November 18, 2026
How CDW Scaled Its Artificial Intelligence Program From Pilot to Production
To operationalize an AI program that now supports thousands of employees, CDW followed a holistic, methodical strategy that put all the right pieces in place.
CDW’s salesforce services the needs of more than 250,000 customers. As a leader in the technology industry, the company has turned to artificial intelligence to help it achieve that objective.
Just seven months after CDW launched Seller Assistant 1.0, an AI agent for its 4,500 U.S.-based sellers, the company is transitioning to Seller Assistant 2.0. The new version can handle millions of tasks but has 17 built-in "skills" that orchestrate more complex workflows for sellers. It also identifies which aspects of a sale an agent supported so that CDW can measure revenue and ROI — a capability that stymies many AI adopters.
CDW has achieved AI momentum at scale because it has spent the past three years building a solid, strategic foundation. Today, CDW’s AI program is intentional yet adaptable, ambitious and well protected.
According to Ben Weiss, vice president of enterprise AI platforms and products, and Yasim Abbas, senior manager of AI product strategy, CDW has moved from isolated pilots to operationalized production. As leaders of CDW’s strategy, development and delivery of high-impact AI products, Weiss and Abbas have helped scale these efforts through a deft combination of bottom-up empowerment and top-down alignment.
“The bottom-up would be providing a broad set of coworkers with the access to build agents for themselves,” says Abbas. “The top-down is scaling it for the whole enterprise. That’s where you get a lot of questions, and it’s critical that you define the business outcome.”
CDW’s AI foundation also addresses ownership, data readiness, adoption and governance. Ownership ensures accountability; without it, teams may be tempted to abandon a pilot when they encounter friction. Data readiness addresses integration, monitoring and scaling. Adoption strategy considers how roles and workflows will change and addresses the “people” side of the equation. And, of course, risk management establishes critical guardrails and keeps humans in the loop at well-defined points.
Another factor in CDW’s success was an intentional effort to help employees understand and get comfortable with AI as a source of empowerment, says Weiss.
“That helps people get up the learning curve faster, to a point where they are less fearful about this ‘taking away their jobs’ and start to embrace the idea that this is just a tool for them to achieve more,” he says.
Enterprisewide Access, Supported by Multilayered Controls
As CDW tells its own customers, AI is a journey. CDW has been advancing steadily on this journey since late 2023, when it first began deploying Copilot to thousands of coworkers. That has grown now to 13,000 employees, or three-quarters of CDW's global workforce. Employees also have access to Microsoft’s agent-building platforms: M365 Agent Builder, a custom no-code tool for beginners; Copilot Studio, a configurable, low-code builder for intermediate users; and Azure AI Foundry, an expert-level engineering platform.
So far, employees have created more than 10,000 personal agents, democratizing AI across the enterprise. For a company of CDW’s size and scope, centralized monitoring and management of agents is essential. To help, Microsoft’s Agent 365 provides visibility and controls across all of the agents in CDW’s environment.
“At any moment, Yasim or I could go in, and in about four seconds, shut off any one of those agents,” Weiss says, if a problem arises with an agent’s operation.
Microsoft Purview, a unified platform for governance, risk and compliance, sends alerts to the security team about agent actions or conversations that require human review. The agent builder tools use data loss prevention policies to systemically enforce governance, and that helps strike the right balance of coworker empowerment with agents and risk management.
A governance committee comprising representatives from CDW’s legal, security, privacy, AI, IT engineering and infrastructure teams must approve any agents with privileged access or movement within sensitive environments.
Strong governance “is what helps us sleep well at night, knowing that, yes, we’ve given thousands of people access to do this kind of thing,” says Weiss. “However, they’re limited in what they can do. Some companies would hear 10,000-plus agents and say that’s chaos. But with the right policy-level enforcement, you can absolutely have that.”
Sellers’ ‘Super Agent’ Frees Time for Customer Relationships
Seller Assistant 1.0, which debuted in February 2026, was a turning point in CDW’s pilot-to-production journey, says Weiss. “That was the first time we had a custom agent for our salesforce that was pushed to thousands of people, and it wasn’t just a generalized productivity tool, like Copilot,” he says.
Instead, the agent’s purpose is to remove friction from workflows underpinning CDW’s core business: selling products and services and acting as a trusted technology adviser. The agent launched with three use cases, including providing tailored recommendations for customers based on their needs and challenges. While useful, CDW wanted the agent to perform more substantive tasks.
Seller Assistant 2.0, launching in fall 2026, will deliver. It can fulfill 17 “complex skills” on sellers’ behalf, such as creating go-to-market strategies around high-impact opportunities or helping sellers navigate Salesforce CPQ (configure, price, quote) to create faster, more accurate quotes. It can also handle far more skills than that through the use of its connected tools.
