Tom Arbuthnot: Hi everybody. Welcome to this expert briefing. We are talking about the topic of the moment, uh, definitely the topic of the, the first half of 2025, and it feels like, uh, continuing to be the topic for the, the rest of the year in the new Microsoft FY as well. We're getting into agents and agentic and.
What does it mean? And, and particularly Microsoft have some really powerful agents in the box. So we're gonna focus on some of those and some use cases, uh, familiar faces in the community. We've got a and Chris, uh, for the, the couple of people that may not know you guys a do you wanna introduce yourself?

Alev Tamer: Yes. Hi everyone. My name is Alev Tamer. I am a senior Copilot technical specialist based in the uk and I am aligned with most of the industrial and pharmaceutical customers. Chris. 

Chris Wheeler: Yeah. Hi everyone. Chris Wheeler. I am also another Copilot technical specialist aligned to public safety and national security in our public sector space.
Tom Arbuthnot: Awesome. And, and, and agents really has been the, the conversation at the moment. Maybe we can start with like just defining, we've got AI and we've got Copilot, and now we've got agents. How do you define agents versus Copilot? What's the difference? 

Alev Tamer: I would define them as my digital colleagues, so it's just.
Having a set of information, which is different than graph API, which is Microsoft 365 ecosystem when it comes to data just to interact either as a chat bot, which is we can create with, uh, create agents as the data agents, but also there are some autonomous agents can act or create tasks or automate mundane tasks.
So I just define them. Like this, so, so 

Tom Arbuthnot: it's by your, you are talking about by use case. So it could be a scoped chat conversation. Mm-hmm. Uh, or it could be a more autonomous or semi-autonomous. It goes and does a thing for me potentially. Yes. 
Chris Wheeler: I think what makes it really special is it's all specialized and instructed in the way that we work.
So obviously in terms of Copilot Microsoft 365 Copilot, we've made it as, um, agnostic as possible regardless of what industry you work in. But if there's it, the board is possible 
Tom Arbuthnot: use case, right? Exactly. It's a giant LLM that's sucked up all this information, all this data, and it can have a go at. Any anything really.
Exactly. And 

Chris Wheeler: it still does a really good job. But what we're saying is agents is just that next step in that evolution of more specialized tasks, more specific outcomes. Uh, for, for us as an organization, you know, what is it that we want to achieve? What instructions do we need to ingrain in a specific agent?
What knowledge does it need? What, uh, connectors does it need? So this is where it gets a lot more customized, and I think this is where it gets really creative and some of the, um, some of the new agents that are coming outta the woodwork are really, really enhanced and really making a huge impact in our, in our customer base.

Tom Arbuthnot: Awesome. And, and then the other kind of piece of that puzzle is reasoning and reasoning models and kind of multiple turn type scenarios. So we've just recently as a recording seen Researcher, Analysts come in, which are the first kind of reasoning agents into the M365 stack as well, which is really exciting.
Alev Tamer: Yes, so we are releasing prebuilt or what we called first part agents as Microsoft, and they will be included in Microsoft 365. Copilot like Facilitator, ai, interpreter skills, employees have service. Um. Project manager. So Copilot will basically help you co-manage your project because we are not trying to remove the human element.
But the two agents are groundbreaking. 
Yeah. Researcher and Analyst. 

Tom Arbuthnot: Yeah. And I, I think that's something that the whole industry's been excited about because you get, rather than you kind of going back forward, back forward, you can kind of set it some homework. So go off do some searching, work things out, think things through, come back to me with an answer.
Chris Wheeler: Mm-hmm. Yeah. The quality's massively increased from the, uh, from the 4o models, even though 4o is still really good. It really depends on what you're trying to achieve. So, I mean, the, the clues in the name Researcher by itself to go off and Researcher and check his findings and go back and forth, like you said, it's just, you know, does such a good job.
And the accuracy and the, the real depth of the information is just really. Astounding. And um, just, just on that subject as well, obviously the models that are driving this as well, we're also making those available if you wanted to build your own agents, if you wanted to use those same models. So it's not just we we're creating our own and that's it, we're, we're giving that capability to our customers as well.

