Thank you for joining me in this session of Global Perspectives. I'm delighted to be joined by the Executive Vice President and Chief Commercial Officer of Microsoft, Judson Althoff. Thank you for being here today.
It's my pleasure, Regina. How are you?
I'm doing well, but let's dive right in, shall we?
So you are in the heart of the rapid pace of technological change, and that technology is evolving to play a critical role in innovation and driving a competitive advantage. We're hearing from CEOs that AI is their main investment regardless of the economic environment. So what's your perspective on this, and what are you hearing from your clients and business leaders about how they're going to move into this world?
Last year, I think more of the technology actually became viable for people to use in a business context. But if you had a time-lapse camera and could capture CEO sentiment over that period, two years ago I think most folks said, 'Well, this is kind of cute. Maybe it can help me with my holiday cards, but it's probably not going to amount to anything in business.' A year later, it was a bit more somber: 'Are the AI robots coming to kill us? Help us, Microsoft.' What's great about this year is the hype cycle is dampening a bit. It's less about anxiety and speculation about capabilities and more about pragmatic business value conversations. The average company has at least a hundred big ideas of what they think they can get done with generative AI, and we find ourselves increasingly providing consultative guidance on where to apply the technology to actually get business value. For us, there are four pillars we consistently focus on in AI transformation. The first is enriching the employee experience. We fundamentally believe AI has to empower human achievement. There's so much you can do today to put AI in the hands of your people to make them more productive, enjoy their work, accelerate the learning curve, and make them more successful at achieving their potential. We have a solid methodology of how we take people through immersion and extensibility into data that matters to them, and then the real measurement of ROI on the investment. If CEOs are going to invest in this economic climate, it's got to make sense from a dollar standpoint. The second category is reinventing customer engagement. Generative AI agents and assistants have tremendous potential for enabling you to engage with your customer more actively, in a more personalized and unique way, and at scale. We're seeing fabulous results there. The third category I find really interesting is reshaping business processes altogether. A business process is an artifact of human workflow today. If we step back and say, 'What if we made it AI first? Where do humans come in to create unique potential within that process?' We're finding a lot of optimization. Think of this as lean for knowledge work—continuous improvement coupled with technology creating great business value. The final category is bending the curve on innovation. At Microsoft, this is a pretty obvious extension of generative AI because we're in the business of writing code. With all the code generation capabilities in GitHub Copilot, we've used generative AI to write over a third of the code of all our new products. It's enabled us to get more product to market over the last year than in the prior five. It's a big leap forward. But it's not just in IT or software development. It's also in a lot of industry scenarios, like healthcare, financial services, manufacturing, and even material research. There's a tremendous amount of potential. I think there's a reason why CEOs are saying, 'How do we put AI to work for competitive differentiation?' The most important thing is to stay locked on the pragmatic approach, because there will continue to be a litany of ideas, and most of them are not going to add business value. I like to use those four pillars as a checklist: is the scenario actually going to move the needle in one of those four categories? That's a good indicator of whether it makes sense to invest.
Those pillars are really well aligned with how we think about AI and the conversations we have with our clients. People at the heart of everything they're doing. It's clear the customer engagement and customer service side is where we're already seeing tremendous use cases. But I love your point about the redesign of the entire value chain, because that's where we see a lot of interoperability. We design workflows based on how humans get work done, but we can totally rethink how that works. And then lastly, the innovation—we've got some great use cases in personalized medicine, very specific to your own human DNA and diagnosis. So great way of thinking about it. Now, as you consult with your clients, they have these hundred ideas. How do you propose they get started? What does that roadmap look like when you're advising them?
It's a great question. The technology landscape and portfolio capabilities are evolving such that you have to have an applied approach. We tend to think of these things in three categories. First, increasingly focusing on AI business solutions—what can you achieve with a copilot, a set of agents, or assistants to reshape a business process? That applies across so many of the categories we just talked about. Second, having an AI platform inside your company so you can develop AI-native applications that are unique to your business and provide differentiation. Third, having a strong cybersecurity foundation. The one thing not slowing down is the rise of bad actors and threat vectors. The more we digitize and leverage AI, the more those vulnerabilities increase, so you have to be prepared. We walk customers methodically through those three technology lenses against the four business pillars to drive a path forward. Generally, there's a propensity for folks to want to dive in and start developing their own new things, especially if the CTO has been given a mandate by the CEO. But the reality is you can achieve so much with readily available capabilities. We encourage customers to start with Copilot—put it in the hands of their people, help them with immersive experiences so they learn to use an AI assistant in their life. Then extend it to the data sources that matter to them. If you run supply chain, connect your Copilot to your supply chain data so you can reason over it and interrogate it as if it were another team member. That Copilot stack is basically the UI for AI. We suggest going first with what you can achieve with Copilot and a product called Copilot Studio, which allows you to point it at different data sources to create a unique experience. After you've exhausted those scenarios, then you ask if you need to develop something custom. If so, we help customers build out a real AI platform with three assets: putting the right tools in front of developers, providing a real app platform (we just launched Foundry, an app server for AI-native applications that allows you to use multiple models, test them for groundedness and content safety, and get observability and ROI), and the data layer. We have a product called Fabric that allows you to reason over data sources in our cloud, AWS, GCP, or on-prem, with one common way of reasoning over all your data. On the cyber side, you need a comprehensive approach—zero trust environment, endpoint protection, a solid identity platform, and incident management. Keeping those three technology pillars in mind and mapping them back to the four business scenarios gives you a structured approach to pragmatically delivering AI solutions.
Sounds very comprehensive from a total tech stack and blueprint perspective. Maybe we could double-click on a couple of the topics you mentioned. I do want to double-click on risks, but let me save that for the second part. Let's stay on the positives and business optimism. I tend to land on the optimistic side of AI. The agent piece is quite interesting. For those of us who don't live and breathe technology, can you tell us a little more about what that means? I even heard somebody say 'nursing agent' because it used to be more back-office oriented, but now they're front-office oriented.
Agents have been the word of the year in AI. When the word started becoming of interest, I actually had a fundamental debate across our senior leadership team about whether this term makes people comfortable with adopting AI. But when you step away from the inherent fear, it's important to understand how these things compose across a business process. Depending on the vendor, some argue for AI assistants or full-blown autonomous agents. The reality is you're going to need these things working together in concert to provide real AI differentiation. Let me give you an example: a customer support scenario with inbound calls. You have expert resources to solve problems, but the volume of inbound versus available subject matter expertise is hard to plan for. By having an agent layer for triage—like your nursing example—the agent can determine if a problem is known with 99% confidence and provide the answer directly, without bothering expert resources. If it's a complicated problem with an important customer, it routes to the right subject matter expert, who is empowered with an AI assistant or Copilot to help solve the problem. Once solved, the Copilot handles the busy work of case management, updating the knowledge base. This enriches the knowledge base, scales customer engagements, and creates a flywheel where the agent gets smarter over time. This idea of agents, AI assistants, and human skill working together provides real differentiated AI. We're firm believers that this is the right approach.
