Microsoft Part 2: How AI Agents Deliver Faster, Smarter Support for Frontline Teams

0:04Irina Parsina:Welcome back to Voices from the Frontline, AI in action in partnership with Microsoft. I'm Ron Thurston, and in this conversation, we explore one of the biggest shifts happening in frontline work today. The expectation of instant answers, real-time support, and information right at your fingertips. I'm joined again by Abby Sweeney and Irina Parsina from Microsoft's engineering teams. And together we break down what this means in practice from fast co-pilot powered product answers to the rise of AI agents that deliver speed, confidence, and consistency on the front line. So let's dive in. We've kind of developed this expectation of easy access and answers to everything at our fingertips.

0:54Irina Parsina:Our watch, our glasses, our phone, we have instant access to information. And if you are a patient or a customer or you're in any business where there's frontline teams, I believe there is now an expectation that they have equal or greater access to everything. That's a customer expectation baseline that has really changed. Absolutely.

1:17Abbie Sweeney:And I think if I can add one tiny little bit onto that, it's, you know, if I can, exactly as you're saying, getting access to information quickly, if I can be, if I'm a customer and I'm asking, hey, do you have this in stock? And they don't have to go back to the back office and go onto the computer and find the information manually. If they can just put into their co-pilot on their device, do we have this shoe in stock? And I can get an answer within two seconds. Amazing. Even better if they can say, want me to order it and send it home. Amazing. So I think, you know, really making those processes efficient is kind of what Copilot is about.

1:52Irina Parsina:Thank you so much. And the other hot topic that I'm going to bounce to you, Irina, is around agents. Agentic AI agents. These are the buzzwords as we end 2025. So help us define it, understand it. And then how is this showing up in front lines in multiple industries? Yeah.

2:16Abbie Sweeney:Yeah, absolutely. I think some of the examples that Abby had shared around maybe the broader support that you can get with generative AI. So if I have maybe a photo of something, customers coming in with a product that says, I'm looking for something similar. My child loves this, but you had it in stock five seasons back. Can I have something similar? And if I'm not, let's say, from the same department, I can take a photo of it and I can sort of find similar looking products or maybe help with some additional things like translation. But these are sort of broad skills that everybody can really benefit from in every workflow.

2:58Abbie Sweeney:So the way I think of agents is not so broad. It's actually bringing that goodness, those skills, the things that you get from these broader outcomes that you're looking for, but putting it into a little package with specific systems, with specific skills that you really want to be prioritized. Is it understanding a lot of text? Is it translating? Is it maybe bringing in something that is very specific to my organization or to the customers that the employee is serving or to the outcome that they're looking to achieve? Putting some guardrails and some logic around it. So it's really sort of packaging it up so that you get this almost outcome goal oriented specificity in action.

3:44Abbie Sweeney:So it's the buzzword in action. It's phenomenal because it's got sort of the responsible aspects to it. I can ask questions and, you know, as an organization, I can be very safe and secure in the knowledge that my employees being safe and secure, right? when they're reaching out and asking these broader questions. But they will need to be like an expert prompter in order to get the things that they need out of this. And that does happen. And we actually see a lot of really great examples of employees really embracing this, the prompt of thorns and being experts in prompting really well, asking the right questions to get the right answer.

4:25Abbie Sweeney:So the agent kind of helps you really deliver that faster so that an employee is not having to necessarily be the best prompt in the world in order to get the best outcomes. And you're enabling consistency in outcomes. So you're enabling consistency in how things are delivered and instead making sure that you're prioritizing specific systems, specific knowledge, specific skills, that you just want to be there at the fingertips of the employees that you're looking to enable. So that's kind of the idea of what an agent is.

4:56Irina Parsina:I'm referencing what Abby just said around easy access and speed. So what you're describing then is an even, if I'm understanding it correctly, a drilled down, even tighter version of that speed, correct? Yeah, absolutely.