Tom Arbuthnot: Yeah, I think the, the, the, the prebuilt or the, uh, your first party models and or agents are a great example of like, okay, Researcher goes and does this, but I could, theoretically, I could build an agent that does the similar level of searching and reasoning and diligence over a different data set. And potentially, obviously with the. Microsoft 365 stack over your own private corporate data 'because it's all skewed and controlled. So that's really interesting. Is the ability to take this kind of power against your corporate data as well. 

Alev Tamer: That's what the first that from Copilot chat work or web and the great thing, you'll be able to use Researcher in Microsoft 365 apps.
So when you are trying to create a documentation or a PowerPoint presentation, you'll have the capability to select Researcher so you don't have to. It's separate to the Copilot 
Tom Arbuthnot: interface as well. Well, maybe we could start Researcher, actually. I know you've got an example of one you've done before, but while I pulls up, Chris, maybe you could talk, set the scene.
What is Researcher, what's it for? Why would I go to Researcher agent rather than just Copilot? 

Chris Wheeler: Yeah, so Researcher in a nutshell is our deep reasoning agent that's built on oh three, uh, by, by, uh, OpenAI. So it's, uh, a more enhanced, uh, large language model that's designed to be really, um, iterative to go back and fact check itself.
Where, where previously like models have been like a, a, a question and answer, so it's like a one time. Uh, round trip. This one actually goes through a load of sub layers of processing to make sure that it is getting the right answers. Maybe it needs to go and question those answers as well to get even, uh, an even better response as well.
So we're all about kind of the accuracy, but the intensity of the data and the output that's generating for us. And then what you'll notice is when you use it instead of just like a couple of paragraphs, this is so well structured and so. It's, it's, it is deep, man. It's, it's, it's like really, like it's, it's intense, but in the sense that, you know, some of the things that it comes up with is just phenomenal.
Just how, as if you were talking to someone with like a PhD in a specialized subject, um, the way it breaks down certain topics and structure and it goes back. And the other thing as well, I really like about Researcher is it goes and you can actually see what it's reasoning over. So it actually tells you its logical steps of thinking before it produces the output as well.
So you can actually go through that with, with the Researcher agent as well. So, yeah, see where it's 

Tom Arbuthnot: gone in its thinking, you know, just to use a human term. Like, like I, it's, it's, it's gone down this road or it's gone down that road. And that can be really interesting as well because that can teach you about.
Oh, I might need to change my prompt slightly. Or, you know, like, uh, or it finds out a whole bunch of things that, that you can then go down the road 

Alev Tamer: about that it'll actually help you enhance your prompts with a button. 

Tom Arbuthnot: Oh, nice. Oh, a, why don't you bring up your example? 'cause that that will help bring it to life for people.

Alev Tamer: Of course, let me share my screen. So this is one of the examples. So these are the topics that we are going to discuss today. And Researcher will help everyone because sometimes we do spend hours and hours amount of time just to do some research about new specific things. And I dunno, Tom, you are one of the experts in this, UC world, now it's just the Copilot world. But think about keeping up with all of the recent announcements with Microsoft Build with Microsoft technology. 

Tom Arbuthnot: Yeah, I use it a lot. I, I like invo com off the back of Incom I'll be like, find me all the Teams related announcements and news. Provide me a summary.
Um, that things like that is super useful for me 'cause it will diligently go and do. All particularly reasoning over the web content, all the search stuff, which is really useful. 

Alev Tamer: And now it's also bringing your work content. So I just ask you to create me a talking tracks to cover these topics for our podcasts.
And I just list all the topics in the email and. Researcher will gimme some different recommendations. Do you have any specific examples or scenarios you want to include for each agent? I can just ask it to, oh, please go through the web and gimme the specific use cases for a specific industry if I'm speaking with one of my customers.

Tom Arbuthnot: Ah, so this is, so this is what you're talking about when you're talking about enhancing the prompt, like Researcher is trying to get you to do the best it can do there. 

Alev Tamer: No. So Enhancing Prompts is another button that I can maybe show you with an example. 

Tom Arbuthnot: Oh, yeah, let's see it. Yeah, yeah. Sorry. Let's go through this example first then, 

Alev Tamer: so I can just ask it to be specific for a specific, uh, use case or topic.
Or I can just go ahead and draft the talking track and then I can see. An incredible level of steps and detailed, it's actually taken. 

