CIO Sharon Mandell transforms Juniper Networks for the AI era

CIOs have elevated their stature in the business by driving digital transformations for nearly a decade. Now, top-level IT chiefs are taking this formula a step further by putting digital initiatives to work in helping transform the business itself.

Juniper’s Sharon Mandell is one such CIO, as evidenced by her efforts to support the traditional switching and routing products supplier’s evolution into an AI networking company.

When Mandell joined Juniper five years ago, the company had just acquired Mist Systems, the maker of a Wi-Fi assurance solution that enabled Juniper to enter the lucrative campus and branch-office markets and AI for IT.  Since then, Mist has “evolved tremendously” and expanded its identity into an AI-cloud native platform and network control point for Juniper’s end-to-end portfolio, with insight, automation, and assurance across the full stack, Mandell says.

Traditionally, Juniper’s business processes were targeted toward very large, complex, and long sales cycles. Mist required the exact opposite: a more bundled package sale to the enterprise, Mandell says. It would not be easy, but the CIO — recognizing business alignment was the key — enlisted both the business and technology sides of the house and got to work.

“I locked on to the need for a business transformation, which required a change to many, if not all of our systems,” Mandell says. “Business transformation is a team sport and not always easy. IT can’t make this transformation alone. Leaders and subject matter experts in other impacted functions have to come along with this change in approach as well.”

Much input and planning were required to evolve Juniper’s business model and prepare for a services future, she says. The company started by selecting and implementing new products to better integrate teams. For example, she and her IT team modified Juniper’s Salesforce Opportunity Management system, implemented new sales forecasting approaches using Clari, re-engineered the company’s use of Oracle CPQ, and updated its SAP Order Management systems.

“All of these systems are now more seamlessly integrated into a true end-to-end workflow, which did not exist before, “Mandell says. “Turnaround time to partners and customers vastly improved. This supports a more standardized, product-bundle-driven sales process, which is essential in the enterprise space. That’s why I’ve referred to it as a business transformation as much as a digital one.”

The high-tech difference

To facilitate the transformation, Mandell had to shake up how IT was done.

“We moved away from functionally siloed IT teams and shifted to a product operating model, creating cross-functional teams built around business capabilities,” the CIO notes.

Next up was altering the company’s product development methods. “We also shifted from traditional waterfall methods to a more agile development approach with sprints and ongoing testing,” Mandell says. “This allowed us to share demos and get feedback every two weeks, making the process far more integrated and iterative.”

Being the CIO of a technology company differs from helming IT in other industries in several ways, one of which is that Mandell is working with a very technically savvy business base, many of whom think they could do the job better than the CIO “and I suspect they would approach the job with much more humility once they actually tried doing it,” she quips.

The tougher part of being a CIO of a high-tech company is the breadth of technologies and systems. “As an engineer, I had a deep focus and a specific toolset,” says Mandell, who came to Juniper after serving as CIO of Tibco and has held engineering roles throughout her career. “IT is where product vision meets actual human use — and that gap is often wider than people imagine.”

Cloud-native foundation, AI advantage

Juniper’s previous CIO migrated to the AWS cloud but, like most enterprises, also operated pockets of Google Cloud for data and Azure when specified by the customer. The company also embraced SAP Analytics and Snowflake for its data architecture and employed machine learning to automate some processes.

Having this cloud-native foundation enabled Mandell to tackle the AI networking goal much faster, she says, as it facilitated the collection of data that relayed how the network was functioning and helped make the product better — without the customer knowing or objecting to upgrades.  

Being cloud-native also enabled Juniper to understand the network in new ways, and make adjustments quickly, notifying users before the help desk started getting calls, or getting lucky enough to fix problems before users even experienced the problem, Mandell says.

“Collecting and having AI give you information helps you build a much more responsive and resilient network with less downtime and less truck rolls,” she says. “You use that control point in the cloud.”

Integrating multiple acquisitions has been Mandell’s key challenge at Juniper. Mist’s Wi-Fi assurance solution has expanded to wired networks and brought AI-native capabilities to Juniper’s switching portfolio, she says. Juniper also integrated 128 Technology’s SD-WAN into the platform and more recently embedded NAC capabilities from another small acquisition directly into the Mist cloud. Its acquisition of Apstra, moreover, contributed data center networking to Mist, giving customers better visibility and ease in upgrades.

