WEBVTT 00:00:00.000 --> 00:00:02.600 Hey, everyone. Today, I have a very special guest 00:00:02.600 --> 00:00:06.679 from Microsoft, Michael Goad, who is a systems 00:00:06.679 --> 00:00:09.439 or solutions engineer. I was going to say systems 00:00:09.439 --> 00:00:11.859 engineer. Yeah, I appreciate it. Thanks for having 00:00:11.859 --> 00:00:15.660 me here. Man, I remember I've been at Microsoft 00:00:15.660 --> 00:00:19.480 for a year and a half now, and I remember meeting 00:00:19.480 --> 00:00:22.359 you. uh, early on when I first joined Microsoft, 00:00:22.480 --> 00:00:23.980 cause you kind of reached out to me, noticed 00:00:23.980 --> 00:00:26.519 we all had a YouTube channel and, uh, kind of 00:00:26.519 --> 00:00:28.519 started talking about, uh, our YouTube channels 00:00:28.519 --> 00:00:30.579 and stuff like that. But I've been with Microsoft 00:00:30.579 --> 00:00:33.240 a year and a half. I'm a solution engineer, like 00:00:33.240 --> 00:00:35.979 you just mentioned. Uh, but I, uh, I met, I work 00:00:35.979 --> 00:00:38.100 on our HLS side. So I work with a lot of our 00:00:38.100 --> 00:00:39.880 healthcare providers and stuff like that. So 00:00:39.880 --> 00:00:42.219 got some experiences from that side of the street. 00:00:42.960 --> 00:00:46.280 That's awesome. So HLS health, what, what's the 00:00:46.280 --> 00:00:49.759 LS for just healthcare life sciences. I used 00:00:49.759 --> 00:00:51.740 to be with United Healthcare at one point in 00:00:51.740 --> 00:00:54.759 time. Now I'm with Blue Shield of California. 00:00:55.420 --> 00:00:59.359 Blue Shield, yeah. Okay, so you have customers 00:00:59.359 --> 00:01:01.740 like that basically, right? Yeah, absolutely. 00:01:02.079 --> 00:01:06.019 Generally, we talk about AI, AI in general. It 00:01:06.019 --> 00:01:08.879 can be co -pilot. It can be other vendors. It's 00:01:08.879 --> 00:01:12.120 open to whatever we want to speak about. You 00:01:12.120 --> 00:01:14.439 know, a lot of times when I use AI, I'm like, 00:01:14.500 --> 00:01:17.500 OK, well, it's giving me a good response or it's 00:01:17.500 --> 00:01:20.359 not giving me a good response. But like my what 00:01:20.359 --> 00:01:22.859 I do isn't like super mission critical. I'm not 00:01:22.859 --> 00:01:25.540 dealing with people's lives. I'm not I'm not 00:01:25.540 --> 00:01:27.760 a doctor. I don't have a patient. So when I'm 00:01:27.760 --> 00:01:29.859 thinking your field, I'm thinking like if I'm 00:01:29.859 --> 00:01:33.480 using AI, it's a lot more critical for it to 00:01:33.480 --> 00:01:36.620 give you good, accurate information. Is that 00:01:36.620 --> 00:01:38.620 something that your customers bring up when you 00:01:38.620 --> 00:01:41.650 speak about AI with them? Yeah, absolutely. I 00:01:41.650 --> 00:01:44.310 mean, healthcare is a regulated industry. So 00:01:44.310 --> 00:01:47.250 we have to factor in those security, you know, 00:01:47.250 --> 00:01:49.469 making sure everything your data is secure and 00:01:49.469 --> 00:01:52.829 encrypted and things like HIPAA compliance come 00:01:52.829 --> 00:01:56.030 into play. The thing that I, you know, when you 00:01:56.030 --> 00:01:58.810 kind of start working with AI and whatever platform 00:01:58.810 --> 00:02:00.909 you're looking at, especially in regulated industries 00:02:00.909 --> 00:02:02.430 like healthcare, you have to pay attention to 00:02:02.430 --> 00:02:04.549 like, how do they, you know, what are the security 00:02:04.549 --> 00:02:07.430 guardrails that can be set up? And how do I govern 00:02:07.430 --> 00:02:10.629 this from an administration side? I think, I 00:02:10.629 --> 00:02:12.650 mean, no matter healthcare anywhere, like everyone 00:02:12.650 --> 00:02:17.189 wants to use AI and sometimes the organization 00:02:17.189 --> 00:02:19.650 wants to let people use AI, but they're going 00:02:19.650 --> 00:02:21.169 a little slower than users want because they 00:02:21.169 --> 00:02:22.870 want to make sure that they put the right guardrails 00:02:22.870 --> 00:02:25.009 in place. So you start seeing shadow AI and all 00:02:25.009 --> 00:02:27.389 that kind of stuff. But, you know, healthcare 00:02:27.389 --> 00:02:30.250 is just like. At the end of the day, it's very 00:02:30.250 --> 00:02:32.830 similar to other business structures where, you 00:02:32.830 --> 00:02:34.750 know, we're setting up the right infrastructure, 00:02:34.949 --> 00:02:36.789 setting up the right governance controls to make 00:02:36.789 --> 00:02:39.789 sure that users can use this tool because we 00:02:39.789 --> 00:02:42.229 wanted them to take advantage of it. But we have 00:02:42.229 --> 00:02:44.030 to make sure that we have certain things set 00:02:44.030 --> 00:02:46.590 up appropriately to make sure that our data is 00:02:46.590 --> 00:02:48.750 secure and so forth. So just a little bit more 00:02:48.750 --> 00:02:50.870 when it comes to highly regulated institutions 00:02:50.870 --> 00:02:53.550 like health care. That makes sense. And, you 00:02:53.550 --> 00:02:55.229 know, when you, as you were speaking, I remember 00:02:55.229 --> 00:02:58.270 like HIPAA compliance is a big one, right? Yep. 00:02:58.449 --> 00:03:00.750 Absolutely. Yeah. That's a big one. The exchange 00:03:00.750 --> 00:03:04.330 of like information that might be personal, it's 00:03:04.330 --> 00:03:06.969 about the patient and it can only be in a very 00:03:06.969 --> 00:03:09.629 specific circle. Yeah. So there's that side of 00:03:09.629 --> 00:03:12.789 it. And then there is like just AI itself, right? 00:03:12.849 --> 00:03:15.750 I'm curious, your customers, again, being more 00:03:15.750 --> 00:03:18.719 on the healthcare side, like. How are they starting 00:03:18.719 --> 00:03:20.840 to even use it? You know, whether it's on the 00:03:20.840 --> 00:03:23.360 productivity side or whether it's on the agent 00:03:23.360 --> 00:03:26.300 side, like what are some ideas that are being 00:03:26.300 --> 00:03:28.120 shared with you about how they want to use it? 00:03:28.490 --> 00:03:30.590 I think a lot of users are still trying to figure 00:03:30.590 --> 00:03:33.110 out how to build their proper skills, understanding 00:03:33.110 --> 00:03:38.069 what is a good prompt, what comes into the different 00:03:38.069 --> 00:03:40.169 ingredients that make up a good prompt, and then 00:03:40.169 --> 00:03:42.650 others are trying to figure out things like agents 00:03:42.650 --> 00:03:45.909 and so forth. In healthcare, I've actually seen 00:03:45.909 --> 00:03:48.669 some really cool things with people kind of experimenting 00:03:48.669 --> 00:03:52.150 with agents. When an organization allows agent 00:03:52.150 --> 00:03:55.770 creation, especially within Copilot, it's really 00:03:55.770 --> 00:03:57.810 easy. You've done some really great videos on 00:03:57.810 --> 00:03:59.509 YouTube. YouTube, we're kind of showcasing how 00:03:59.509 --> 00:04:01.610 easy it is to create an agent directly within 00:04:01.610 --> 00:04:04.710 the Copilot web UI. And we've seen some really 00:04:04.710 --> 00:04:06.870 interesting things. One of the most powerful 00:04:06.870 --> 00:04:08.849 ways and the thing that I started to see the 00:04:08.849 --> 00:04:11.189 most within a lot of different organizations 00:04:11.189 --> 00:04:14.469 is creating like a document library search agent. 00:04:14.629 --> 00:04:17.050 And what I mean by that, like, I think a lot 00:04:17.050 --> 00:04:18.730 of people, like even when I came to Microsoft, 00:04:18.810 --> 00:04:21.750 I'm like, how do I, there's so many HR documents, 00:04:21.949 --> 00:04:24.980 policy documents, like how do I? figure out all 00:04:24.980 --> 00:04:26.600 these and understand all these and there's still 00:04:26.600 --> 00:04:29.199 times where i'm like just just today i went to 00:04:29.199 --> 00:04:31.740 i was at a customer on site and i couldn't remember 00:04:31.740 --> 00:04:37.379 like um How much is it? When do I need a receipt? 