Episode 402 – Declarative Agents with Andrew Connell

May 22, 2025

Episode Transcript

Welcome to Episode 402 of the Microsoft Cloud IT Pro Podcast. In today’s episode, we’re thrilled to have Andrew Connell (AC) joining us once again. If you’ve been with us for a while, you’ll recognize AC as a repeat guest and one of the early voices who encouraged us to launch this podcast. In this episode, we’re diving deep into the world of Microsoft Copilot agents from a bit of a developer’s perspective . We expand on insights from Episode 400 and helping us understand some additional intricacies of how Copilot works under the hood. We’ll break down the different billing models and some limitations in what they provide and why having at least one $30 license is key to unlocking the semantic index. We’ll also explore how Copilot uses semantic indexing to ground responses in your organization’s data, and how it respects user permissions through access control lists. But it’s not all smooth sailing—AC also walks us through some of the current limitations, like missing metadata and large file indexing. He also shares a potential workaround using Azure AI Search for those needing more comprehensive coverage. Finally, we this weeks Microsoft Build Conference and tease a potential future episode where ACw will return to talk about custom Copilot development.

Andrew Connell

Your support makes this show possible! Please consider becoming a premium member for access to live shows and more. Check out our membership options.

Show Notes

About the sponsors

Would you like to become the irreplaceable Microsoft 365 resource for your organization? Let us know!