Tom Arbuthnot: And the big difference here, you are showing an example where this has already run through it. It can take Yes. 5, 10, 15 minutes to gather this data. So whereas that, that first gen of, uh, LLM, like the 4o type model was all about fast response.
Like we were like answer question, answer question. Mm-hmm. This, you're giving the, the machine, the LM time. To, to reason to do the searches, to cross reference all the data, and that's where you get this, this, as Chris said, this really deeper result. 

Alev Tamer: That's correct, and sometimes I, I was literally impressed because sometimes it was just gathering information and it was actually, I can see, oh no, this information is irrelevant.
Let me check and go through this website. Oh, yes, this might be a useful one. Yeah. It's, 

Tom Arbuthnot: I, I, I do a lot of, we do a lot of obviously, research on Empowering.Cloud and it's like comparing teams, rooms, bars or whatever. And you'll be like, find the number of screens it supports and you'll see it do variants of different queries.
It's like, oh, I, I screen didn't work. I'll try monitor. Monitor didn't work. I'll try display. Like it's really trying to get the best result it can. 

Alev Tamer: And now everybody's talking about, yes, we need AI in our day-to-day work, but it's a huge investment. How can AI help with my business goals or my key KPIs, or how can I actually get the actual return of investment using generative AI technologies?
It can be with Microsoft, it can be with others. Now I can actually, so some of time when I'm doing live, live demonstration to our customers C levels, I'm just. Hi Copilot or hi Researcher, please act as my strategical Researcher and just go through all the web, all the Microsoft build announcement and create me a report where you can actually help.
We just have return of investment. 

Tom Arbuthnot: Yeah. And, and the big thing, you mentioned it earlier, there is like other, other LLMs have had reasoning models. So obviously chat gpt has got it. Gemini's got it. But the big difference here is this is all baked into Copilot. So it's, it can access, I. Your work data subject to you allowing it in the permissions and everything else.
But if I'm, like, for example, I'm a, a product manager and I look after a product, it's got access to the, the PowerPoints about the product. It can start to reason over the internal data as well as the web data. 

Chris Wheeler: The other thing as well. Yeah, sorry. The other thing as well, which is quite interesting is that obviously now this is, that this is included in your Microsoft 365 Copilot license.
So there's a huge amount of value compared to if you start using those separately with third party, then the cost of actually using a deep reasoning model is quite substantial. So what we've done is we, we've set. The, um, the, the fair usage for using these agents, but well within the actual budget of the actual license cost of Microsoft 365.
So there's no additional cost of using these agents as well, as well as obviously being grounded in the data, and obviously that's where the benefit is as well. So all the graph data. 

Tom Arbuthnot: Yeah, at time of recording we get, uh, I think 25 is the current public number for a number of deep research queries you can do within the M365 Copilot license.
So these are quite computationally expensive, and all the other models have some kind of limit as well. So you've got in quite a lot in your box for, for your 30 bucks there. 

Chris Wheeler: It's still early days in that regard because obviously like this is a new model and you know, over time it should come down. So it should be quite a lot more, um, you know, cost effective or processing effective to, to do that.
So I guess like if you have a look inside of, say, like Copilot studio or even Azure. AI, you can see those models and the deeper reasoning in regards to like the number of messages or tokens it consumes, it's obviously a lot more than just using your standard GPT-4 or, you know, 4o models basically.
So we're still at that high precipice, but for us to include it in that per user, per month license at and at 25 um, attempts or 25 interactions, um, it's, it's highly beneficial without having to pay any more. So yeah, there's a lot of value in there. 
Tom Arbuthnot: Yeah. And obviously when you go into the Copilot studio mode, you can.
Have as many as you want. 'cause you're paying on a consumption basis in that scenario. Right? 

Chris Wheeler: Yeah, no, absolutely. Yeah. No Copilot studio you could just go above and beyond, you know, 'cause you're talking about loads of other things like patching into different topics, different systems, different knowledge sources.
Um, we've been doing a lot of work with like multi-agent functionality as well, so that's where like agents can talk to each other. We've obviously got things like MCP, which is gonna be a great tool for patching into different systems and knowledge sources as well. So, yeah, 

Tom Arbuthnot: I feel a second briefing coming on order.
Chris, that's a whole, there's a whole me up order up. 