00:04:37.519 --> 00:04:39.800 What's the dollar value that I need a receipt? 00:04:39.879 --> 00:04:42.399 So I just went to our agent and asked that question. 00:04:42.500 --> 00:04:45.459 And creating those type of policy agents where 00:04:45.459 --> 00:04:47.319 we take a whole bunch of SharePoint document 00:04:47.319 --> 00:04:49.319 libraries and create a real simple agent that 00:04:49.319 --> 00:04:51.220 people can use natural language to ask, like, 00:04:51.259 --> 00:04:53.420 what's a parental leave or what are my insurance 00:04:53.420 --> 00:04:55.779 benefits? I've seen a lot of those. Yeah, no, 00:04:55.899 --> 00:05:00.860 that's great. And I think that, like, the research, 00:05:01.040 --> 00:05:05.139 summarizing, reasoning. That's kind of like the 00:05:05.139 --> 00:05:09.100 bread and butter of AI and what we use Copilot. 00:05:09.279 --> 00:05:12.379 And it's been very good at like looking at a 00:05:12.379 --> 00:05:15.360 specific knowledge repository of documents, whether 00:05:15.360 --> 00:05:17.439 they're SharePoint sites or OneDrive or what 00:05:17.439 --> 00:05:20.959 have you, and then giving you information back 00:05:20.959 --> 00:05:23.779 so you can have like an HR agent. You know, one 00:05:23.779 --> 00:05:26.339 of the... The things that I always see in the 00:05:26.339 --> 00:05:30.420 news, right? And I'm sure everyone's seen this. 00:05:30.839 --> 00:05:35.220 There's so much fear about AI and kind of like 00:05:35.220 --> 00:05:38.779 the future and like people losing their jobs 00:05:38.779 --> 00:05:41.620 and all this stuff and automation. And I don't 00:05:41.620 --> 00:05:44.139 know, like it's interesting because you and I 00:05:44.139 --> 00:05:48.360 are like on the front lines of AI and using it 00:05:48.360 --> 00:05:51.540 and talking to customers. And at the same time, 00:05:51.560 --> 00:05:54.100 like I see these. other people that have these 00:05:54.100 --> 00:05:56.439 really fancy titles you know like i'm a data 00:05:56.439 --> 00:05:58.699 scientist for this i'm the ceo of that and they're 00:05:58.699 --> 00:06:00.740 like be careful this is coming that's coming 00:06:00.740 --> 00:06:04.959 i still haven't seen what it is that they're 00:06:04.959 --> 00:06:07.040 talking about that's coming but they're definitely 00:06:07.040 --> 00:06:10.670 like doing a good job of scaring everyone And 00:06:10.670 --> 00:06:13.889 I always I just still look at AI as like a tool 00:06:13.889 --> 00:06:17.750 that's helping me get rid of or get rid of stuff 00:06:17.750 --> 00:06:21.009 that I don't want to do. That's, you know, like 00:06:21.009 --> 00:06:23.990 over and over again or information that I'm looking 00:06:23.990 --> 00:06:26.730 for, like the example you gave when you're on 00:06:26.730 --> 00:06:29.610 a business trip and you want to know what your 00:06:29.610 --> 00:06:32.689 expenses are. So I still see a lot of this stuff 00:06:32.689 --> 00:06:35.699 that. for the most part is benign. Now what's 00:06:35.699 --> 00:06:38.480 going to happen five or 10 or 15 or 20 years 00:06:38.480 --> 00:06:42.720 from now, who knows, right? Like, but I wonder, 00:06:42.759 --> 00:06:46.839 like, do you, do you talk to people that are 00:06:46.839 --> 00:06:49.839 kind of worried about their, their, their jobs 00:06:49.839 --> 00:06:53.439 and their careers and, and AI, or is everyone 00:06:53.439 --> 00:06:56.259 more like relaxed and calm? Like, yeah, you know, 00:06:56.279 --> 00:06:58.889 this is just a tool that I'm using. Yeah. It's, 00:06:58.889 --> 00:07:01.189 I think it's really hard for like us in our roles 00:07:01.189 --> 00:07:03.449 because we live in AI, we live and breathe it. 00:07:03.470 --> 00:07:06.449 But I think, yeah, I get there. There is a definitely 00:07:06.449 --> 00:07:09.269 a fear out there. Like what, what is the future 00:07:09.269 --> 00:07:13.889 going to hold? And. at least right now where 00:07:13.889 --> 00:07:15.529 I'm at, like, I can't tell you like what two 00:07:15.529 --> 00:07:17.550 years, what two years is going to look like from 00:07:17.550 --> 00:07:20.230 here, but I don't know. Like if, if we kind of 00:07:20.230 --> 00:07:22.689 like look back at history a lot, like let's take, 00:07:22.730 --> 00:07:24.990 for instance, like when, when the internet really 00:07:24.990 --> 00:07:27.389 started to take hold, like everyone is like, 00:07:27.430 --> 00:07:28.810 we're going to lose our job. The internet's going 00:07:28.810 --> 00:07:30.870 to replace everything. And what actually happened 00:07:30.870 --> 00:07:33.870 was like, we learned to evolve and we started 00:07:33.870 --> 00:07:37.449 digital businesses and stuff like that. yes i 00:07:37.449 --> 00:07:39.350 think there is a there is a fear out there because 00:07:39.350 --> 00:07:42.430 we still don't know like this ai like is a complete 00:07:42.430 --> 00:07:46.050 revolution but i think what we will start seeing 00:07:46.050 --> 00:07:48.790 is like ai being like you said like an assistant 00:07:48.790 --> 00:07:51.910 helping helping the human and i think we'll just 00:07:51.910 --> 00:07:54.129 kind of evolve with how we work with that like 00:07:54.129 --> 00:07:56.810 it ai has definitely changed my life like honestly 00:07:56.810 --> 00:07:59.889 i could not like see myself working without copilot 00:07:59.889 --> 00:08:02.949 because it's it's so helpful but like i'll give 00:08:02.949 --> 00:08:06.899 you an example like um uh i was I was working 00:08:06.899 --> 00:08:11.779 with some executives a couple days ago, and we 00:08:11.779 --> 00:08:14.699 were kind of showcasing this really cool agent 00:08:14.699 --> 00:08:17.759 inside of Excel. And I think a lot of people 00:08:17.759 --> 00:08:19.579 think when we talk about agents, we think, oh, 00:08:19.600 --> 00:08:21.740 an agent inside of Copilot. And yes, that's a 00:08:21.740 --> 00:08:26.800 place where agents live, but agents can reside 00:08:26.800 --> 00:08:30.199 in a lot of different areas like Excel. And we 00:08:30.199 --> 00:08:32.440 were kind of showcasing the power of agent mode. 00:08:32.500 --> 00:08:33.759 And if you're not familiar with agent mode, what 00:08:33.759 --> 00:08:36.539 it is, you did a really great video on it. It's 00:08:36.539 --> 00:08:39.120 an agent that's embedded directly in Excel that 00:08:39.120 --> 00:08:42.000 has like really, it's like almost like a data 00:08:42.000 --> 00:08:44.620 scientist at your fingertips. It has like, could 00:08:44.620 --> 00:08:47.259 do backend Python code and really complex formulas. 00:08:47.460 --> 00:08:49.500 But what we did is we kind of took this scenario 00:08:49.500 --> 00:08:54.700 like, hey, my CFO needs a Q2 analysis of this 00:08:54.700 --> 00:08:58.519 thing. And we had all this raw data and I'm like, 00:08:58.620 --> 00:09:01.940 hey, I need a data scientist to take a look at 00:09:01.940 --> 00:09:03.559 this. And I gave it to my data scientist and 00:09:03.559 --> 00:09:06.149 she kind of, She opened up Alpha's agent mode 00:09:06.149 --> 00:09:08.110 and she put in like six prompts and she turned 00:09:08.110 --> 00:09:10.730 this raw data into like all of these cool insights. 00:09:10.870 --> 00:09:13.549 She's like created a dashboard, an SBAR report. 