Fletcher Previn says advancements in generative artificial intelligence are leading him to rethink workflows not just for Cisco Systems’ 10,000 IT workers, a department that operates on a $1.54 billion annual budget, but also the broader employee base of around 90,400.He favors a collaborative approach that takes into consideration input from both the C-suite and workers.“There’s the bottom up inertia of people asking for access to tools, and then there’s the top-down inertia of saying, ‘Here’s how we see this role evolving,’” says Previn, who has served as chief information officer at the networking-equipment company since 2021. “You sort of have both trying to meet in the middle.” What that means for his work as CIO is finding ways to improve worker productivity, including giving developers more access to AI coding tools like Cursor, Windsurf, and GitHub Copilot. Cisco Systems is closely monitoring adoption of these tools, as well as the amount of code that’s accepted by developers. Around 70% of the company’s 20,000 developers log into AI coding tools at least once per month. The acceptance rate for AI-generated code is around 24% and while that may not sound too impressive, Previn says it is a big leap from 4% nearly a year ago.“AI is getting better,” he adds. “It’s supporting more programming languages. Developers are getting more comfortable with it.” Over time, he hopes that as much as 70% of Cisco Systems’ code will be AI generated.There’s also the ongoing work that Previn is doing to rethink the IT function’s work and responsibilities. He is reevaluating the primary work done for each role, as well as job titles, the ways to reevaluate work with AI tools, and what future training may be required to support such big, sweeping changes.Cisco Systems has also made some investments in AI applications that improve productivity to support the non-technical workforce. Some examples include using AI to do more intelligent onboarding, which allows for Cisco to more accurately assess exactly what tools or software are needed to support the role a new employee is hired for.Cisco Systems is also using AI for planning hardware updates. Most companies refresh their laptops every two-to-four years. But by using AI to detect a laptop’s memory, application performance, and network telemetry, Cisco Systems can decipher the difference between performance problems that can be fixed by the IT team versus when a device may fail and will need to be replaced.Upgrading laptops is a “significant amount of money at a company of our size,” says Previn. “And sometimes a lot of people are perfectly happy with the laptop they have.”While Cisco Systems generates nearly $57 billion in annual revenue, large enough to rank 83rd on the Fortune 500, the company competes aggressively for tech talent, an arms race that has accelerated this year as the generative AI boom continues to expand.To lure skilled workers, Previn says he prioritizes creating a workplace that encourages “emotional safety for people to be able to innovate and experiment and fail fast.”Another key pillar of Previn’s strategy involves recognizing IT’s operation to do all work by a singular platform. There are separate teams organized cross functionally—one to support Workday, another for SAP, and yet another for Oracle—that are made up of around six-to-ten employees who have all the necessary skills to advance an internal tech project into production. This allows small teams to operate independently. “There’s a huge productivity loss when you’re kind of dynamically forming and un-forming teams by project,” asserts Previn.Previn says Cisco Systems fields a lot of pitches from the software-as-a-service providers it works with to add on more expensive AI capabilities. He progresses cautiously on that front. “You can consume all of those things, but run the risk of controlled cost,” says Previn. “It can create employee experience confusion, similar to what we had when chatbots started to become a thing.” He expects that generative AI will increasingly evolve to become an agent-to-agent world, a digital workforce that can perform tasks on behalf of workers, sometimes autonomously and in coordination with each other. That involves an inverted approach to how employees engage with technology. Rather than an employee asking an AI search tool, “Where’s the link to Workday?” they could instead say, “I’m taking a vacation on Friday.” Agentic AI could then do the rest: block off calendars, cancel meetings, and put up an out-of-office message. Cisco Systems has developed the company’s own internal digital AI teammate called “CIRCUIT,” an AI assistant that helps the company’s employees find and understand general information more quickly. Rather than an employee being asked to toggle between different language models or select if the data they want to access is publicly available or is internal information, “CIRCUIT” infers intent and determines the appropriate selections.Previn says this approach is an evolution of his thinking that Cisco Systems can’t assume that workers are up to date on which AI models are most advanced and best match the tasks they are trying to complete.“The rate of innovation in AI is happening so quickly that if you find out your developers are using a language model that is six months old, then effectively, all the software you’re writing is already six months out of date,” says Previn.John KellSend thoughts or suggestions to CIO Intelligence here.This story was originally featured on Fortune.com