See full episode transcriptTranscript is autogenerated by AI

1 00:00:03,439 --> 00:00:06,879 Welcome to episode 402 of the Microsoft Cloud 2 00:00:06,879 --> 00:00:11,539 IT Pro podcast recorded live 05/16/2025. 3 00:00:11,919 --> 00:00:14,240 This is a show about Microsoft three sixty 4 00:00:14,240 --> 00:00:16,394 five and Azure from the perspective of IT 5 00:00:16,394 --> 00:00:18,635 pros and end users, where we discuss a 6 00:00:18,635 --> 00:00:20,954 topic or recent news and how it relates 7 00:00:20,954 --> 00:00:21,614 to you. 8 00:00:21,914 --> 00:00:24,394 This week, Ben and Scott have Andrew Connell 9 00:00:24,394 --> 00:00:25,134 or AC 10 00:00:25,515 --> 00:00:27,754 back on the show to loop back into 11 00:00:27,754 --> 00:00:31,239 some topics from episode 400 around custom agents, 12 00:00:31,300 --> 00:00:33,399 but from a bit of a developer perspective. 13 00:00:33,780 --> 00:00:35,239 They discuss some differences 14 00:00:35,619 --> 00:00:38,420 besides price in the various billing models and 15 00:00:38,420 --> 00:00:38,920 limitations 16 00:00:39,299 --> 00:00:41,879 around creation of the semantic index for Copilot. 17 00:00:42,340 --> 00:00:45,114 They also discuss how Copilot uses the semantic 18 00:00:45,114 --> 00:00:47,914 index and access control lists for security when 19 00:00:47,914 --> 00:00:51,114 retrieving your content. Another topic of conversation is 20 00:00:51,114 --> 00:00:54,494 potential issues you may encounter around large files, 21 00:00:55,034 --> 00:00:58,155 ASPX pages, and missing metadata when it comes 22 00:00:58,155 --> 00:00:58,734 to Copilot 23 00:00:59,170 --> 00:01:01,969 responses, as well as potential workarounds by using 24 00:01:01,969 --> 00:01:03,350 Azure AI search. 25 00:01:03,809 --> 00:01:05,890 Finally, they wrap up the show by talking 26 00:01:05,890 --> 00:01:07,810 about some topics you may want to pay 27 00:01:07,810 --> 00:01:11,189 attention to from this week's Microsoft Build Conference. 28 00:01:11,569 --> 00:01:13,189 Let's dive into the show. 29 00:01:15,215 --> 00:01:17,215 For all of you listening, it's not just 30 00:01:17,215 --> 00:01:19,295 Scott Knight today. There's a new I would 31 00:01:19,295 --> 00:01:21,215 say a new voice. It's kind of an 32 00:01:21,215 --> 00:01:22,814 old voice. I don't know how many times 33 00:01:24,334 --> 00:01:25,075 Wow. Just 34 00:01:25,579 --> 00:01:28,540 digging that lightning. Out swinging today. It's a 35 00:01:28,540 --> 00:01:29,359 Friday morning. 36 00:01:29,739 --> 00:01:32,379 Said, we have a popular repeat guest. Andrew 37 00:01:32,379 --> 00:01:34,939 Connell's back on the show. But then, no, 38 00:01:34,939 --> 00:01:35,979 you just went to the end of it. 39 00:01:36,060 --> 00:01:37,900 Just said it all yeah. I shoulda had 40 00:01:37,900 --> 00:01:39,439 you kick it off then, Scott. 41 00:01:40,405 --> 00:01:42,245 Now I'm all we're we're all off today. 42 00:01:42,245 --> 00:01:43,844 This is what happens on a Friday morning 43 00:01:43,844 --> 00:01:46,165 when it's been a long week. But, yeah, 44 00:01:46,165 --> 00:01:47,685 we do have Andrew Connell back on the 45 00:01:47,685 --> 00:01:50,245 show to talk more about CoPilot agents. So 46 00:01:50,245 --> 00:01:51,685 I don't know how many times we've had 47 00:01:51,685 --> 00:01:53,685 you on now, Casey. It's probably like four 48 00:01:53,685 --> 00:01:55,590 or five. Has it been that many? I 49 00:01:55,590 --> 00:01:56,790 know it's, like, one or two. I didn't 50 00:01:56,790 --> 00:01:57,510 know I didn't know if it's been I 51 00:01:57,510 --> 00:01:59,430 don't know. But we're go hey. It's a 52 00:01:59,430 --> 00:02:01,670 long time listener. Let's just say second or 53 00:02:01,670 --> 00:02:04,090 third or fourth time caller. Yeah. Exactly. 