00:09:13.629 --> 00:09:15.669 She even created like a what if scenario dashboard 00:09:15.669 --> 00:09:20.429 for the CFO. And it looked really cool. But the 00:09:20.429 --> 00:09:23.389 thing I really took away from that is. Back to 00:09:23.389 --> 00:09:26.549 your idea of fear is, I don't think AI is going 00:09:26.549 --> 00:09:28.870 to fully replace the human because with that 00:09:28.870 --> 00:09:31.669 data scientist, yes, that data scientist knows 00:09:31.669 --> 00:09:34.289 how to create all of those really cool dashboards 00:09:34.289 --> 00:09:36.389 and all of these cool pie charts and stuff like 00:09:36.389 --> 00:09:40.870 that. But the thing that Copilot did is... She 00:09:40.870 --> 00:09:42.850 knew how to create all that, but she also knew 00:09:42.850 --> 00:09:45.710 how to prompt co -pilot. So we didn't degrade 00:09:45.710 --> 00:09:49.210 her role. She, we just, we kind of reframed of 00:09:49.210 --> 00:09:51.370 how she works with the data where instead of 00:09:51.370 --> 00:09:54.129 trying to take hours upon hours of creating all 00:09:54.129 --> 00:09:57.070 this cool content, she knows she, she knew what 00:09:57.070 --> 00:09:59.669 she wanted out of that content and was able to 00:09:59.669 --> 00:10:01.889 use natural language to get the agent to do that. 00:10:01.929 --> 00:10:04.110 So I think there's that data scientists, it was 00:10:04.110 --> 00:10:06.879 important that. You know, she knows what she's 00:10:06.879 --> 00:10:09.620 trained to understand the data and understand 00:10:09.620 --> 00:10:11.860 what she wants out of it. And then she articulate 00:10:11.860 --> 00:10:14.279 that to an agent, to an assistant, to be able 00:10:14.279 --> 00:10:17.399 to pull that out. And for that, it's, she saved 00:10:17.399 --> 00:10:20.019 hours to know. So now like she can do so much 00:10:20.019 --> 00:10:23.240 more with that time, maybe another analyst of 00:10:23.240 --> 00:10:25.159 another document or something like that. But 00:10:25.159 --> 00:10:28.480 I think that's where I really see like AI still 00:10:28.480 --> 00:10:30.519 being an assistant to human. Like there could 00:10:30.519 --> 00:10:33.610 be. areas where we fully automate things. But 00:10:33.610 --> 00:10:37.289 I think for a lot of things, I don't think AI 00:10:37.289 --> 00:10:39.129 is really to the point where it replaces everything 00:10:39.129 --> 00:10:41.549 that a human can do, but it's going to assist 00:10:41.549 --> 00:10:44.350 that human to make them more productive. Yeah. 00:10:44.529 --> 00:10:47.190 And I love that example. I actually, you're right, 00:10:47.230 --> 00:10:49.990 I did some content, the video on agent mode in 00:10:49.990 --> 00:10:52.850 Excel, and I'm very impressed with agent mode 00:10:52.850 --> 00:10:57.029 because up until agent mode came out inside Excel, 00:10:57.250 --> 00:11:00.070 Word, and PowerPoint, Copilot would tell you, 00:11:00.090 --> 00:11:02.409 give you instructions. Like it would say, click 00:11:02.409 --> 00:11:04.929 here and go to this menu item and copy and paste 00:11:04.929 --> 00:11:07.700 this. And I was like. geez like just do it you 00:11:07.700 --> 00:11:10.600 know like i i don't know like because some people 00:11:10.600 --> 00:11:13.860 like they use excel and they use word and powerpoint 00:11:13.860 --> 00:11:15.899 and they're masters at it right they know like 00:11:15.899 --> 00:11:19.259 keyboard shortcuts and they're just like wizards 00:11:19.259 --> 00:11:22.039 inside these applications i'm not like maybe 00:11:22.039 --> 00:11:24.179 i do presentation maybe i do this and that but 00:11:24.179 --> 00:11:26.419 i don't know every single feature in these applications 00:11:26.419 --> 00:11:29.539 even though i work for microsoft that's right 00:11:29.539 --> 00:11:32.820 right so i don't But, like, agent mode, it's 00:11:32.820 --> 00:11:34.559 kind of like you don't need to, right? You just 00:11:34.559 --> 00:11:37.179 tell it what you want. Create a chart for me. 00:11:37.240 --> 00:11:41.019 Do a bar graph. You know, format this column 00:11:41.019 --> 00:11:43.539 in this fashion. And it does it for you. You 00:11:43.539 --> 00:11:46.100 don't have to know where all the features are, 00:11:46.159 --> 00:11:50.659 right? So, yeah, I often call agent mode. I like 00:11:50.659 --> 00:11:52.700 to call it vibe coding, essentially, in Excel. 00:11:53.860 --> 00:11:56.889 Because you just do it in natural language. have 00:11:56.889 --> 00:11:58.909 it do something. I think that's a really good 00:11:58.909 --> 00:12:01.190 example of what agent mode is like by coding 00:12:01.190 --> 00:12:03.649 directly in Excel. Yeah, no, that's awesome. 00:12:03.769 --> 00:12:07.009 And I think back to your point, like I also look 00:12:07.009 --> 00:12:10.289 at AI and all these different technologies as 00:12:10.289 --> 00:12:14.450 tools. I create my video thumbnails sometimes 00:12:14.450 --> 00:12:19.000 using open AI. There's all kinds of use cases 00:12:19.000 --> 00:12:21.940 that I have for AI, but now I don't have to be 00:12:21.940 --> 00:12:25.559 like a graphic artist to create the perfect thumbnail 00:12:25.559 --> 00:12:29.299 for a video or know every feature inside Excel. 00:12:29.399 --> 00:12:33.759 I can use this assistant, to your point, to help 00:12:33.759 --> 00:12:37.279 me, right? And so for those reasons, I'm also 00:12:37.279 --> 00:12:43.860 not fearful of AI and replacing jobs. I'm excited 00:12:43.860 --> 00:12:47.879 that... I now have access to a tool that can 00:12:47.879 --> 00:12:51.740 help me, whereas before I would have to call 00:12:51.740 --> 00:12:54.740 someone or hire a professional to help me. Are 00:12:54.740 --> 00:12:58.100 you familiar with, what is it called, just -in 00:12:58.100 --> 00:13:00.399 -time manufacturing, Jim? Are you familiar with 00:13:00.399 --> 00:13:04.450 that concept? I'm guessing. where you order something 00:13:04.450 --> 00:13:06.669 and it's not made yet and they build it for you, 00:13:06.710 --> 00:13:08.870 something like that? Yeah, pretty much. Apple's 00:13:08.870 --> 00:13:11.429 a really good example of this, where they kind 00:13:11.429 --> 00:13:13.769 of, just now in manufacturing, is the idea of 00:13:13.769 --> 00:13:17.470 you're ordering and building the stuff just in 00:13:17.470 --> 00:13:18.870 time, where it's not like you're ordering all 00:13:18.870 --> 00:13:20.309 this stuff and keeping it on a shelf and have 00:13:20.309 --> 00:13:22.470 to pay for all that stuff. I kind of, when I 00:13:22.470 --> 00:13:24.950 first joined Microsoft, I kind of, I don't think 00:13:24.950 --> 00:13:26.450 I coined this term, I think I heard it somewhere, 00:13:26.549 --> 00:13:30.720 but I've used AI in... I call it just -in -time 00:13:30.720 --> 00:13:32.379 learning. I think I picked this up from someone 00:13:32.379 --> 00:13:35.600 else. When I first joined Microsoft, I was an 00:13:35.600 --> 00:13:40.860 MDM guy. I worked with Intune, Jamf, Workspace 00:13:40.860 --> 00:13:42.820 ONE, and stuff like that. Then I got hired in 00:13:42.820 --> 00:13:46.600 as a Teams SE, or TS at that time it was called. 00:13:46.820 --> 00:13:49.639 I'm like, oh man, they're taking an MDM guy and 00:13:49.639 --> 00:13:51.940 making me a collaborative guy. Eventually, they... 00:13:52.190 --> 00:13:54.350 they molded into that what we call now are they 00:13:54.350 --> 00:13:57.169 the co -pilot se but uh when i when i joined 00:13:57.169 --> 00:14:00.809 myself like i've got to learn so much in a very 00:14:00.809 --> 00:14:04.370 little time. And AI was such a powerful tool 00:14:04.370 --> 00:14:07.649 for me because I was able to like use tools like 00:14:07.649 --> 00:14:10.009 researcher and stuff like that to like really 00:14:10.009 --> 00:14:12.230 quickly learn things like just in time learning. 