Chris Wheeler: Sign me up. 

Tom Arbuthnot: Yeah, yeah, yeah, yeah. I mean, it's a whole, I think, and this, this one, we'll get a really good foundation of what Microsoft's doing with it, but we should talk about that stuff because then, then you are able to, if you think through what we're showing here, imagine applying this kind of. Logic, but to your specific use case? 

Chris Wheeler: Yeah, a development level, low code, pro code. Absolutely. 

Alev Tamer: And if you see this result, it's very professional. So it's not just baked into working formation, it's actually baked into Microsoft 365 ecosystem. So you can just click to open the word and the word documentations will be available without doing copy and pasting.
And the great news is you'll be able to see. Um, PowerPoint presentation. So this will actually create a PowerPoint presentation. 

Tom Arbuthnot: Oh, interesting. That's new to me. I haven't tried that yet. That's exciting. 

Alev Tamer: And also if you like, listen, in podcasts you'll be able to create an audio overview from the work documentation.
So these kind of things are really, really. It's just, it'll change how you work. 

Tom Arbuthnot: Yeah, I'm really excited about audio summaries. You've got that coming to meeting transcripts as well. So they're being able to pick multiple meetings to generate a summary. But actually a Researcher report plus an audio summary is a really great personal workflow for me 'cause I've got about a half hour drive to the office.
It's like that's a perfect window for like gen me up on the summary of that Researcher will be really useful. 
Alev Tamer: Support enhancing prompts. You will have the buttons to enhance your prompts.
Tom Arbuthnot: Oh, interesting. So that's running your prompt probably through a a a a four hour similar model tuning it up, ready to run your Researcher report. Yes. And 
Alev Tamer: this will be available in all of the prompts windows. So it's not just specific for, I can just regenerate. 
Chris Wheeler: Yeah. We, we, we had some, uh, previous out of the box agents, so you would've seen things like writing coach, prompt coach.
So very similar sort of process, but just instead of going to that agent, this is just a click button. 
Tom Arbuthnot: Yeah, you bring it right into the UI. That's wicked Awesome. Anything else on Researcher or should we jump into Analyst 
now? I 
Chris Wheeler: don't think so. No. I think Researcher is, um. Yeah, it's, it's gonna be, um, yeah, it's gonna be extremely impactful I think, in terms of, you know, where we position it as well.
Um, any new changes, obviously we'll communicate that out as well. There's, there's a few things that we're thinking about in terms of, uh, future roadmap, but, um, yeah, for the moment using those O three models, it's, um, yeah, it's definitely a, a step forward, I think. 
Tom Arbuthnot: Nice. Okay, so Analyst, so this is a different agent.
So what's the proposed use case for, for Analyst agent? 
Alev Tamer: So Analyst agent, we integrated with Python, so it'll be able to just give really advanced data insights within complex Excel spreadsheets as well as other data sets.
And just I can use a prompt. Create a table with the volume of planets at a column to show the volume is multiple of earths and a road, um, to show how the sun compares to the volume of the sun. So like you see, you'll be able to see the advanced reasoning capabilities, and the great thing is you don't have to know anything about.
It's just you using your natural language to ask anything, create, uh, ask it to create bar charts, diagrams, or different things. If you are in the sales, you'll be able to give it the last quarter, three quarter sales results, and you'll be able to ask it. Please create me a forecast for Q4. What are the things that I can do to improve my sales pipeline?
Tom Arbuthnot: That's awesome. So this is like a Excel expert on demand. So normally you would've gone to that person who can really be, be a wizard with the data sets, but now you're getting that, that kind of capability as an agent. Yes. 
Alev Tamer: Um. If I think about myself, I can only use maybe 25% of Excel's capabilities. 
Tom Arbuthnot: 25 is high Alev that.
Excel is a beast. I mean, I'm impressed at 25, I'd say I'm probably at the like five, and even that's me asking Copilot how to do it,
Alev Tamer: and it just gives you diagram or cool bar charts that you can copy and paste into a PowerPoint presentation. 
Tom Arbuthnot: So, where's the line here? Because obviously we've got Copilot in Excel as well. So do I start at, uh, here with Analyst agent or am I in Excel? What, what, how would you explain that? I, 
Alev Tamer: so if you are working on an Excel spreadsheet, it's better to use Copilot in Excel.
But if you're planning to work on different kind of data sets, different or multiple spreadsheets, I think on the list might be better to use. But Chris. 
