Sheila Jordan is such a big believer of generative artificial intelligence that she’s already put it into the hands of all 95,000 employees at Honeywell.
“There is not a function in the company that is not thinking about gen AI,” says Jordan, senior vice president and chief digital technology officer at Honeywell since 2020, after long leadership tenures at Symantec, Cisco, and Walt Disney Co.
While Jordan acknowledges there remains an ongoing debate between those that have bought into the hype of generative AI and the opposing naysayers that aren’t yet sold on the technology, she is such an ardent believer that Honeywell has already put 16 generative AI use cases into production. To generate excitement internally, Honeywell allows employees to submit ideas for how to use the technology and hundreds of proposals have been submitted.
But Jordan is also keen to remain fiscally responsible when thinking through when to leverage generative AI. She prioritizes the biggest, most impactful ideas first, those that can increase revenue, boost productivity, or make employees more satisfied at work.
Generative AI has cascaded across Honeywell in quite a few ways, with some of the earliest use cases stemming from the company’s relationship with Microsoft, who Honeywell relies on for Azure cloud computing services. Around 5,300 of Honeywell’s employees have access to the Microsoft 365 copilot and over 4,500 software engineers are using GitHub to write 90,000 lines of code per week. The engineers find the productivity gains appealing, with Honeywell reporting a 65% usage rate for GitHub.
Honeywell also rolled out some tools, including an AI copilot offering from Moveworks, to automate some IT help desk requests. There’s been a 80% drop in inbound tickets as virtual assistants are able to handle a bulk of the workload, leaving only the most complex issues to be handled by the human IT help desk.
Honeywell has also rolled out a generative AI virtual assistant, called Red, which can quickly pull data from the company’s archive, including 350,000 pages of product manuals and over 50,000 internally crafted articles. Red offers responses in more than 100 languages and is accessible to every employee with a laptop computer.
Jordan says that one of the biggest problems that companies like Honeywell face today is that they are swimming in too much data that’s not always properly organized. Generative AI can take that data, including unstructured data from video or word documents, and organize it for better insights that “gives you a whole new view of what’s happening in your organization,” says Jordan.
Her efforts to recognize data stored across Honeywell predates the generative AI boom, but was opportune given the importance of data to make generative AI work properly. Four years ago, Jordan started to streamline the number of software applications used at Honeywell. When certain vendors, like Salesforce or SAP, are selected, that solution is put into use across the entire organization.
As a result, Honeywell now has just over 1,000 software applications in use today, sharply lower than 4,500 when Jordan came on board.
Honeywell now leans on Snowflake for all key and critical data, including bookings, billings, inventory, HR, and engineering data. The data-driven approach allows Honeywell to expedite decision making, make more accurate predictions, and even get more inventive on dynamic pricing, which are tactics that allow companies like Honeywell to charge higher prices when factoring inputs including regional preferences, seasonality, and surges in product demand.
“You can’t have a gen AI strategy unless you have a data strategy,” says Jordan.
Beyond working with Microsoft, Honeywell is also exploring generative AI tools from AWS and has had conversations with Google and others. “I actually think the technology is not as important as the use,” says Jordan, who believes all the large players will provide great AI technologies.
That said, Jordan doesn’t want a plethora of large language models put in use across Honeywell, as it would make it difficult to adhere to the company’s responsible AI governance and security principles. “If you have too much technology, it can get hard to manage,” says Jordan.
And while a recent survey conducted by Honeywell found that only 17% of AI decision makers in the industrial sector have fully implemented their initial plans for the technology, Jordan says she’s been in technology long enough to remember when corporations were worried about letting their employees use their personal phones for business use. They eventually worked through that problem, and they will again with generative AI.
“You’re not going to get everything right,” says Jordan. “Figure out the best use case and continue to use it. Because it’s not going to go away.”
John Kell
Send thoughts or suggestions to CIO Intelligence here.