5:12Abbie Sweeney:It's speed with now. I now have a map where I'm heading to. So I don't just have the speed on this incredible vehicle and engine, but I actually know exactly where I'm heading to. So I know where I need to stop and I know which passengers I need to pick up on the way. So it's that specific nature of really being able to define how this agent delivers the outcome, so helps the humans to deliver the outcome that they're looking for. I'm going to use this analogy because we started that with the car. You have like the car that I'm driving, then you have potentially a car that I'm driving and I have, you know, a load of passengers there.

5:49Abbie Sweeney:So that's sort of my team that I'm working and I'm driving. and I'm using the agent for my team. But then you might have some self-driving vehicles. You have autonomous agents that are doing things on our behalf and are transporting people where they need to go to. And we have examples of various levels of autonomy when it comes to building agents, depending on what function they serve. So one example, we have maybe like a pricing agent. And this pricing agent leverages generative AI to understand sort of the pricing strategy, then pulls into real data from ERP systems and understands exactly if something else is happening.

6:32Abbie Sweeney:Flights have been canceled. You know, these sort of signals are coming in and then it applies specific business rules that matter to our organization. So there's some specific business logic and rules that you can apply that really matter to you, that really reflect your brand, what you as a business are about. And then that can all be packaged up so that the employee that is maybe there actually acting on those pricing changes with that agent are able to do just that very quickly without necessarily having to deal with all of that complexity. And it can be connected into your point of sale systems.

7:09Abbie Sweeney:And so those are more complex examples, but this is where you have sort of the power of that agent being in there.

7:15Irina Parsina:So let's think a little bit deeper. So Abby, you know, a couple of the others here that are part of this around just co-pilot in OneNote and media analysis. So where does that now take this conversation and those tools? Yeah, really good question.

7:32Abbie Sweeney:I think this is around the different types of content and ways that we can interact with an agent. So we know that we have people are different types of learners, whether they're visual learners, whether they're audio learners, whether they like to do things, and that's how they learn. So we know that everyone is different and everyone likes to learn differently. And so being able to integrate things like voice and media into a platform like Copilot. So if let's say I am a mechanic and I am going to a store to fix the air conditioning unit, as an example, I might be up a ladder with tools in my hands and I don't have a third hand to be able to take out my cell and be able to read off the instructions or how I'm going to fix this air conditioning unit?

8:21Abbie Sweeney:What if I have a headset and it's voice enabled so I can say, hey, co-pilot, tell me how to fix this air conditioning unit. Or hey, agent, tell me how to fix this air conditioning unit. And it will then read out the policy or the instructions step by step in my ear and I'm just able to use my hands to actually fix the air conditioning unit. How awesome would that be? Or maybe it's that my first language is Spanish and I like to interact with all of my policies and my colleagues in Spanish. So if I could just speak into my agent and be able to, I won't do it in Spanish, but if I was able to ask a question in Spanish and it was able to respond in Spanish and maybe come back with an English translation that I was able to show a customer or something of those sorts so i think it's kind of about being able to have this multi-modal concept of um co-pilot and the platform so that whether it is um putting in pictures to ask for analysis whether it is using my voice to get um better answers whether it is being able to have an audio output so that i'm able to as i say follow steps without having to read it off of a cell and makes it really really valuable for you know any worker but very um

9:39Irina Parsina:specifically field workers because i also would add abby in that you know this idea and often in frontline is a multi-generational workforce so what what i'm immediately comes to mind is all the different learners based on you know different generations of maybe learning styles yeah speed the speed at which information comes might be different, the how it looks different, all of it is really interesting because I think for decades, we've, in front lines, we've tried to create content that was as generic as possible and hope that like 80 % of the audience and what you're saying is, I can actually create specific content that's just for you based on how you like to learn.

10:28Irina Parsina:Imagine that from an employee retention, employee productivity, customer experience, that's actually incredibly valuable. Thanks for joining us. And a special thank you to Abby and Irina for unpacking how AI is evolving from fast product lookup to intelligent agents and support that adapts to every learning style. What we're seeing isn't just better technology, It's frontline teams gaining speed, clarity, and confidence in every moment. Join us for the next episode as we continue to explore how AI is transforming the way people work, learn, and serve.