54 00:02:04,469 --> 00:02:07,030 A popular that popular caller on our show 55 00:02:07,030 --> 00:02:08,490 that just keeps coming back. 56 00:02:09,110 --> 00:02:11,025 But, no, we did, like, episode 57 00:02:11,564 --> 00:02:13,564 I think it was was it 3 no. 58 00:02:13,564 --> 00:02:16,064 It was 400 was the Copilot agent once. 59 00:02:16,125 --> 00:02:17,724 And you were listening to it, and you 60 00:02:17,724 --> 00:02:19,405 sent us a message and said, hey, Ben. 61 00:02:19,405 --> 00:02:21,405 Let's talk about some more stuff around Copilot 62 00:02:21,405 --> 00:02:22,384 agents that 63 00:02:22,750 --> 00:02:24,189 you've come to discover. And I would say 64 00:02:24,189 --> 00:02:26,989 even from a developer's perspective, like, Scott and 65 00:02:26,989 --> 00:02:29,150 I are very IT pro, which everybody knows, 66 00:02:29,150 --> 00:02:30,989 and you are very developer. But I think 67 00:02:30,989 --> 00:02:33,169 that also gives you a new perspective sometimes 68 00:02:33,310 --> 00:02:34,989 on how some of the stuff works under 69 00:02:34,989 --> 00:02:37,455 the covers. You uncover some stuff. You're like, 70 00:02:37,455 --> 00:02:39,135 Ben and Scott, let's talk about some other 71 00:02:39,135 --> 00:02:39,955 stuff that 72 00:02:40,335 --> 00:02:42,254 relates to that. So we're gonna take a 73 00:02:42,254 --> 00:02:44,435 little bit of a journey adding some 74 00:02:44,974 --> 00:02:47,455 color, some additional information to what we discussed 75 00:02:47,455 --> 00:02:50,335 on episode 400. So if you haven't listened 76 00:02:50,335 --> 00:02:51,814 to that one, you may even wanna listen 77 00:02:51,814 --> 00:02:53,550 to that one first. And then even maybe 78 00:02:53,550 --> 00:02:55,310 dive into some of the development stuff with 79 00:02:55,310 --> 00:02:57,949 Copilot agents and how you can expand Copilot 80 00:02:57,949 --> 00:02:59,409 agents even more from 81 00:02:59,789 --> 00:03:01,469 if you have access to a developer, if 82 00:03:01,469 --> 00:03:03,629 you are a developer listening to the show. 83 00:03:03,629 --> 00:03:05,655 Yeah. That sounds great. But yeah. You guys, 84 00:03:05,655 --> 00:03:07,194 first of all, like, congratulations 85 00:03:07,495 --> 00:03:09,995 on 400 episodes doing this for eight years. 86 00:03:10,294 --> 00:03:11,754 That's a huge accomplishment. 87 00:03:12,055 --> 00:03:13,735 It is I've done eight I've done a 88 00:03:13,735 --> 00:03:15,655 podcast for that long and that many episodes, 89 00:03:15,655 --> 00:03:17,254 and I will I know how big how 90 00:03:17,254 --> 00:03:19,449 much effort it is and how how much, 91 00:03:19,449 --> 00:03:21,289 like, staying power you've done with it. So 92 00:03:21,289 --> 00:03:23,050 huge kudos to you guys for doing this. 93 00:03:23,050 --> 00:03:24,489 I remember when you started this, and it's 94 00:03:24,489 --> 00:03:26,330 like it's really I I was listening to 95 00:03:26,330 --> 00:03:28,250 this on your episode 400 on the way 96 00:03:28,250 --> 00:03:29,610 to the airport, and I, like, did a 97 00:03:29,610 --> 00:03:30,969 note to myself. I'm like, oh my god. 98 00:03:30,969 --> 00:03:33,129 Gotta congratulate these guys. That's freaking awesome. I 99 00:03:33,129 --> 00:03:34,784 didn't realize they were at 400. And then 100 00:03:34,784 --> 00:03:35,664 I listened to the rest of the episode, 101 00:03:35,664 --> 00:03:36,625 and I was like, I'm gonna get to 102 00:03:36,625 --> 00:03:37,905 the airport, and I gotta send a loom 103 00:03:37,905 --> 00:03:39,504 over to to to you guys and be 104 00:03:39,504 --> 00:03:40,864 like, I got some stuff I wanna talk 105 00:03:40,864 --> 00:03:41,364 about. 106 00:03:41,985 --> 00:03:44,465 Yeah. Well, thanks. And thanks for encouraging us 107 00:03:44,465 --> 00:03:45,985 to start it because I was gonna say 108 00:03:45,985 --> 00:03:47,800 you're one of the catalysts. So Yeah. There 109 00:03:47,800 --> 00:03:50,039 you go. You, like, propelled us into doing 110 00:03:50,039 --> 00:03:51,560 this. You're like, you really should do it. 111 00:03:51,560 --> 00:03:52,840 And then you got me to go chase 112 00:03:52,840 --> 00:03:54,520 down Scott and convince Scott to do it 113 00:03:54,520 --> 00:03:57,479 with me. So thank you. And CJ as 114 00:03:57,479 --> 00:03:59,560 well. You guys were both an inspiration to 115 00:03:59,560 --> 00:04:02,655 starting it episode one eight years ago. I 116 00:04:02,655 --> 00:04:04,814 missed doing it, and I'm looking forward to 117 00:04:04,814 --> 00:04:06,734 starting another one by the time this goes 118 00:04:06,734 --> 00:04:08,174 live. I think I might have it might 119 00:04:08,174 --> 00:04:10,014 be live by then, so we'll see. Oh, 120 00:04:10,014 --> 00:04:12,094 alright. Keep our eyes out for us. Yeah. 121 00:04:12,094 --> 00:04:13,294 Put a link. We'll throw it in the 122 00:04:13,294 --> 00:04:15,830 show notes. Alright. Sounds good. Yeah. Where do 123 00:04:15,830 --> 00:04:17,850 you wanna start? Some of the stuff that 124 00:04:17,990 --> 00:04:19,589 we talked about. I don't know. There were 125 