00:14:12.289 --> 00:14:15.070 And I think like back to that fear thing, like 00:14:15.070 --> 00:14:17.110 sure there's some fear, but there's a lot of 00:14:17.110 --> 00:14:19.190 really cool advantages that we can take. And 00:14:19.190 --> 00:14:22.250 I think as we evolve and kind of figure out where 00:14:22.250 --> 00:14:24.950 AI is going to live with humans and all of that. 00:14:25.559 --> 00:14:28.419 Things like being able to use AI to quickly learn 00:14:28.419 --> 00:14:31.639 and catch up and stuff like that is going to 00:14:31.639 --> 00:14:33.860 be super important. And I could see, you know, 00:14:33.860 --> 00:14:36.080 future students and stuff like that, really leveraging 00:14:36.080 --> 00:14:38.620 those type of tools to be able to learn quickly 00:14:38.620 --> 00:14:41.940 and evolve and, you know, get new jobs that. 00:14:42.639 --> 00:14:44.279 They have to learn a whole bunch of stuff in 00:14:44.279 --> 00:14:46.500 like me. But, you know, I think that's a really 00:14:46.500 --> 00:14:49.580 powerful way of using AI that's, you know, if 00:14:49.580 --> 00:14:53.340 we're fearful of it, look at ways to really take 00:14:53.340 --> 00:14:57.220 advantage of it and evolve yourself to learn 00:14:57.220 --> 00:15:00.299 new skills and so forth. Exactly. And that's 00:15:00.299 --> 00:15:03.379 the other thing, right? Like if you can create 00:15:03.379 --> 00:15:06.539 automation or do something faster because now 00:15:06.539 --> 00:15:09.759 you have this powerful tool. That doesn't mean 00:15:09.759 --> 00:15:12.659 that you can take that half an hour saved. I 00:15:12.659 --> 00:15:14.679 mean, it can, but it doesn't necessarily mean 00:15:14.679 --> 00:15:18.200 you go to sleep or take a nap or whatever, extend 00:15:18.200 --> 00:15:22.159 your lunch by an hour. You could, once for a 00:15:22.159 --> 00:15:24.320 moment, get an extra half hour, just breathe. 00:15:24.659 --> 00:15:29.360 You could then up -level yourself, listen to 00:15:29.360 --> 00:15:33.850 a podcast, go watch a video. You know, go go 00:15:33.850 --> 00:15:36.850 read a blog post somewhere that helps you get 00:15:36.850 --> 00:15:39.509 up to speed and educated on something. So there 00:15:39.509 --> 00:15:42.389 are proactive things that you can do to continue 00:15:42.389 --> 00:15:45.289 to kind of build yourself. So so, yeah, like 00:15:45.289 --> 00:15:48.230 it saves you time and it just it's kind of up 00:15:48.230 --> 00:15:51.450 to you, like how you end up spending that time 00:15:51.450 --> 00:15:53.629 that you just saved with this new assistant. 00:15:54.269 --> 00:15:56.669 This also reminds me like actually tomorrow morning, 00:15:56.809 --> 00:15:59.909 I have a customer presentation where I'm talking 00:15:59.909 --> 00:16:03.110 to this leadership team that is in charge of 00:16:03.110 --> 00:16:07.529 cloud ops or cloud operations. And I'm talking 00:16:07.529 --> 00:16:10.789 about agents and I don't know much about cloud 00:16:10.789 --> 00:16:13.509 ops. And I've never been in that side of the 00:16:13.509 --> 00:16:16.210 business. Right. So this is an example of where 00:16:16.210 --> 00:16:18.389 I want to be able to relate to my customers. 00:16:18.389 --> 00:16:21.860 So I'm going to use Copilot to ask it. Hey, if 00:16:21.860 --> 00:16:23.940 I'm talking to a team that's focused on cloud 00:16:23.940 --> 00:16:27.980 ops, what are some types of workflows that they 00:16:27.980 --> 00:16:30.899 have that I can help them automate? And so that 00:16:30.899 --> 00:16:33.879 would be something that would be a lot more difficult 00:16:33.879 --> 00:16:36.159 for me to find that information on before. But 00:16:36.159 --> 00:16:38.399 now I can just ask Copilot and it will probably 00:16:38.399 --> 00:16:41.320 give me some good examples. I really like that. 00:16:41.379 --> 00:16:44.480 Actually, man, one of the things I've started 00:16:44.480 --> 00:16:46.519 to do recently, like I've always known this is 00:16:46.519 --> 00:16:48.539 there, but like. Just recently, I really started 00:16:48.539 --> 00:16:50.279 taking advantage of because I've been in a couple 00:16:50.279 --> 00:16:54.340 executives meetings and some of them are pretty 00:16:54.340 --> 00:16:56.960 large. So I'm like, I'm just like nervous. Like, 00:16:56.980 --> 00:16:59.419 is this the right like when you talk to executives, 00:16:59.559 --> 00:17:01.799 their minds are different. Like they just the 00:17:01.799 --> 00:17:03.720 way that they think and stuff like that is just. 00:17:04.000 --> 00:17:06.599 is is a lot different and so one of the things 00:17:06.599 --> 00:17:07.839 one of the recent things i've been doing kind 00:17:07.839 --> 00:17:09.119 of like what you're doing with your cloud ops 00:17:09.119 --> 00:17:11.559 is like i'd had a powerpoint or i had like a 00:17:11.559 --> 00:17:14.059 talk track i was thinking about and i would dump 00:17:14.059 --> 00:17:16.240 it into copilot and i'd frame copilot like hey 00:17:16.240 --> 00:17:18.519 i'm going to an executive meeting so i need you 00:17:18.519 --> 00:17:21.440 to think like an executive and like analyze this 00:17:21.440 --> 00:17:24.700 this powerpoint is am i in track and a copilot 00:17:24.700 --> 00:17:27.940 or or whatever ai tool that you use it's it's 00:17:27.940 --> 00:17:30.980 really cool to be able to like It can help you 00:17:30.980 --> 00:17:32.920 get a different perspective. And like a lot of 00:17:32.920 --> 00:17:34.640 times it'll like help me like, Hey, you, you 00:17:34.640 --> 00:17:36.839 had this question or you had this title on this 00:17:36.839 --> 00:17:39.660 slide. think about it this way and maybe this 00:17:39.660 --> 00:17:41.700 is a better way to word it to really get that 00:17:41.700 --> 00:17:44.359 impact. So I was using AI to get different perspective 00:17:44.359 --> 00:17:46.119 based on like the meeting you're going into, 00:17:46.240 --> 00:17:48.720 the people that you're talking to. I was working 00:17:48.720 --> 00:17:51.180 with a customer a few months ago and one of the 00:17:51.180 --> 00:17:53.680 really interesting ways they use Copilot was 00:17:53.680 --> 00:17:55.779 they were kind of like in a, they were doing 00:17:55.779 --> 00:17:57.920 some negotiations with the vendor and they were 00:17:57.920 --> 00:18:00.839 just in a mind block. They couldn't get past 00:18:00.839 --> 00:18:03.440 anything. So they actually opened up Copilot 00:18:03.440 --> 00:18:05.440 and said, hey, I want you to act as this vendor. 00:18:05.559 --> 00:18:07.859 Here's all the contacts. just going to have a 00:18:07.859 --> 00:18:10.059 back and forth and your job here is to kind of 00:18:10.059 --> 00:18:12.339 help me negotiate this and get me past my brain 00:18:12.339 --> 00:18:14.319 block. And I'm like, that was a really interesting 00:18:14.319 --> 00:18:17.420 way of using Copilot to kind of like get a different 00:18:17.420 --> 00:18:19.259 perspective on things. And I think that's one 00:18:19.259 --> 00:18:22.339 of the really cool ways of that I found recently 00:18:22.339 --> 00:18:25.500 of just using AI. Exactly. And I think part of 00:18:25.500 --> 00:18:28.380 it is people are still learning all the different 00:18:28.380 --> 00:18:30.680 ways to use it. And even going back to what you 00:18:30.680 --> 00:18:32.640 said, like the prompts and prompt engineering, 00:18:32.859 --> 00:18:35.839 which was talked about a lot when AI came forth. 00:18:36.250 --> 00:18:38.849 A lot of people are still might be stuck at that 00:18:38.849 --> 00:18:41.990 stage. Like what, what is the art of the possible? 