Mandell’s IT team “drinks its own champagne” and is Juniper’s top customer, testing out early innovations, providing technical feedback to product development teams, and resolving issues faster. Juniper runs its own MPLS network on prem, which Mandell says is unusual for a company of its size but allows Juniper’s internal and external IT pros — a headcount of 600 internally and externally — to work directly with Juniper’s product development teams. 

“One of the benefits of working at a networking company is having engineering on speed dial,” she says.

Since then, Mist has “evolved tremendously” with Juniper’s aligned business and technology teams.

“A key differentiator of Mist is that it’s built on a proven microservices cloud architecture. This provides the elastic scale, resiliency, and performance demanded by the world’s largest retailers, healthcare systems, universities, and enterprises,” Mandell says. “Mist runs in the public cloud across multiple providers, giving customers the flexibility to meet their security or compliance preferences. Being AI-native isn’t just a tagline; it’s foundational to how we deliver a better networking experience.”

HPE made a move to acquire Juniper for $14 billion last year, but the DOJ rejected the move. HPE is awaiting news of its appeal and has until October 2025 to finalize the deal.

Mandell remains focused on employing AI and other features to keep Mist on top. “While the legal process plays out, there will be no disruption to the way we do business,” she says. “Our operations will continue as usual, and we remain focused on delivering AI-native innovation for enterprise, service providers, and cloud customers.”

Brandon Butler, a senior research manager at IDC, says Juniper’s early focus on AI networking has positioned the company well.

“From the beginning, Juniper has been thinking about how they can use AI natively within the platform to ease the operations of the Mist platform, to reduce the errors and the problems that can arise in the network. The Mist platform is really good at being able to fix those automatically without human intervention,” Butler says, noting that Cisco and HP have also embraced AI but “Juniper has really made it the centerpiece of their marketing strategy and their go-to-market strategy. Everything is AI-driven.”

Generative AI inside

Juniper uses generative AI within the company and in its products and will continue deploying the rapidly advancing technology, Mandell says.

Internally, the company uses Microsoft Copilot, GitHub Copilot, Copy.ai, and other tools across multiple functions. In marketing, generative AI helps Juniper generate and personalize content based on user behavior. The company also uses a specialized RFP generative AI service in support and go-to-market to draft RFP responses, improving both efficiency and scale. Gen AI is also used in documentation — for example, to create drafts from product specs that are then refined by humans and turned into training materials, including multilingual voice tracks for video modules.

Externally, generative AI enhances how Juniper supports its customers. “It powers our support chatbot and simplifies access to documentation and knowledge bases,” Mandell says, noting the company’s AI assistant, Marvis, has used natural language processing and understanding for over a decade — and now incorporates gen AI to expand those capabilities even further.

“Customers benefit from faster issue resolution and more intuitive self-service experiences,” Mandell says.

Agentic AI on the horizon

“While our own products don’t require gen AI in the same way a consumer application might, we build the networks that allow our customers to run gen AI workloads, including training, inference, and storage clusters,” Mandell notes. “That infrastructure is a critical part of enabling gen AI innovation across industries.”

Juniper is evaluating upgrading its ERP system but Mandell plans to wait and see what agentic AI brings to the table.

“We haven’t made that decision yet — and honestly, I expect ERP will look very different in the age of agentic AI. We’re still early in that exploration. Right now, we’re evaluating ERP vendor roadmaps and beginning proof-of-concept work with agentic platforms that could sit alongside the core transactional system,” the CIO says. “The architecture in this space is evolving rapidly, and we want to be well-informed and ready when the time comes to make that choice.”

She adds: “By the time we put our hands on keyboards, agents are going to become a more common thing. AI may be able to help us accelerate that process. I want to do that in advance of anyone getting to this fixed notion of what our ERP is going to be, because I think the world looks a lot different in a few years than it does today, given the pace at which this technology is on. We have to be open to everything.”