00:04:19,670 --> 00:04:21,269 there's kind of a couple things. I don't 126 00:04:21,269 --> 00:04:22,709 know if we wanna start with some of 127 00:04:22,709 --> 00:04:25,269 the stuff around deploying it, different models for 128 00:04:25,269 --> 00:04:27,050 pricing it, or if you wanna start with 129 00:04:27,110 --> 00:04:28,949 some of the other topics and what's going 130 00:04:28,949 --> 00:04:30,915 on in the background. Where if you were 131 00:04:30,915 --> 00:04:32,834 gonna explain this to somebody, where would you 132 00:04:32,834 --> 00:04:34,435 start that journey? Doctor. Let me build off 133 00:04:34,435 --> 00:04:36,274 what you guys said in episode 400 because 134 00:04:36,274 --> 00:04:37,714 there was a few things that I thought 135 00:04:37,714 --> 00:04:40,035 were not that you missed anything, but there 136 00:04:40,035 --> 00:04:41,634 was a few extra things that I've seen 137 00:04:41,634 --> 00:04:43,014 that when I talk to a customer 138 00:04:43,314 --> 00:04:45,475 that it's a it's like a you should 139 00:04:45,475 --> 00:04:46,214 know this 140 00:04:46,649 --> 00:04:49,129 before you go into things. So one of 141 00:04:49,129 --> 00:04:50,410 them was you guys were talking a little 142 00:04:50,410 --> 00:04:51,850 bit about the billing, and there was, like, 143 00:04:51,850 --> 00:04:54,410 the difference between $30 per user per month, 144 00:04:54,410 --> 00:04:56,009 or do I spend $20.200 145 00:04:56,009 --> 00:04:58,649 for 25,000 messages? Doctor. Yep. Doctor. And or 146 00:04:58,649 --> 00:05:00,250 do I go with the consumption model and 147 00:05:00,250 --> 00:05:02,204 pay a penny per message? Of course, all 148 00:05:02,204 --> 00:05:03,024 USD pricing. 149 00:05:03,404 --> 00:05:05,404 Sorry. All my numbers are USD pricing. I'm 150 00:05:05,404 --> 00:05:06,925 sure you could buy this outside of outside 151 00:05:06,925 --> 00:05:08,365 of The US. Doctor. Yeah. Doctor. There's one 152 00:05:08,365 --> 00:05:09,964 thing that I see when people do that 153 00:05:09,964 --> 00:05:11,805 that is a huge blind spot that a 154 00:05:11,805 --> 00:05:13,164 lot of people don't realize. And if they 155 00:05:13,164 --> 00:05:14,784 go with the consumption model first, 156 00:05:15,245 --> 00:05:18,519 they will have this copilot sucks type attitude 157 00:05:18,519 --> 00:05:19,479 to it because it's like, I thought it 158 00:05:19,479 --> 00:05:20,680 was supposed to be grounded in all my 159 00:05:20,680 --> 00:05:22,680 data. So there's there's two aspects to it. 160 00:05:22,680 --> 00:05:24,519 So first of all, one thing to understand, 161 00:05:24,519 --> 00:05:25,639 and I think you guys talked a little 162 00:05:25,639 --> 00:05:27,879 bit about this, is that when Microsoft says 163 00:05:27,879 --> 00:05:30,279 that it is grounded in your data and 164 00:05:30,279 --> 00:05:32,654 they always point to Microsoft Graph, that is 165 00:05:32,894 --> 00:05:35,795 kind of true. I mean, it's not incorrect. 166 00:05:36,014 --> 00:05:37,694 But to be a little bit more technical 167 00:05:37,694 --> 00:05:40,175 about it is, like in SharePoint, you put 168 00:05:40,175 --> 00:05:42,175 content in SharePoint, and then for search to 169 00:05:42,175 --> 00:05:45,055 work in SharePoint, there's an indexer that creates 170 00:05:45,055 --> 00:05:46,449 an index so so that when you do 171 00:05:46,449 --> 00:05:48,769 execute a keyword search, it doesn't look at 172 00:05:48,769 --> 00:05:50,370 the content in SharePoint. It looks at the 173 00:05:50,370 --> 00:05:52,449 index, finds the results, and it gives you 174 00:05:52,449 --> 00:05:53,970 gives you the results and it's links to 175 00:05:53,970 --> 00:05:56,550 the content in SharePoint. Yep. So that's how 176 00:05:56,769 --> 00:05:59,490 Copilot is grounding your information. So it's not 177 00:05:59,569 --> 00:06:01,410 it's doing a similar model. What it does 178 00:06:01,410 --> 00:06:04,904 is it gets data from Microsoft Graph about 179 00:06:04,904 --> 00:06:06,824 you and your organization. So that's data from 180 00:06:06,824 --> 00:06:08,045 SharePoint, data from 181 00:06:08,425 --> 00:06:11,944 chat messages, Teams transcripts, meeting transcripts, all that 182 00:06:11,944 --> 00:06:13,384 kind of stuff. And what it does is 183 00:06:13,384 --> 00:06:15,225 it does another kind of indexing called a 184 00:06:15,225 --> 00:06:18,370 semantic index, which is not keyword index, but 185 