00:18:42.109 --> 00:18:46.450 What can I do? What do I type? And we, we have 00:18:46.450 --> 00:18:49.390 like different tools to show people. And there's 00:18:49.390 --> 00:18:51.630 even something called the prompt coach. Now that 00:18:51.630 --> 00:18:54.509 you can type in an initial prompt and the prompt 00:18:54.509 --> 00:18:58.710 coach expands that prompt and puts additional 00:18:58.710 --> 00:19:01.230 instructions in there. So, and then you can take 00:19:01.230 --> 00:19:04.569 that and feed it back into Copilot. So some of 00:19:04.569 --> 00:19:07.890 that stuff is. starting to improve people are 00:19:07.890 --> 00:19:10.750 getting to understand more use cases and there's 00:19:10.750 --> 00:19:13.410 more agents that are like specific being built 00:19:13.410 --> 00:19:15.569 for people but there's still a lot of people 00:19:15.569 --> 00:19:18.730 that are starting to learn the basics if you 00:19:18.730 --> 00:19:22.369 will and i think over time um you know people 00:19:22.369 --> 00:19:24.869 will be more comfortable and all the different 00:19:24.869 --> 00:19:29.130 use cases it sounds weird but um This is, this 00:19:29.130 --> 00:19:30.990 is a way that I, so Microsoft, we have like a 00:19:30.990 --> 00:19:33.710 four pillar of the prompt ingredients. And this 00:19:33.710 --> 00:19:37.170 is how I always remember it. Drafts can see everything. 00:19:37.349 --> 00:19:39.430 And it sounds really weird, but draft stands 00:19:39.430 --> 00:19:42.269 for G goal. So like, what goal do you want? So 00:19:42.269 --> 00:19:43.650 these are like the ingredients, like when you're 00:19:43.650 --> 00:19:45.390 building a prompt, think about these four things. 00:19:45.769 --> 00:19:48.009 Drafts can see everything. Goal, context, source, 00:19:48.130 --> 00:19:50.109 and expectations. That's how I remember that 00:19:50.109 --> 00:19:52.289 one. But whenever like building your prompt out, 00:19:52.369 --> 00:19:54.839 I think, I think a lot of people are still. you 00:19:54.839 --> 00:19:56.579 know, building out their, their prompting skills 00:19:56.579 --> 00:19:59.539 and learning how to interact with AI. You know, 00:19:59.539 --> 00:20:02.740 look at your goal, the context, the source, like 00:20:02.740 --> 00:20:05.380 add in things and your expectations, what you 00:20:05.380 --> 00:20:07.579 want out of the output in there. But you know, 00:20:07.579 --> 00:20:09.519 that's, I've, I've been, I've been educating 00:20:09.519 --> 00:20:11.900 that a lot. But I think one of the most powerful 00:20:11.900 --> 00:20:13.779 ways when we kind of like think about prompting, 00:20:13.799 --> 00:20:16.579 I think AI has gotten to the point where it's 00:20:16.579 --> 00:20:18.779 actually really, really good at understanding. 00:20:19.630 --> 00:20:21.970 the intent of the user. I think there's still 00:20:21.970 --> 00:20:23.809 times where you want to make sure that you're 00:20:23.809 --> 00:20:26.089 abiding by those prompting gradients of goal, 00:20:26.190 --> 00:20:28.809 context, sources, expectations. But some of the 00:20:28.809 --> 00:20:30.849 ways that I've built some prompts or just gotten 00:20:30.849 --> 00:20:33.029 stuff out of there is just brain dump into Copilot 00:20:33.029 --> 00:20:37.630 or brain dump into your AI. And I think just 00:20:37.630 --> 00:20:39.950 naturally, I've kind of abided a lot of the way 00:20:39.950 --> 00:20:41.769 that I brain dump kind of goes into that stuff. 00:20:41.970 --> 00:20:44.230 But, you know, just brain dump in there and you'll 00:20:44.230 --> 00:20:46.150 be, a lot of times you'll just be surprised how 00:20:46.150 --> 00:20:49.309 well an AI system can understand that. then i 00:20:49.309 --> 00:20:51.230 think one of the more important pieces is when 00:20:51.230 --> 00:20:54.730 you do that it's like no whatever ai you're in 00:20:54.730 --> 00:20:57.650 it's it's don't stop at like don't stop at the 00:20:57.650 --> 00:20:59.490 first thing like a lot i think a lot of users 00:20:59.490 --> 00:21:01.380 will put their prompt in they get the output 00:21:01.380 --> 00:21:04.660 they don't like it and they just stop um continue 00:21:04.660 --> 00:21:07.279 like collaborating with your ai like if you if 00:21:07.279 --> 00:21:09.359 you miss the source or you mix or you didn't 00:21:09.359 --> 00:21:11.799 add the right context to it add it in the next 00:21:11.799 --> 00:21:13.740 prompt in line with that conversation because 00:21:13.740 --> 00:21:15.960 all the ais have a memory so they have a memory 00:21:15.960 --> 00:21:17.740 of that conversation so as you continue to talk 00:21:17.740 --> 00:21:20.920 to it it can continue to get the right output 00:21:20.920 --> 00:21:22.619 that you want. I think a lot of people miss that. 00:21:22.700 --> 00:21:24.880 So like, um, if you don't like, I always say 00:21:24.880 --> 00:21:26.839 like, if you don't like it, try, try again, keep, 00:21:26.880 --> 00:21:28.700 keep working in line with that until you get, 00:21:28.700 --> 00:21:31.170 until you get to the. To the prompt or the output 00:21:31.170 --> 00:21:33.630 that you want. Exactly. And that's kind of me, 00:21:33.670 --> 00:21:37.650 actually. Like, I don't remember that one sentence 00:21:37.650 --> 00:21:39.650 that you just shared. Like, I don't remember 00:21:39.650 --> 00:21:42.130 all the different parts of the prompt. So a lot 00:21:42.130 --> 00:21:44.569 of times I give it what I think it needs. And 00:21:44.569 --> 00:21:47.470 if it's not quite right, I keep adding to it 00:21:47.470 --> 00:21:49.769 and keep adding to it until, to your point, it 00:21:49.769 --> 00:21:52.309 gets it right. And one of my coworkers actually 00:21:52.309 --> 00:21:55.750 said that eventually when, if I've done like 00:21:55.750 --> 00:21:59.170 a lot of back and forth, I might come back to... 00:21:59.500 --> 00:22:02.359 co -pilot or AI at the end of it and say, give 00:22:02.359 --> 00:22:05.740 me the one prompt that would have gotten me to 00:22:05.740 --> 00:22:08.539 where we've gotten to at this point. And then 00:22:08.539 --> 00:22:11.920 the AI goes back and just generates like a two 00:22:11.920 --> 00:22:14.119 or three sentence prompt. Yeah, I've done that 00:22:14.119 --> 00:22:16.319 before. Yeah, that should have been what you 00:22:16.319 --> 00:22:18.519 put in the beginning, you know. Based on everything 00:22:18.519 --> 00:22:21.240 we've talked about so far, create a prompt for 00:22:21.240 --> 00:22:23.900 this whole conversation. Exactly, right? So then 00:22:23.900 --> 00:22:26.400 you can like... Copy and paste that or bookmark 00:22:26.400 --> 00:22:28.380 it and reuse it next time. Save that to your 00:22:28.380 --> 00:22:31.279 prom gallery. There you go. That's right. Now, 00:22:31.420 --> 00:22:34.400 one thing I'm curious what your thoughts are 00:22:34.400 --> 00:22:36.180 on this. This is from my personal experience. 00:22:36.279 --> 00:22:39.819 So a lot of times when I'm creating a video, 00:22:39.940 --> 00:22:45.440 I go to ChatGPT and I ask it like, hey, I'm creating 00:22:45.440 --> 00:22:49.119 a video about this. What's a good title for the 00:22:49.119 --> 00:22:51.579 video? And it'll give me like a bunch of titles. 00:22:51.660 --> 00:22:55.160 It'll give me some ideas, right? And on that 00:22:55.160 --> 00:22:58.279 particular day, I settle on one of them and I'm 00:22:58.279 --> 00:22:59.640 like, okay, this is good. I'm going to keep it. 00:22:59.940 --> 00:23:01.940 And then like a day or two later, I'm like, ah, 00:23:02.039 --> 00:23:04.359 I don't really like it. I go back to that same 00:23:04.359 --> 00:23:07.279 chat and I'm like, well, you know what? I don't 00:23:07.279 --> 00:23:09.500 really, you know, what about this one instead? 00:23:11.530 --> 00:23:13.410 Where I'm going with this is like, sometimes 00:23:13.410 --> 00:23:16.910 I feel like AI is just very agreeable. Like when 00:23:16.910 --> 00:23:19.549 I, when I, when I say to it, like, okay, this 00:23:19.549 --> 00:23:21.910 is great. And it's like, and it comes back in 00:23:21.910 --> 00:23:23.410 on the first day and it's like, yeah, let's lock 00:23:23.410 --> 00:23:25.970 it in. You're good to go. Then the next day I'm 00:23:25.970 --> 00:23:27.589 like, well, you know what, what about this? And 00:23:27.589 --> 00:23:29.390 you're like, oh yeah, the old one sucks. You 00:23:29.390 --> 00:23:32.369 know, like, I'm like, what? You told me the old 00:23:32.369 --> 00:23:34.329 one was great. Now you're telling me it sucks. 