Q&A | September 2, 2026

Co-innovating enterprise AI: When DXC deployed AI to 115,000 employees, here's what AWS learned

A conversation with Russell Jukes, Chief Digital Information Officer at DXC Technology, and Jose Kunnackal, Director of Amazon Quick at AWS

In February, DXC Technology completed a global rollout of Amazon Quick, an AI platform from AWS that lets anyone in the organization, from finance to HR to marketing, query their own tools and data in plain language and act on what they find.

The deployment spans more than 115,000 employees across 70 countries, and the real-world feedback it generates flows directly back into how Amazon Quick is built. Because DXC operates the platform at the same scale it recommends to customers, it can offer something most system integrators can't: first-hand data to guide other enterprises through the same process faster and with fewer surprises.

In the following conversation, DXC’s Russell Jukes and AWS’ Jose Kunnackal discuss why deploying AI is fundamentally different from deploying software, how they measure success beyond adoption numbers and what it takes to build a co-innovation partnership at enterprise scale.

What is the hardest part about building a product together?

Russell: Building a product together in this era is genuinely challenging and well worth the investment. There is no shortage of good ideas — we have them, AWS has them, every customer has them. Everyone arrives with use cases they want solved. The real work is prioritization: deciding which use cases to tackle, in what order and then figuring out how to solve them together. That takes discipline from both sides. But when you get it right, the results reflect it. DXC's partnership with AWS isn't a buyer-supplier relationship. It's genuine co-innovation.

Jose: From the AWS side, building together means constant contact — staying close to the use cases that actually matter. When we look at how something needs to roll out across 100,000-plus users, the conversation is always about how to make it secure without sacrificing the end-user experience. Getting both right is harder than it sounds, and you only learn what that balance looks like by deploying at real scale. The Quick business desktop is the latest addition to the platform, and DXC already has it deployed across thousands of users while we continue shaping it together for joint customers.

Q: Amazon Quick is powered by agentic AI. What was most surprising about deploying AI that most enterprises don't typically encounter with traditional software?

Russell: Deploying AI is not the same as deploying SaaS. I learned that the hard way. I rolled out our first AI capability the way I would roll out any SaaS application, and it simply doesn't work like that. As an industry, we have spent enormous sums on AI without seeing the returns everyone is chasing, and my role offers me a unique vantage point to try and understand why. As I see it, it starts with sponsorship at the very top of the organization, but it doesn't stop there. You have to work down through every level and ask honestly: how does this technology actually work and where will it create specific value for us? Without that alignment — executive, managerial, and all the way through the organization — the platform never delivers what it's capable of.

Jose: From a product standpoint, scale is a requirement, not a feature. Can the platform roll out to hundreds of thousands of users, perform reliably and deliver the controls an enterprise needs? You have to be able to answer yes to all of that before you can credibly go to market. And you need proof, not a promise.

The other question that comes up in almost every organization is: why wouldn't we just build this ourselves? It's a fair question. But the pace at which AI is evolving makes it genuinely difficult to keep up, not just with the technology, but with user expectations as they shift. By the time an internal build is production-ready, the landscape has moved. A proven platform at scale lets enterprises stay ahead of that curve rather than chase it.

Q: Once the technology is ready, what does it take to get 115,000 people to change how they work?

Russell: At DXC, we deployed Amazon Quick to the entire corporation, and that comes with real responsibility — responsibility that ultimately sits with me. The foundation has to be right: safe, secure and scalable, with the appropriate guardrails in place. We started with use cases where people could verify the answer themselves, so they could build confidence before relying on it for higher-stakes decisions. From there, adoption tends to follow a natural arc. People start by asking questions, looking for answers on demand. But for that to work, they have to trust what they're getting. They need to know the answer is accurate, and that it's being used to inform a decision, not make one. Because AI supports decision-making. It doesn't replace it.

Jose: You're not deploying just another tool. You're deploying something that changes how people work. That distinction matters. The goal is to introduce it in the natural flow of work, so it feels like an evolution rather than an interruption. That starts with helping people understand what the tool can actually do and how it fits into their daily habits. Then you layer in guidance, not just on Quick specifically, but on AI more broadly. How do you think about it? How do you prompt effectively? Those become the foundational exercises. And as people grow more proficient, you meet them at the next level. That progression, from awareness to fluency to genuine capability, is what sustainable enterprise adoption actually looks like.

Q: Most enterprises default to tracking AI adoption when measuring for success. What should they be tracking instead?

Russell: The number of agents deployed across an enterprise is not the measure of success. I've come to divide the world into two categories, and AWS has reinforced this view: personal agents and professional agents. Personal agents help individuals drive their own productivity through everyday tasks, conversations, and workflows. Professional agents are enterprise-grade capabilities built by IT, with far more rigorous guardrails, governance, security and data connectivity. What matters is the AI fluency of the workforce.

The traditional measure of adoption asks a simple question: we deployed it, so now are people using it? You track it, put it in a scorecard, and report who's using it and who isn't. That isn't fluency. Fluency asks how people are using it and whether it’s making them more effective and more efficient. That's a fundamentally different question from volumetric adoption. One belongs to the traditional world of software deployment. The other belongs to an AI-native world.

Jose: Fluency means you are proficient not just in using the tool daily, but in executing work you would have done anyway, only with meaningfully better results. Take an account manager using Quick to transform how they approach account planning or sales planning, and then how they feed those insights into their CRM. That's fluency, where people have genuinely changed how they work, not just added a new tab to their browser.

Logging in is not a measure of success. What matters is how much more people can do, and how differently they can do it. We see this across healthcare, life sciences and financial services. Customers who track how Quick changes their day-to-day and how much time it returns to their people see the difference compound at scale across hundreds of thousands of users. The difference is not just measurable — it's something people actually feel in their day-to-day work.

Q: Is this shift creating new kinds of roles inside DXC?

Russell: Absolutely. There are new roles emerging across every function that simply didn't exist before. I have business partners in finance and HR who used to be deep in technology, reporting and analytics, who now partner with me directly on AI. That's driving a real decentralization of decision-making power, which is exactly what this capability is built to support. New roles are appearing inside IT itself, around governance, guardrails and gateways.

To give an example, we have an agent that handles resource requests for our Resource Management teams and another that does lead routing for Sales. Those didn’t exist a year ago, and they’re already saving teams hours of personal productivity that used to be entirely manual.

The analytic capability now available to people in finance and other functions goes far beyond what anyone thought possible even a few years ago. But none of it compounds unless people actually engage with it, which is why fluency matters more than adoption. The organizations that will pull ahead are the ones building it systematically, at every level, right now.

Q: What does success look like when all of those pieces come together?

Jose: True success for a company with AI comes when you have a tool that really works and scales, you have change management in place, and you are able to roll this out and manage that deployment in a way that is compliant with everything that the company wants. This partnership across DXC and AWS has really helped us build a product that has resonated with customers and that we are able to roll out at that scale. The learnings we've had across Amazon and DXC over the last few months have been immense, and we are taking that to build the product and roll it out across our customers.

Q: If you had to leave enterprise leaders with a few key takeaways, what would they be?

Russell: Four things. First, you need an exponential framework to harness the full value from AI. Second, drive fluency over adoption — it changes what you measure and how you act. Third, scale doesn't matter if no one actually uses what you've built. And fourth, AI supports decision-making. It doesn't replace it.

The capabilities we keep gaining, and the speed at which we're gaining them, is a direct result of how this partnership is structured. Co-innovation at this level is not common. When it works, you feel it.