Chris Wheeler: No, I, I completely agree. I think in terms of where we first started with Copilot in Excel, um, you know, within just say like one set spreadsheet with structured data, say like in a, in a table format, um, you know, for the first like year or so, it was great at doing basic things, which is fine.
And then as you saw, we started adding more advanced analytics in there, but still just in the body of that one spreadsheet, um, I think where Analysts really kind of goes. A way of a step further is obviously that once again, there's that deeper reasoning and the advanced like logic in what, what that data's representing and especially different data sources.
Um, and probably the more complex calculations that need to happen. So, you know, once again, like the Analyst, same as Researcher, will go and check its findings. It will basically do a lot more deeper diving into that data and understand it a lot better. Um, so yeah, the, the, the Excel in, um, the Copilot in Excel, it's kind of.
If you're just working on one set of data and it's quite, um, say it's, it's quite a standard format, then yeah, absolutely. But as you start going more into like the data scientist sort of roots, more complex structures, more, you know, bespoke formulas, then I think Analysts would probably win hands down in that regard.
Tom Arbuthnot: Okay. That's good that, and can I put multiple files into Analyst? So like you said, you mentioned that I'm already in workbook there. Potentially I could put different, different combinations of workbooks or a workbook and a PDF potentially. 
Alev Tamer: Yes, you'll be able to select people, files, meetings, as well as emails, and you'll be able to use or upload from your bond drive as well as your local disc.
Tom Arbuthnot: Nice. The meetings one's really powerful as well. Like being able to pull in, uh, like we talked about what we're trying to achieve here as a transcript is interesting. Nice. So that's, that's Analyst. And I think for me that's a, a, like I feel like a Researcher is a really broad use case. Analyst is a bit more like I.
I work with data, um, to kind of switch it up completely. Then we've got interpreter agent, which is really interesting because this is the first agent we're gonna talk about. That's, uh, dealing with like audio, video, and obviously teams, which is close to my heart. So, uh, how would you explain interpreter agent?
Chris Wheeler: I think it's like. We, we, we've had translation services for a very, very long time. You know, we, we've done like, you know, speech to text, we've got live, you know, translation. We, we've had that through like Bing for quite a while as well. And, um, this is kinda like the, the, the one agent where you're not essentially.
You're not interacting with it. It's kind of just like building on what you are already contributing to. So it's, it's, it's pretty much an autonomous agent. You don't really need to do anything other than talk. So this is kind of where we've taken that one step further in that translation service where not only will it translate in real time.
You know, whatever you are saying into another language, but also it replicates your spoken tone and your voice as well, which is like an even better form of like understanding and more, um, you know, building on that, um, inclusivity and that connection as well. So even though I'm not speaking, say like, you know, Alev's Turkish, if I'm not speaking Turkish, then I don't want to hear like a separate.
bot speaking in like, you know, that sort of sense anymore, which, you know, we, we've had previously and it's, it's nothing wrong with it, but what, what we wanna do is just add a bit more personalization to that translation where it sounds like I'm actually speaking Turkish or like, say like if, if Ali is speaking in Turkish.
It translates to English. It's like, wow, okay, this is phenomenal. Because I don't have to have to listen out to something else talking to me, I'm, it's like Alev speaking straight to me. So there's some really profound impact to actually being there in the moment and actively contributing rather than having to read a tra a transcript at the bottom or waiting for a separate bot service to come in where, or you know, previously we had things like interpreters where you had to look at them and they had to translate for you, and then you had to talk and engage.
And this just kind of bridges that gap really effectively. I. Um, so yeah. Alev is there anything else on that one? I know we've been using it quite a lot. 
Alev Tamer: Yeah. I, I am a fan Facilitator because we introduced Copilot as like the personal assistant, and then we, I. Involved and evolved it to become a team, a assistant, because it's now going into meetings and starts generating tasks or start generating meeting notes without you even asking for it.
