AI Copilot Podcast: Microsoft’s Ray Smith on the Potential of Technology, AI Agents
Welcome to the AI Copilot Podcast, analyzing the latest AI Copilot and agent developments. In this episode, John Siefert is joined by Ray Smith, VP of AI Agents at Microsoft. Smith will be taking part in the AI Copilot Summit on March 18th to 19th, specifically speaking at the kickoff keynote on March 18th in San Diego, California.
This episode is sponsored by the AI Copilot Summit NA, taking place March 17-19, 2025 in San Diego, California.
Highlights
Journey to Microsoft (01:31)
Smith has been in the business apps and AI space for about 20 years. After years of conducting research and studying tech, he took a job at Accenture. After learning about big implementations, ROI, and driving efficiency hands-on in that job, he decided to set up a startup that was focused on data and AI. “I would call that my kind of cynical phase as it relates to the potential for AI to change the game,” Smith said. That business ultimately ended up being acquired by SAP, leading to him driving a sales cloud product. “I’ve always been in the CRM space, is the short version,” he noted.
He arrived at Microsoft five years ago to step into CRM as well as taking time working in supply chain sustainability. A year ago, he wrote a paper about how AI is not just a chat interface but how it’ll really change how business processes work. Now, he descirbes himself “almost like Ted Lasso, where I believe everything is going to be different and just the pace of disruption and innovation,” he said.
He shares his journey because customers sometimes feel like different technologies are “just another hype cycle.” But for Smith, “it’s like every day, I have the hairs on the back of my neck moment where this is going to change the game.” He believes this will be a fundamental disruption on how we, as humans, live and work.
Big Learnings From Early Career Stages (05:47)
Siefert shared parts of his journey into the tech world. He describes a sense of “naïve moments of inspiration” that he had in his early experiences, believing that what they were doing was changing the world. “It is sort of a magical moment when you’re in one of those types of environments because there’s elements of that,” Siefert said. Once you experience that, you can’t stop searching for those moments over and over again.
Starting from creating a paper to now his role inside a $3 trillion organization, Smith provided insight on big learnings that defined a new future that he experienced for the early stages of his career. He provides an analogy that demonstrates how, with the right funding, humankind has the ability to overcome so many challenges. “There’s just such an amount of funding going in” and that impacts networking, data centers, you name it…every part is going through some major, major breakthroughs – not just the models,” Smith stated. “Everything is getting changed.”
The Potential of Technology (10:00)
Taking AI plus cloud and imagining what’s possible, you can recognize that this is one of the first times in the technology world where it starts to look a little infinite, with how many things are possible to reimagine how industries work, how businesses work, how processes run, how people do their jobs, and more, Siefert described.
Smith shared, “I’ve always been one of those dreamer guys, like ‘What if you imagine the future?’” He’s had someone saying, “Ray, back down to earth…this is what the technology is possible [of doing].” There are so many engineers and professionals in the field that will limit your imagination and the vision of what’s possible. “I think this shift is towards obviously agents but agents really will just become the new apps, the new solutions,” he explained. This is “really just opening the canvas for so many to get as creative, as inventive as possible, that really does challenge the traditional apps, the traditional industries.”
AI in Action (13:53)
Siefert asked about AI use cases Smith has seen. Smith said organizations often outsource mundane business processes like credit collections, invoice processing, and insurance claims. So, notable AI use cases straight out of the gate can be found in things like claims processing — anything that involves either lots of documents or research or some cognitive reasoning over variables.
Some of the earliest use cases were also almost software or sales development reps, where inbound leads are being processed — whether from a form or an email — based on the product of interest, engaging on that topic, offering product support, and then maybe setting an appointment for an account executive, and so on.
Many of the earliest use cases stemmed from what would have been early bots on websites, internal help desks, or B2E (business-to-employee) customer support. Then it shifted to real, autonomous back-office work or assistant-like behavior then back to customers.
Smith sees the most success when someone tries to do one thing and does it well. The biggest mistakes are when early customers want to build an agent that can run an entire department. The ones that surprised him the most were in complex areas like supply chain or M&A.
Unified Agents, Streamlined Interfaces (16:56)
There’s the potential of stitching different agents together — to have them almost networked —where they’re all managing individual processes in parallel. People often ask, “Is this going to be fully autonomous?” The answer is yes, but there’ll always be humans in the loop so they can oversee the agents. Humans who now perform a lot of these mundane tasks will likely shift to overseeing these agents, managing them to accomplish these long-running or complex tasks. No single agent will accomplish these processes by itself. They’ll be broken up atomically, like functions in an organization. Many agents will be stitched together.
Finally, what’s the interface back to humans? Humans want a reduced number of “panes of glass” to get work done, so ideally, it’s reduced to a single interface. This will elevate the job function, allowing users to manage tasks and improve outcomes effectively.
AI’s Rapid Shift (20:03)
The future of business is exciting, as no two organizations are identical. They have horizontal functions, like finance and HR, and vertical business lines. The integration of AI agents can enhance operations, enabling a reimagined structure. Siefert is eager to see new developments and insights at the AI Copilot Summit in March.
How do we start interacting with humans? How do we manage agents? Smith said that at the top layer, he thinks we’ll see an explosion in the same way we saw an explosion of apps when the iPhone came out. These apps are just going to be agents, and they’re going to be agents everywhere.
Every business is unique. Every business has different nuances and processes, and that has always presented a challenge for app and software-as-a-service (SaaS ) builders like Smith in the past, because it’s like, “I can’t get one size fits all.” And then the SIs sit and say, “Okay, let us make this extensible and configure it in this new world.” It’s as simple as saying, “Don’t do this for these rules; I want you to do something different.” Those few simple, declarative statements mean that the agent will take that guidance and behave differently. Think about how easy that is to affect that app layer.
Smith guaranteed there’ll be lots more innovation by the time March comes around.
The Future of Business (24:05)
Siefert underscored how we live in an exciting time where businesses can achieve unique personalization and customization effortlessly. This innovative approach enhances ease of use and execution, promising great discussions and insights when they gather in March. It’s a fantastic opportunity for exploration.
Even in the consumer world, there will be a personal assistant that remembers birthdays, user preferences, and memories. Smith described an assistant that will cultivate a tailored experience when one searches or shops, for instance. In the consumer world, these assistants will redefine search and retail and then bridge into the enterprise. Instead of bring-your-own-device to work, it will be bring-your-own-assistant. The future is bright.