“Where the original Seller Assistant could handle a task that might take one or two minutes, it can now take on tasks that would take 20, 30 or 40 minutes and work to achieve a goal, so you can give it much harder problems,” Weiss says.
The agent records its own memories, which enables it to track each seller’s customers and preferences, so every interaction becomes more tailored. “It feels like it just knows you, and that makes it a really good, personalized agent,” says Weiss.
Seller Assistant 2.0 also enables ROI tracking, closing another critical gap, says Abbas.
Weiss notes, however, that measuring how much agents contribute to revenue is just one metric of success. “The ultimate outcome is that the seller has more time to spend with their customer, building relationships and being a trusted adviser, and less time in front of a screen,” he says. “We’re in a sales business, and relationships are where things are won or lost.”
What Separates Production AI From a Pilot?
The scale of AI deployment is one measure of progress: Is AI restricted to a handful of users or operationalized across the enterprise? But numbers alone don’t tell the whole story. What truly separates an AI pilot from sustainable production are other markers of maturity.
Governance and risk management: CDW uses multiple layers of control across people, processes and technologies — including sandbox environments, data loss prevention policies and centralized approval of higher-risk agents — to establish guardrails and reduce risk.
Business outcomes: Clear objectives and knowing how to measure success keep teams on track. CDW uses quantitative and qualitative assessments. “Every CFO has to ask, ‘Are we spending responsibly on these investments?’ That’s real, and we need to be able to provide clear signals to that,” says Ben Weiss, CDW’s vice president of enterprise AI platforms and products.
Data readiness: Getting data architecture ready for AI agents encompasses models, integration and infrastructure. Connecting systems and data at scale enables advanced use cases and deepens AI’s organizational impact.
Adoption: In addition to its Viva Engage channel, CDW offers employees workshops, videos and presentations to facilitate adoption and AI fluency.
Ownership: AI initiatives often hit snags, such as technical challenges and funding shortfalls. Committed ownership keeps things moving over and around obstacles.
High-Level Ownership Keeps Projects Moving Forward
Throughout CDW’s AI journey, project ownership has proved crucial. When CDW began prioritizing ownership at the outset, momentum followed. “There’s been an effort that when we rally behind a particular area that we want to improve by bringing AI to bear, we get senior-most leaders together to align and say, ‘This is where we’re going to focus, we’re putting intentional dollars and a team behind this,’” says Weiss. “And the owners can’t be just technologists. It’s got to be the business.”
CDW also resolved challenges related to data readiness. Weiss says that AI prototypes often look great but may have limited functionality because they lack the necessary data. CDW solved that by building Model Context Protocol servers that enable agents to consult systems of record, such as pricing data and the sales catalog.
“Agents really need that MCP infrastructure to be able to use all those different tools, whether it’s for gathering data or taking action within the underlying system,” Weiss says.
“The ultimate outcome is that the seller has more time to spend with their customer, building relationships and being a trusted adviser, and less time in front of a screen.”
— Ben Weiss, Vice President of Enterprise AI Platforms and Products, CDW
Beta Tests and Employee Engagement Drive Adoption at Scale
Even when organizations get everything else right, they may struggle to move past pilots for a deceptively simple reason: an inability to overcome employees’ reluctance to actually use AI.
“Building trust and confidence is crucial,” says Abbas, adding that CDW tests new agents with more than 1,000 members of a beta community of coworkers before rolling out the agents enterprisewide. “We get their feedback, we make sure we’re listening, and we make changes based on that. That ensures the product is successful as it goes into the end users’ hands.”
Meanwhile, up to 6,500 employees collaborate on AI through a community in Microsoft Viva Engage, a level of participation well beyond Weiss’s initial expectations. The community also supports adoption by making it easy for employees to share notes and learn about new AI features.
Three years into CDW’s AI journey, Weiss and Abbas say that democratization is what makes AI powerful, and CDW has been intentional in leveraging that across the enterprise. Microsoft’s suite of agent platforms allows someone with no technical knowledge to write a skill using plain language. When someone with domain expertise can tailor agent skills to real-world workflows, AI use cases can deliver true value, Abbas says.
That type of thinking permeates CDW’s bottom-up/top-down strategy. “The things that people do in the ‘bottom-up’ that are really good rise to the top,” Weiss says. “And, we can incorporate them into those top-down initiatives. That’s where this gets really exciting.”
CDW’s experts can help your organization move AI
forward at every stage of the journey.
Amy Burroughs
Writer