00:23:34.470 --> 00:23:37.559 And you know, so like, I'm like. So I feel like 00:23:37.559 --> 00:23:40.099 it's just like very agreeable and it can like 00:23:40.099 --> 00:23:43.119 agree with anything that you say. So have you 00:23:43.119 --> 00:23:45.940 run into something like that before? Yeah, absolutely. 00:23:46.099 --> 00:23:48.660 All the time. Sometimes I specifically have to 00:23:48.660 --> 00:23:51.930 like frame or like tell the AI like. No fluff, 00:23:52.069 --> 00:23:53.910 like be real or something like that. Actually 00:23:53.910 --> 00:23:56.690 fun. I built an agent just for this. Like I built 00:23:56.690 --> 00:23:59.190 an agent that helped me rewrite stuff. And inside 00:23:59.190 --> 00:24:00.990 of the agent, I specifically put instructions 00:24:00.990 --> 00:24:03.230 where like, I don't want you to agree with me 00:24:03.230 --> 00:24:05.230 or anything like that. Or cause I'll have it 00:24:05.230 --> 00:24:06.849 rewrite stuff or help me get new perspective. 00:24:06.950 --> 00:24:09.369 So I like built it in agents just so that I, 00:24:09.410 --> 00:24:11.509 it wouldn't give me fluff. But actually one of 00:24:11.509 --> 00:24:13.549 the things I've done that I think I wouldn't 00:24:13.549 --> 00:24:16.470 say it has a hundred percent help, but has slimmed 00:24:16.470 --> 00:24:19.279 down a little bit. I've actually. gone through 00:24:19.279 --> 00:24:22.059 the copilot personalization stuff so I kind of 00:24:22.059 --> 00:24:24.500 and inside of that copilot personalization you 00:24:24.500 --> 00:24:26.059 know when you go to settings and personalize 00:24:26.059 --> 00:24:28.460 and edit custom instructions inside of my custom 00:24:28.460 --> 00:24:30.019 instructions I put some of those areas where 00:24:30.019 --> 00:24:32.200 I don't want it to like be totally agreeable 00:24:32.200 --> 00:24:34.259 with me and stuff like that doesn't work 100 00:24:34.259 --> 00:24:36.339 % of the time but I have noticed it's been a 00:24:36.339 --> 00:24:39.500 lot better but yeah I run into that a lot exactly 00:24:39.500 --> 00:24:42.259 yeah it's like talking to my mom you know she's 00:24:42.259 --> 00:24:47.150 very supportive just tell me the truth I can 00:24:47.150 --> 00:24:48.890 almost, like, hear the AI just saying, like, 00:24:48.910 --> 00:24:52.009 you can't handle the troops. Exactly. So, yeah, 00:24:52.049 --> 00:24:53.869 so that's one of the things that I run into. 00:24:53.990 --> 00:24:56.529 Maybe all that will get better over time as well. 00:24:56.829 --> 00:25:00.809 I think OpenAI's 5 .2 has done a good job of 00:25:00.809 --> 00:25:03.930 that. I think it's been less, like, super positive 00:25:03.930 --> 00:25:06.589 mom a little bit. But, yeah, I think they'll 00:25:06.589 --> 00:25:09.470 continue to get better because I think that's 00:25:09.470 --> 00:25:10.690 probably one of the bigger things a lot of people 00:25:10.690 --> 00:25:15.730 run into. Okay. Yeah, I hope that improves. The 00:25:15.730 --> 00:25:18.890 other thing that you mentioned before we started 00:25:18.890 --> 00:25:21.720 this podcast was. You were talking about how 00:25:21.720 --> 00:25:25.339 we have with Copilot, especially in studio when 00:25:25.339 --> 00:25:27.500 you're building agents, we have access to different 00:25:27.500 --> 00:25:29.680 large language models. You can kind of go to 00:25:29.680 --> 00:25:32.160 this dropdown. And even in different places inside 00:25:32.160 --> 00:25:35.000 Excel, you can say, hey, I want to use OpenAI 00:25:35.000 --> 00:25:37.960 GPT -5 too, or I want to use Anthropix Cloud, 00:25:38.160 --> 00:25:41.619 or I want to use other models that might be coming 00:25:41.619 --> 00:25:44.480 in the future, right? So I think that's kind 00:25:44.480 --> 00:25:49.000 of cool. And I'm wondering if you are experiencing... 00:25:49.880 --> 00:25:51.799 kind of like different bang for your buck right 00:25:51.799 --> 00:25:54.500 as you as you try these models or if you're just 00:25:54.500 --> 00:25:57.619 settling on auto or what your experience is on 00:25:57.619 --> 00:26:00.740 that side One of the things I've noticed is Anthropic's 00:26:00.740 --> 00:26:02.980 been getting a lot of, a lot of, a lot of people 00:26:02.980 --> 00:26:04.539 are really interested in what Anthropic's doing 00:26:04.539 --> 00:26:07.119 in Claude. And so I've been seeing like a lot 00:26:07.119 --> 00:26:08.960 of like on TikTok and YouTube and all that, all 00:26:08.960 --> 00:26:11.339 these people like inserting the Claude add -in 00:26:11.339 --> 00:26:13.299 in Excel to do all this cool stuff. Because Claude's 00:26:13.299 --> 00:26:15.460 really good at kind of understanding structured 00:26:15.460 --> 00:26:17.920 data, which is why it makes sense to do it in 00:26:17.920 --> 00:26:20.390 Excel and PowerPoint and stuff like that. Um, 00:26:20.650 --> 00:26:22.430 and I think one of the cool things with Copilot 00:26:22.430 --> 00:26:25.529 is, um, I think one of the, a lot of the value 00:26:25.529 --> 00:26:27.890 that Copilot brings is, uh, what Microsoft is 00:26:27.890 --> 00:26:30.529 doing is we're kind of being, um, kind of multimodal 00:26:30.529 --> 00:26:34.670 where, um, you, we, we, we're giving users and 00:26:34.670 --> 00:26:36.569 organizations choice in the, in the type of models 00:26:36.569 --> 00:26:38.890 that they use. So like by default, you know, 00:26:38.890 --> 00:26:41.390 we use open AI models, but you also have the 00:26:41.390 --> 00:26:43.690 choice now to use Anthropix cloud models. And 00:26:43.690 --> 00:26:45.750 like back to that agent mode, like if I was to 00:26:45.750 --> 00:26:47.950 do that, like kind of the whole demo again, like. 00:26:48.640 --> 00:26:50.980 You could choose if you want to use OpenAI or 00:26:50.980 --> 00:26:53.039 Anthropic to do all of this cool stuff in Excel. 00:26:53.220 --> 00:26:54.559 And I think that's really powerful. And then 00:26:54.559 --> 00:26:56.579 you can also do that in like Copilot Studio where 00:26:56.579 --> 00:26:57.900 you can have all these different models that 00:26:57.900 --> 00:27:01.309 are, you know. I think that have different strengths. 00:27:01.490 --> 00:27:03.730 So like if we're doing something in Excel, like 00:27:03.730 --> 00:27:05.609 I want to use Anthropic because it's just better 00:27:05.609 --> 00:27:07.569 at understanding structured data at this moment 00:27:07.569 --> 00:27:10.190 in time. Like I think LLMs are so up and down 00:27:10.190 --> 00:27:12.329 of who's high, who's low and stuff like that. 00:27:12.430 --> 00:27:14.569 But I think that's one of the things I've enjoyed 00:27:14.569 --> 00:27:16.950 at Microsoft. And I think some of our customers 00:27:16.950 --> 00:27:18.569 are kind of like understanding like, hey, there's 00:27:18.569 --> 00:27:21.289 options with Copilot. We've got a multimodal 00:27:21.289 --> 00:27:23.029 option to pick whichever one's going to be best 00:27:23.029 --> 00:27:25.750 for me because I think especially like everybody 00:27:25.750 --> 00:27:28.210 wants to use Cloud. It's like, oh, I can use 00:27:28.210 --> 00:27:31.269 it. I did not know that. It's like, yes, absolutely. 00:27:31.789 --> 00:27:34.589 Yeah. I like that. Right. The way you put it 00:27:34.589 --> 00:27:37.829 like multimodal and you can bring in your own 00:27:37.829 --> 00:27:40.109 large language model of choice, even different 00:27:40.109 --> 00:27:43.490 versions, you know, like four to five to deep, 00:27:43.529 --> 00:27:47.170 deep reasoning, quick response, uh, anthropic 00:27:47.170 --> 00:27:50.210 open AI. And there's rumors of other ones coming. 