The great thing is it only generates work related things. So if you are speaking about a movie in the beginning of the meeting, it'll left them out and, uh, organized. 
Tom Arbuthnot: So this is Facilitator, isn't it? So this is Facilitator. Yeah. Yeah, so, so this is moving into, if interpreter is language translation and it goes back and forward and just, just to round out interpreter, the person who enables that is the, is is speaking and having, or how should I say, the person enabling that is choosing the language they want to receive the communications in?
Is that right? 
Chris Wheeler: Yeah, so the, so the organizer picks the language that's being spoken to, and then obviously the people receiving will select their language. They want to hear it in as well. So, and also as well as that, it's not just one single language translation. You can have multiple going on at the same time.
So you can have French, German, Spanish, like all these different translations happening simultaneously. So whoever's listening in will be able to hear it in their language, which I think is quite. Very, very impactful. 
Tom Arbuthnot: Big for accessibility, isn't it? Where you're like, uh, or like, like most of the orgs I work with are multinational, but English is the preferred language for it.
But actually I can, I, I can choose to flip it into my local language. I would prefer just 'cause I find that easier to, to keep, keep up. Yeah. 
Chris Wheeler: Such an easier mental burden. 'cause like Alev, I'm guessing it's just such a, I mean, my wife's polish, so it's just like having to have that mental, uh, capacity to translate in real time in your head.
And then obviously you got them like speaking English and stuff. So just natively, it just makes so much more sense and less, less mental strain as well, I think. 
Alev Tamer: Yes, it's, so it's, I think that's one of the thing that will require some time to adjust, because if you enable a interpreter for yourself, you'll also hear your own voice.
So it'll help when to start speaking, when to stop speaking, but also it'll replicate your voice without you even actually recording your voice. It's really great to hear your own voice, but it definitely makes you more engaged with the conversation. You'll be able to select up to nine, nine different languages and that will remove some of the maybe fractured conversation because you'll be able to use your own.
Yeah. 
Tom Arbuthnot: And there is, there is some interpretation lag there isn't there naturally, like you are, you are saying it and then the system is interpreted. So that's why it kind of gives you a playback. 'cause you're like, like, and it is visually when you turn on, it's visually in teams. Like someone's using interpreters.
That's your cue to be like, okay, let's leave a couple of seconds runoff at the end of each sentence. Yeah. For the different interpreters to hit. 
Chris Wheeler: So that performance improves if you do upload a voice profile. So what we're, what we've got is obviously, you know, we've got intelligent speaker, so if you've got your voice profile, if that's uploaded, then that actually helps with the performance of the interpreter agent as well because it has a knowledge source based on your previous, um, whatever it is you've recorded for yourself for that profile.
So, yeah, that we, we are making some improvements there as well. Yeah, definitely. 
Alev Tamer: And you'll see like an interpreter, um, sign on your video, so I know if I turn on the interpretation. 
Tom Arbuthnot: Nice. Alright, let's jump into Facilitator. So Facilitator is another one of these agents that's broadly applicable, right?
We're all in meetings we could all do with a bit more structure around the meeting and, and it's, it is doing meeting notes, but Copilot does meeting notes for us too. So what's the difference of Facilitator versus my Copilot? 
Alev Tamer: So we introduce Microsoft 365, Copilot as your personally assistant, but the agents like Facilitator.
We are actually evolving Copilot to become your team, a assistant. So during your team meetings, it'll start generating the notes automatically without you asking for it, it'll introduce itself. Why it's there and then it'll just start generating notes. It'll create a really good organized table if you have action items or follow up tasks.
And the final results if you are using pages is amazing. 
Tom Arbuthnot: And it's generating it for the whole team, isn't it? So we're all seeing the same notes in real time, and we can edit and add to those notes and that that's the big difference, is I don't come away with my Copilot summary and you come away, your Copilot summary.
We are all in real time. Agreed. Those are the minutes and the ability to correct them as you go is really important as well. 'cause AI is valuable. Yes. It's like, oh, it's not quite, got that right. We'll, change that is is really useful 
