch.
- [7]
"Maybe we will start running in this mode for probably next two to three years," Venkatarayalu said.
- [8]
After the two-to-three-year period, customers decide when to make the switch to autonomous operation.
- [9]
"For me, I wouldn't deploy these agents without my control system in place," Venkatarayalu said.
- [10]
In Voyager's security and access-control system, close to 90% to 95% of alarms were false, and Nvidia had eight or nine people per shift clearing them.
- [11]
Honeywell says the result at Voyager is up to a 90% reduction in unnecessary alerts related to the Building Management System.
- [12]
Operators can open each alarm the system classified and mark whether they agree.
- [13]
"That's all about bringing in an operator in the loop," Venkatarayalu said, so the operator can learn to trust the system.
- [14]
"We have identified close to 10% of energy efficiency," Venkatarayalu said.
- [15]
The roughly 10% figure is potential savings Honeywell and Nvidia have identified through better understanding and optimization of operating conditions; it has not been achieved yet.
- [16]
If 90% to 95% of access-control alarms were false, real alarms were 5% to 10% of volume.
Sources
2 independent publishers whose own reporting we read for this story.
Topics and entities
Topics
- Building Automation
- AI agents in physical operations
- Digital Twins
Entities
- NVIDIA Nemotron
- Nvidia
- Honeywell
- Suresh Venkatarayalu
- Dell Technologies Inc.