00:27:50.349 --> 00:27:54.910 Right. So it's like. You still have the application 00:27:54.910 --> 00:27:57.630 and you have Copilot in there and then you are 00:27:57.630 --> 00:28:00.750 not locked into any LLM. You can use the one 00:28:00.750 --> 00:28:03.490 that you prefer or even set it to auto and then 00:28:03.490 --> 00:28:06.410 it decides. And I think to your point, they're 00:28:06.410 --> 00:28:09.349 all kind of like good at something, better at 00:28:09.349 --> 00:28:11.869 something than others. So like, for example, 00:28:11.869 --> 00:28:16.990 when I use OpenAI, back to the video stuff. I 00:28:16.990 --> 00:28:19.269 think it does a really good job on the pictures, 00:28:19.490 --> 00:28:21.430 you know, because I get help on the thumbnails 00:28:21.430 --> 00:28:23.349 and it does a very good job because when I give 00:28:23.349 --> 00:28:26.089 the same prompt to like Gemini and these are 00:28:26.089 --> 00:28:28.049 both the free versions, by the way. So I know 00:28:28.049 --> 00:28:30.369 maybe the, you know, the paid versions do a better 00:28:30.369 --> 00:28:33.230 job. But like Gemini doesn't do that good of 00:28:33.230 --> 00:28:35.950 a good of a job with the pictures as as OpenAI 00:28:35.950 --> 00:28:38.109 does. Right. So like they're all good at something 00:28:38.109 --> 00:28:41.910 versus another. Yeah, I would agree with that. 00:28:42.460 --> 00:28:45.539 i remember it was christmas time uh and i saw 00:28:45.539 --> 00:28:47.299 like on tiktok some kind of prompt where you 00:28:47.299 --> 00:28:49.460 can like take a picture and turn it into something 00:28:49.460 --> 00:28:51.640 i forget the name but it was like uh some kind 00:28:51.640 --> 00:28:53.799 of like artist like profile like uh something 00:28:53.799 --> 00:28:56.160 like uh from like the 1900s and i'm like i put 00:28:56.160 --> 00:28:58.059 it in the gemini because at the time i was i 00:28:58.059 --> 00:28:59.440 was i was working with gemini because it was 00:28:59.440 --> 00:29:02.140 super cool and it's still get a lot of like eyes 00:29:02.140 --> 00:29:04.140 on it but and it just didn't work out and then 00:29:04.140 --> 00:29:06.619 i put it in chachi bt and like it worked flawlessly 00:29:06.619 --> 00:29:08.740 so yeah i think that's that's the you're gonna 00:29:08.740 --> 00:29:11.319 i think that's uh the thing we see in the industry 00:29:11.319 --> 00:29:13.859 is like All these different large language models 00:29:13.859 --> 00:29:15.839 have a lot of different strengths and weaknesses. 00:29:15.960 --> 00:29:18.539 Some are better than others and stuff like that, 00:29:18.579 --> 00:29:22.859 but it goes up and down so, so much. And by the 00:29:22.859 --> 00:29:25.980 way, just thinking about the different strategies 00:29:25.980 --> 00:29:29.420 that these companies have, like Microsoft, Google 00:29:29.420 --> 00:29:34.259 with Gemini, Anthropic with cloud, OpenAI, and 00:29:34.259 --> 00:29:38.099 even Apple with all the hardware products that 00:29:38.099 --> 00:29:40.940 they have, right? Their phones and laptops and 00:29:40.940 --> 00:29:43.779 everything that they can put Apple intelligence 00:29:43.779 --> 00:29:47.730 in. It's kind of interesting, right? Microsoft 00:29:47.730 --> 00:29:51.809 is more of this partnership with all these LLMs 00:29:51.809 --> 00:29:54.990 and they don't necessarily have their own large 00:29:54.990 --> 00:29:57.109 language model, but they have Copilot that's 00:29:57.109 --> 00:29:59.930 using them on the backend. And then, you know, 00:29:59.950 --> 00:30:02.349 I think Apple is supposed to partner with Google 00:30:02.349 --> 00:30:04.890 or Gemini in the future. So they all have their 00:30:04.890 --> 00:30:07.589 own. And then there's the creators of these large 00:30:07.589 --> 00:30:11.329 language models, even Grok by XAI, right? So 00:30:11.329 --> 00:30:13.329 there's all these different companies and strategies 00:30:13.329 --> 00:30:16.799 and Anthropic. comes the other way right they're 00:30:16.799 --> 00:30:19.359 like we have our own ai and we're gonna have 00:30:19.359 --> 00:30:22.660 hooks into the microsoft product so you know 00:30:22.660 --> 00:30:26.180 you can add you know put the ad into excel and 00:30:26.180 --> 00:30:28.059 then we're like well why put the ad into excel 00:30:28.059 --> 00:30:31.200 when you can just select cloud as as the option 00:30:31.200 --> 00:30:34.480 inside agent mode or co -pilot so everyone's 00:30:34.480 --> 00:30:36.500 kind of coming at this in different directions 00:30:36.500 --> 00:30:39.619 and it's going to be very interesting like what 00:30:39.619 --> 00:30:42.099 is the better strategy or the model or is all 00:30:42.099 --> 00:30:46.660 of the above Is all of the above going to work 00:30:46.660 --> 00:30:51.279 anyways, right? I think the big thing is, to 00:30:51.279 --> 00:30:54.900 me, which AI do I choose? If you're coming from 00:30:54.900 --> 00:30:57.599 a personal side of things, you've got so many 00:30:57.599 --> 00:31:00.700 choices. When you look at the work side of things, 00:31:00.779 --> 00:31:02.500 it becomes a little bit more difficult because, 00:31:02.599 --> 00:31:07.819 to me, where Copilot makes a lot of sense. is 00:31:07.819 --> 00:31:10.519 like when you start with an AI, like you're starting 00:31:10.519 --> 00:31:13.240 fresh, like the AI knows nothing about you and 00:31:13.240 --> 00:31:15.519 you have to build that kind of relationship and 00:31:15.519 --> 00:31:18.259 memory versus like Copilot. I think that the 00:31:18.259 --> 00:31:20.299 immediate value a lot of organizations get is 00:31:20.299 --> 00:31:22.380 like. say you've been at the organization for 00:31:22.380 --> 00:31:25.380 two, five, 10, 20 years or so, like it has all 00:31:25.380 --> 00:31:28.380 of that backend history about you and can like 00:31:28.380 --> 00:31:30.579 correlate the relationship between your data, 00:31:30.680 --> 00:31:33.119 your history at the company and what you're looking 00:31:33.119 --> 00:31:35.460 to get out of whatever you're looking for or 00:31:35.460 --> 00:31:38.019 prompting against. I think that is a big value 00:31:38.019 --> 00:31:41.140 that having some of that immediate like history 00:31:41.140 --> 00:31:44.539 and that work data. So you don't have to like 00:31:44.539 --> 00:31:47.220 start fresh with a new AI, even though like there's 00:31:47.220 --> 00:31:49.079 connectors into like Microsoft and stuff like 00:31:49.079 --> 00:31:51.359 that. You know, there's, it's just, it's, it's 00:31:51.359 --> 00:31:53.680 not the same as having like that direct pipe 00:31:53.680 --> 00:31:56.559 into the Microsoft graph and stuff like that. 00:31:56.579 --> 00:31:58.680 That's true. That's, that's a good point. So. 00:31:59.880 --> 00:32:02.160 there's different terminologies for that on the 00:32:02.160 --> 00:32:04.319 technical side right it was in the beginning 00:32:04.319 --> 00:32:07.680 Microsoft used like the semantic index and now 00:32:07.680 --> 00:32:09.779 it's like work IQ and like there's all these 00:32:09.779 --> 00:32:13.119 different terms that basically say that we know 00:32:13.119 --> 00:32:16.099 about you about your job about your work history 00:32:16.099 --> 00:32:19.980 so the context of the response from Copilot is 00:32:19.980 --> 00:32:22.160 going to be more relevant because of all this 00:32:22.160 --> 00:32:24.680 information that we have sometimes I wish like 00:32:24.680 --> 00:32:26.259 there was a reset button though like I don't 00:32:26.259 --> 00:32:29.599 want it to know anything about me but it's cool 00:32:29.599 --> 00:32:31.420 i mean like you see like like gemini is doing 00:32:31.420 --> 00:32:34.059 the same thing like They see the power of like, 00:32:34.180 --> 00:32:36.380 you know, you can add on all your personalization 00:32:36.380 --> 00:32:38.400 into Gemini now from the consumer side of things, 00:32:38.500 --> 00:32:40.859 