Introduction
Enterprise AI adoption has moved from a discovery and experimentation phase to a more demanding execution phase, where the challenge is operationalizing AI at scale. A recurring theme in analyst briefings is that AI proofs of concept often stall in pilot and never reach production. In a state-of-AI survey, McKinsey reports that, early on, two-thirds of respondents say their organizations have not yet begun to scale AI across the enterprise. Although organizations have access to a range of AI capabilities, only some have converted AI assets into production-grade business transformation solutions. Much of the market is now stuck between sandbox success and workforce-scale execution. While there is a desire for measurable returns, most companies still struggle with delivering AI at scale.
On February 10, 2026, DXC announced the completion of an enterprise-wide deployment of Amazon Quick for 115,000 employees in 70 countries, along with a new DXC Amazon Quick Practice to help customers do the same. In an analyst briefing, Russell Jukes, DXC’s chief digital information officer, described what worked, what failed, and how DXC has turned that deployment experience into a customer offering. DXC’s deployment of Amazon Quick was a hard-earned enterprise transformation rather than a normal software rollout. This is because products like Amazon Quick change how employees work day to day, and implementing such a solution requires enterprises to understand how IT governs systems that can act with increasing autonomy.
The sections below examine DXC’s strategy and experience, the process of deploying an AI solution, the new services it is selling, and the hurdles ahead.
The Playbook DXC Had to Throw Out
Before Amazon Quick, DXC attempted to roll out an AI capability using the same playbook it applied to previous software rollouts, with scorecards and adoption metrics, but was unsuccessful. The lesson learned was that AI capabilities need a completely different approach.
DXC’s rollout departed from the traditional playbook, in which success is declared once licenses are provisioned. Instead, the company built the program around what it calls AI fluency. With this approach, adoption is treated as a learning curve, with gates that employees must successfully navigate to advance. Employees’ use of an AI product they do not fully understand can create a material business risk. DXC made fluency goals mandatory across the entire workforce and deliberately held off on formal productivity metrics until the foundation was in place. The rationale is that when output is measured too early, employees optimize for the metric rather than learning. Ideally, the value case comes later, once fluency has taken hold.
Trust in outputs is the other difference from conventional IT projects. With AI results feeding business decisions, DXC invested heavily in grounding its AI in organizational truth.
To achieve this, all traffic flows through an AI gateway, where answers are checked against known truths, a predefined set of facts. The view is that if the truth layer is wrong, nothing downstream matters, including governance and architecture. That discipline extends to how agents are built. DXC drew a hard line between personal agents, which any employee can create for their own use, and professional agents, which only IT can build and which operate under stricter security and governance standards.
Measuring AI Value with Personal Productivity and Process Redesign
DXC uses simplified messaging to clearly communicate the value gained from AI projects. Personal productivity targets time saved by everyone, while process improvement challenges existing processes to look for inefficiencies.
DXC says employees were sending 2500 to 3500 emails a day with simple questions like ‘Where is my Purchase Order?’ Now, one AI agent handles these routine questions, cutting email volume and freeing time for higher-value work. DXC also cited a threefold improvement in invoice processing using invoice-reading agents. DXC used these examples of AI agents delivering value at the individual employee level.
In a process improvement effort, DXC initially sought to accelerate new-supplier onboarding. It then addressed the underlying need by creating an agent that searches for existing suppliers across 130 countries before adding a new one. By reusing qualified suppliers whenever possible, DXC reduced onboarding work and delivered the same outcome with less effort.
When small productivity gains are multiplied across DXC’s roughly 40,000 engineers who use a single agent, the results are measurable. When multiplied across multiple agents and the entire workforce, the bottom-line value of individual productivity becomes apparent in enterprise profitability. Another lesson is that the goal is not simply to automate existing steps, but to examine why those steps exist and redesign the process where needed.
How Internal Deployment Became a Go-to-Market Offering
DXC took a risk in pressure-testing a brand-new product, working hand in hand with AWS’s product development team. AWS and DXC both benefited from working together on a product that was just a few weeks old. Feedback from DXC employees, along with Amazon’s own employees, helped AWS build a better product as it added functionality at a high rate. In the process, DXC developed implementation expertise it now applies in customer engagements.
The experience of using the product, along with integrating it with their own systems, data, and workflows, enabled DXC to build the operating models and governance around it. The internal experience was used to create a new DXC Amazon Quick Practice that helps other enterprises deploy the product at scale. Using a product like Amazon Quick introduces new challenges, such as how long to leave an agent in production before retiring it. Another challenge is handling token spending when considering the product’s ROI.
By framing the rollout as both difficult and repeatable, DXC is positioning that experience as a marketable services offering.
Three Hurdles DXC Says Are Important for Every Enterprise to Take
DXC shared several challenges the company faced and still faces when using Amazon Quick. It’s important for other large organizations to be aware of these.
1. Executives demand to see the value of the product early on during the implementation cycle, but building AI fluency takes runway and sustained executive support.
2. Because DXC runs tools from multiple vendors, its opportunity to consolidate onto a single AI workspace is limited. The choice for a single AI tool can lead to large organizations adopting AI more easily.
3. Deploying an AI workspace across 70 countries and a large workforce creates localization, terminology, and governance challenges. For example, the knowledge graph needed to be taught to say “offerings” instead of “products”. Implementing an enterprise-grade AI workspace required significant configuration changes.
DXC’s experience shows that reaching every employee across multiple geographies with a common workspace is hard, but achievable with a different approach.
Conclusion
To deploy an AI workspace, DXC had to throw out the old SaaS implementation playbook and adopt new practices, such as employee AI fluency. While DXC encouraged individual innovation, it also raised the bar for enterprise agents. Its ROI framework was designed to help decision-makers easily judge success. DXC took advantage of the first-mover opportunity with Amazon Quick adoption and launched its own practice.
Because AI platforms are evolving quickly, DXC’s rollout may offer lessons for other large enterprises, but the transferability of the knowledge still needs to be tested. IT buyers should delve deeper into DXC’s experience and evaluate how their organization can handle a large-scale deployment of new technology. Lessons learned by DXC, like AI fluency and the use of organizational truth for accuracy, must be understood in their own context to evaluate readiness. DXC’s decision to use technology that was not yet generally available was a significant risk. Still, it also demonstrated the organizational capability to assess risk while choosing to be “Customer Zero”. It also highlights the trust DXC placed in AWS’s ability to deliver a product at a rapid pace using AI-native tooling, as reported by AWS. Time will tell whether DXC can replicate its internal success with customers—the company reports strong customer interest, with a first customer implementation well underway.