Alev Tamer: and you'll be able to see if that's not taken by Facilitator or your colleague or by you.
So those will be available and also it'll give you a halfway point and it'll let you know if your meeting is about the end. So it'll help you with time management. 
Tom Arbuthnot: It feels like we're on the beginning of a journey with Facilitator as well where it can potentially get. More engaged in, in the meeting.
There's been some stuff demoed by Ilya at Enterprise Connect, where we're talking about like potential of like understanding who's engaged, who's not engaged, like being able to see the room through the cameras and be like, oh, the room's not set up right. Um, it, it's like your kind of a, like a awesome PM slash Facilitator, as the name suggests, like available to everybody, whereas previously you'd only get those in certain meetings.
Alev Tamer: Yeah, during one of my meeting I asked Facilitator to assign me that task and 'cause I was using Planner, it created a task in Planner and I received an email by due date. 
Tom Arbuthnot: Oh, nice. Oh, that would be good. Just straight. Straight assignment as well. Awesome. Anything else that was worth talking about on Facilitator?
Again, that license wise, that's part of the Microsoft 365 Copilot, so you've got that license, you've got access to Facilitator as well. So those are some of the ones that I, I use fairly regularly or certainly, uh, Researcher and Facilitator, but there's others I'm less familiar with. So, um, where do we wanna go next?
A in terms of, uh, which one do we wanna talk about next? 
Alev Tamer: Uh, employee self service agents, which is,
Tom Arbuthnot: yeah, I have no idea. I've not used that. So let's get into that. 
Alev Tamer: One of the favorites agents. Within our, some of our customers, so employees have, service agents will be integrated within HR systems, IT systems, including ServiceNow, SAP, Workday.
So you will have one agent to rule them all when it comes to employees have service. So we are starting to release it by the end of this month, which is. Basically now it'll be generally available and admins will have the capability to use Copilot Studio to configure and integrate with different systems.
It can SharePoint sites, it can be third party integrations, and. We are releasing that as a chat bot at the moment, so I can ask, how can I connect to VPN at the same time? I can ask, what is the hybrid working in Microsoft uk? So. This is my platform to go enable, to help myself. 
Tom Arbuthnot: So rather than having to like troll the internet, uh, intranet rather, right.
Like we used to do, to try and find the right doc that had the, the right policy or the right help, it's starting to bring that to the user. 
Alev Tamer: Mm-hmm. And we are expecting it to be evolved to. Take actions, create tasks, um, maybe connecting with expense system as well as travel and vacation systems to just, I would like to book 14 days, starting from this day, will you please my vacations or do my expense on behalf of me?
Creates it tickets as well as creating. 
Tom Arbuthnot: So how's that one work? Because you're, we're talking about in the box agent, so that's in the box agent, but how do I feed it? The right information? 
Alev Tamer: So you'll be able to configure using Copilot studio. 
Tom Arbuthnot: Ah, okay. So this is where we start to dip into some configurations.
So everything we've talked up to now is, uh. Like a kind of end user facing bot. This is the first one which we met where it's a, an a native first party bot, but it does require some config to say, here's the data sets that we want to bring into this conversation. 
Chris Wheeler: Yeah, it needs those specific knowledge sources.
I mean, even if like, I think my favorite one I'm gonna use is probably how do I order a new laptop? Or can you order me a new, new laptop? So that, that'll be my next one because I think I'm getting close to three years now. Actually. I think I'm due one, so I'll be getting on that employee. Uh. Self-service.
The 
Tom Arbuthnot: new service arm ones are amazing. 
Chris Wheeler: Oh, Copilot plus pc. Yeah, and I'm still still waiting, but, uh, yeah, so things like that, it needs to follow, like, you know what the right structure is, you know, the ordering platforms, the systems. So each customer's gonna have different systems and different policies. So we need to tap into that basically.
So even though we'll make it really easy to interface with that employee, self-service agent, it just needs a be a configuration in the backend. 
Tom Arbuthnot: And the last one we wanna talk about was skills agent. Another one I'm not so familiar with. So what skills agent? 
Chris Wheeler: In, in a nutshell, it's, it's like we're, we're all about connecting people.