like all your YouTube history, all your Gmail 00:32:40.859 --> 00:32:42.920 workspace and all that. Like Gemini is doing 00:32:42.920 --> 00:32:44.660 the same thing on a personal level. I'm sure 00:32:44.660 --> 00:32:46.539 they're doing that. I haven't tested on the enterprise 00:32:46.539 --> 00:32:47.960 side of things, but I'm sure it's there as well. 00:32:48.079 --> 00:32:51.579 But I think users are seeing the value of having 00:32:51.579 --> 00:32:54.220 some of that history and context as part of their 00:32:54.220 --> 00:32:57.819 AI instead of trying to build that out over a 00:32:57.819 --> 00:33:00.519 span of many months or years or sometimes. And 00:33:00.519 --> 00:33:03.079 I think there's just a lot of value. having access 00:33:03.079 --> 00:33:05.960 to some of that work data and to be able to like 00:33:05.960 --> 00:33:08.279 i i will like one of the things i do a lot is 00:33:08.279 --> 00:33:11.240 um i actually have an automation setup with copilot 00:33:11.240 --> 00:33:14.640 so every morning i have it automated so at 8 00:33:14.640 --> 00:33:16.680 a .m it gives it does a kickoff prompt and it 00:33:16.680 --> 00:33:19.230 like Tells me all the things I need to catch 00:33:19.230 --> 00:33:21.990 up on, prioritize in my outlook and stuff like 00:33:21.990 --> 00:33:24.509 that. And I see that every day. And that's all 00:33:24.509 --> 00:33:27.049 like tied to my work data and stuff like that. 00:33:27.089 --> 00:33:29.349 And I think that's a, for me, that's like eliminates 00:33:29.349 --> 00:33:31.950 30 minutes out of my morning. Oh, that's awesome. 00:33:32.109 --> 00:33:34.890 That's a great idea. Yeah. And you know, like 00:33:34.890 --> 00:33:37.329 there's so many ways to create these automation. 00:33:37.589 --> 00:33:40.150 Like there's maybe like you're using a scheduled 00:33:40.150 --> 00:33:42.809 prompt or there's like the workflows agent or 00:33:42.809 --> 00:33:45.329 there's agent builder. Because I just, I try 00:33:45.329 --> 00:33:48.220 to explain to people like, You know, there are 00:33:48.220 --> 00:33:50.720 some simple things you can do. I think now you 00:33:50.720 --> 00:33:53.180 can even go back to a prompt that you typed into 00:33:53.180 --> 00:33:55.660 chat and like right click on it and create an 00:33:55.660 --> 00:33:58.539 agent out of it. Right. So there's like, yeah. 00:33:58.940 --> 00:34:01.319 So there's like some simple things you do before 00:34:01.319 --> 00:34:03.500 you get into some of the more powerful tools 00:34:03.500 --> 00:34:07.039 like studio and whatnot. Like, and I think that 00:34:07.039 --> 00:34:10.119 one, that 8am workday thing, that's, that's a 00:34:10.119 --> 00:34:13.269 cool way to start using AI. Yeah, I'm just still 00:34:13.269 --> 00:34:15.550 talking to Copilot a lot. I've used the speak 00:34:15.550 --> 00:34:17.809 button. I do that a lot. I've been doing that 00:34:17.809 --> 00:34:20.010 a lot as well. So that's, yeah. So I used to 00:34:20.010 --> 00:34:21.769 talk to ChatGPT all the time. I would open up. 00:34:21.829 --> 00:34:24.969 So instead of using Siri, I've actually, I have 00:34:24.969 --> 00:34:28.110 the, I had the 15 Pro and I have the 17 Pro Max 00:34:28.110 --> 00:34:30.929 and it has the action button on it. So you can 00:34:30.929 --> 00:34:35.050 map it to a specific action. And at that time, 00:34:35.090 --> 00:34:37.690 ChatGPT, Gemini, and even like the consumer Copilot, 00:34:37.849 --> 00:34:40.250 all you can build a Siri shortcut to like open 00:34:40.250 --> 00:34:41.889 it up that. So instead of using Siri, because. 00:34:42.000 --> 00:34:44.219 As of right now, until they get Gemini's backend, 00:34:44.559 --> 00:34:48.420 I'm like, it's honestly pretty dumb. So I mapped 00:34:48.420 --> 00:34:50.920 out like ChatGP or Copilot and I would just use 00:34:50.920 --> 00:34:52.639 that as my AI assistant, hit the action button 00:34:52.639 --> 00:34:54.980 and I talked to it and it was great. It was awesome. 00:34:55.300 --> 00:34:58.360 I still do it today. That's cool. Yeah, my action 00:34:58.360 --> 00:35:02.559 button on my phone is, it plays this music playlist 00:35:02.559 --> 00:35:06.059 that I have because I listen to music all the 00:35:06.059 --> 00:35:08.280 time. So the first thing I wake up, when I wake 00:35:08.280 --> 00:35:12.630 up is like, I listen to music. all the way until 00:35:12.630 --> 00:35:14.409 like when I'm going to bed basically, right? 00:35:14.789 --> 00:35:18.510 But I like yours. Yours is much more productive. 00:35:19.010 --> 00:35:23.210 So the audience, if you've got a 15 Pro or higher, 00:35:23.369 --> 00:35:24.969 play around that action button. There's some 00:35:24.969 --> 00:35:26.949 really cool stuff that you can do to it. That's 00:35:26.949 --> 00:35:31.530 nice, man. So I think this has been a great conversation. 00:35:31.730 --> 00:35:35.360 Thanks for kind of... sharing all your your ideas 00:35:35.360 --> 00:35:38.360 and your thoughts and um i guess i'll just end 00:35:38.360 --> 00:35:40.960 with one question which is not even necessarily 00:35:40.960 --> 00:35:43.539 related to all of this and you know in the beginning 00:35:43.539 --> 00:35:47.150 you brought up that you have or you had a YouTube 00:35:47.150 --> 00:35:49.750 channel. So how's that going? Are you working 00:35:49.750 --> 00:35:54.150 on that? Are you going to go back to it? Yeah, 00:35:54.269 --> 00:35:56.489 it's still active. I still post a couple of things. 00:35:56.570 --> 00:36:00.210 I've been, but I've actually, when I first joined 00:36:00.210 --> 00:36:02.349 Microsoft, one of the things that they quickly 00:36:02.349 --> 00:36:04.829 got me on because they noticed, hey, you do videos. 00:36:05.230 --> 00:36:08.269 So I actually make videos on the Microsoft's 00:36:08.269 --> 00:36:10.369 HLS YouTube channel. So you'll see me there. 00:36:10.510 --> 00:36:13.030 And then I'm also, I try to keep my channel up 00:36:13.030 --> 00:36:14.659 to date. Like one of the things that, In the 00:36:14.659 --> 00:36:16.579 coming weeks, I'm going to post out is I've been 00:36:16.579 --> 00:36:19.440 at Microsoft for a year and a half. So it's I'm 00:36:19.440 --> 00:36:21.840 going to make a video on. 10 things I've learned 00:36:21.840 --> 00:36:23.780 at Microsoft since being here for over a year. 00:36:24.159 --> 00:36:26.179 Because I think there's, when I first joined 00:36:26.179 --> 00:36:28.460 Microsoft, it's a really cool company to work 00:36:28.460 --> 00:36:31.280 for. And I've learned so much in my first year. 00:36:32.460 --> 00:36:34.119 And so I wanted to kind of share that with the 00:36:34.119 --> 00:36:36.019 world. But yeah, I try to keep both of them balanced. 00:36:36.039 --> 00:36:39.619 But a lot of my time has been spent on the Microsoft's 00:36:39.619 --> 00:36:41.400 HLS YouTube channel. So you'll see me there a 00:36:41.400 --> 00:36:45.679 lot. That sounds good. So I'll put that. hls 00:36:45.679 --> 00:36:48.400 youtube and maybe your other one to your personal 00:36:48.400 --> 00:36:50.219 one i'll put both of them in the description 00:36:50.219 --> 00:36:52.400 of this podcast for anyone that wants to check 00:36:52.400 --> 00:36:55.519 out michael because he has a you know obviously 00:36:55.519 --> 00:36:57.579 he's a great speaker but he also has an awesome 00:36:57.579 --> 00:37:00.219 setup with the lighting and your background so 00:37:00.219 --> 00:37:03.900 you guys can see a real youtuber hey guys 30 00:37:03.900 --> 00:37:07.360 hue bulbs in my in my office here so i spent 00:37:07.360 --> 00:37:10.699 a lot of money at best buy that's awesome so 00:37:10.699 --> 00:37:12.920 michael thanks again man i really really appreciate 00:37:12.920 --> 00:37:16.030 it Awesome. So we'll catch you around. Thanks, 00:37:16.070 --> 00:37:16.190 David.