It's all about empowerment. Um, we've had things like LinkedIn skills for quite a while. We know like what people's skills are based on things like, you know, standard things like your resume, whatever you're putting onto that social media platform. Form. And what we're effectively doing is we've had that already inside of Entra ID as well, where we have that metadata where we can really start to encourage people to connect with other people based on their skills.
So what this agent is effectively gonna do is it's gonna tap into that information and help you discover like people with so set skills, it may be that you've taken on a new project and you need someone with dynamic skills. By interacting with the skills agent is able to uncover you. Like all of those people with said dynamic skills that they have.
Um, they, they have approved themselves. So you can, it's also in your control to say, I, I want people to know that I'm an Office 365 expert, or I'm an expert in Exchange or project management. I'm ITIL certified. 
Tom Arbuthnot: So, and this is within your organization? So this is bigger organization. Like I've jumped into a new project.
Who, who are the people that know about? That skillset set or I'm doing a new project. Who would that, that's really interesting. 
Chris Wheeler: Yeah, so for us at Microsoft, we, we love it because obviously Microsoft is such a massive global company. To find someone that works in like, or specializes in, like Bing for example, we could just interact with this instead of me having to search through account massive of emails or you trying to find people just, 
Tom Arbuthnot: you just put it in the Viva Engage and pray.
You're like, can someone help me with this? 
Chris Wheeler: Yeah, no, exactly. No, it's, it's really good. So it's all like connecting people through, just interacting with the agents to begin with rather than spending. 
Tom Arbuthnot: So how's it picking up who has what skills? How are we feeding that? Is that, uh, like from our profiles? Is that self-service?
How do we, how do we set it up? 
Chris Wheeler: Yeah. So a, I believe this is all based on, um, metadata in Entra ID. 
Alev Tamer: Yes, so we'll just have the new Microsoft 365 profile cards and card editor, and I'll be able to define my skillset, but also if you are using Viva learning, if you're, so we, we have access to all of these data as well as the organization explorer, people companion.
So skills agent will be able to use this.
Tom Arbuthnot: Oh, amazing. So if you've invested in some of that journey already, you're gonna be able to leverage that investment, which is nice. 
Alev Tamer: Yes. So one of the things that we hear about some of the, how can we improve employee experience? And so human resources is all about the employees.
Anyways, within the skill agent, the organization leaders can. Use the skills agent to just create informed strategical workforce planning because now, um, we're in a different age where we can have the skill gaps in between different generations. So if I would like to improve my soft skill about just speaking in a podcast or just presenting in a public environment, I'll be able to search that through my organization.
And in Microsoft, we run on. Coach, model, care. So this is our three pillars when it comes to being a leader. And if I would like to. We have more than 250,000 employees in the world. And if I would like to just have leaders, leadership, uh, skillset, I can actually go and search through my organization and find someone in the uk or I can just find in the us 
Tom Arbuthnot: Yeah.
It's amazing to be able to cross connect between different, different teams in different regions and, and they, and actually very often the person, the, the, the best person to ask is not necessarily gonna be in the same. Be your country as you necessarily, actually they might be in a different, a different country or a different uh, business.
Alev Tamer: And it's all about learning and growing into your day-to-day roles. And this is how AI can just help us. 
Tom Arbuthnot: Awesome. Right Alev, Chris. Amazing introduction into the kind of what's in the box. Uh, we'll do a part two where we'll get into Copilot studio and what you can take these engines and kind of take to the next level because I know both of you have been working with customers doing some really interesting kind of more line of business use cases.
So there's a whole another piece to this puzzle, which is these are almost like, uh, good in the box use cases, but I actually, I think you really unlock the power and you're like. Okay, like let's look at the business and see how we can leverage this same power to do line of business things.
Chris Wheeler: A hundred percent awesome.
Alev Tamer: I cannot remember the days without Microsoft 365 Copilot 
Tom Arbuthnot: Awesome. Well, thanks so much both of you. If anybody's got any questions, do drop them below and look out for part two where we talk about, uh, Copilot studio and building your own agents. Thanks a lot Alev Chris. Appreciate it. 
Chris Wheeler: Nice. Thanks Tom. Than