00:06:18,370 --> 00:06:20,290 a semantic index is more like let's just 186 00:06:20,290 --> 00:06:22,370 say, AI based index that is a little 187 00:06:22,370 --> 00:06:24,610 bit more these words are closer to these 188 00:06:24,610 --> 00:06:26,290 words, and it's just better for, like, AI 189 00:06:26,290 --> 00:06:28,689 based queries. What Microsoft is doing is that 190 00:06:28,689 --> 00:06:30,285 when they look at that content in graph, 191 00:06:30,365 --> 00:06:32,605 when they do a semantic index of it, 192 00:06:32,605 --> 00:06:34,605 they are storing that in what's called the 193 00:06:34,605 --> 00:06:36,685 semantic index. So, that kind of sits on 194 00:06:36,685 --> 00:06:39,084 top of Graph so that when Microsoft when 195 00:06:39,084 --> 00:06:40,464 you run a prompt 196 00:06:40,764 --> 00:06:41,584 against Copilot, 197 00:06:41,964 --> 00:06:45,770 Copilot is looking up content from the semantic 198 00:06:45,770 --> 00:06:48,730 index to get content that's relevant to you 199 00:06:48,730 --> 00:06:49,710 and your organization 200 00:06:50,170 --> 00:06:50,910 to ground 201 00:06:51,210 --> 00:06:52,029 the response 202 00:06:52,410 --> 00:06:54,170 in your data and not just like the 203 00:06:54,170 --> 00:06:56,410 foundational knowledge that the model has that like 204 00:06:56,410 --> 00:06:56,910 ChatGPT 205 00:06:57,745 --> 00:06:59,504 can leverage from, like, the the the training 206 00:06:59,504 --> 00:07:02,064 from the different models. Okay. So Office three 207 00:07:02,064 --> 00:07:04,785 sixty five search uses search index. Copilot uses 208 00:07:04,785 --> 00:07:06,884 semantic index. You've got these two indexes. 209 00:07:07,264 --> 00:07:09,264 That's correct. Yeah. And and and it's important 210 00:07:09,264 --> 00:07:11,730 to, like, call out, like, semantic index is 211 00:07:11,970 --> 00:07:14,850 technically generated with information from the graph. So 212 00:07:14,850 --> 00:07:17,410 you're, like, layers of layers in here. And 213 00:07:17,410 --> 00:07:18,949 and I think it gets a little confusing 214 00:07:19,009 --> 00:07:20,310 to customers because 215 00:07:20,770 --> 00:07:23,910 we maybe throw out terms like AI training, 216 00:07:24,050 --> 00:07:26,985 AI fine tuning, AI inferencing. Like, the reality 217 00:07:26,985 --> 00:07:29,545 is most folks aren't out there doing, like, 218 00:07:29,545 --> 00:07:31,705 AI training typically in the terms you're hearing 219 00:07:31,705 --> 00:07:33,464 maybe, like, I don't know, me, like, Band 220 00:07:33,464 --> 00:07:35,464 Aid about. Like, there there's only so many 221 00:07:35,464 --> 00:07:38,185 open AIs and anthropics and and things out 222 00:07:38,185 --> 00:07:40,205 there that are training these foundational models. 223 00:07:40,660 --> 00:07:42,180 So more and more, I think what I 224 00:07:42,180 --> 00:07:44,180 see in in my day to day is 225 00:07:44,180 --> 00:07:45,560 kind of folks who are 226 00:07:46,660 --> 00:07:49,639 leveraging existing models, and they're effectively 227 00:07:50,100 --> 00:07:50,600 into 228 00:07:51,060 --> 00:07:53,939 some type of augmentation, be it rag that's 229 00:07:53,939 --> 00:07:54,600 out there, 230 00:07:54,925 --> 00:07:56,225 be it leveraging systems 231 00:07:56,605 --> 00:07:58,305 like m three sixty five Copilot 232 00:07:58,845 --> 00:08:01,165 that then build on graph, build on semantic 233 00:08:01,165 --> 00:08:03,085 index, but then can also do rag and 234 00:08:03,085 --> 00:08:06,305 things like that for further grounding, like contextually 235 00:08:06,605 --> 00:08:08,769 in the moment versus, oh, hey. I was 236 00:08:08,769 --> 00:08:11,169 able to fine tune this model or push 237 00:08:11,169 --> 00:08:12,550 out this specific piece 238 00:08:13,089 --> 00:08:13,669 of organizational 239 00:08:13,970 --> 00:08:16,050 knowledge and and kind of do a little 240 00:08:16,050 --> 00:08:17,430 mini retraining run 241 00:08:17,889 --> 00:08:19,889 along the way. So it's all these layers 242 00:08:19,889 --> 00:08:21,555 upon layers in there. And it's funny to 243 00:08:21,555 --> 00:08:22,754 me how it all comes back to, like, 244 00:08:22,754 --> 00:08:24,615 the old days of SharePoint and, like, information 245 00:08:24,675 --> 00:08:26,754 architecture. Like, how did you organize this stuff? 246 00:08:26,754 --> 00:08:28,355 How did you permission it? How how did 247 00:08:28,355 --> 00:08:29,875 that look? And then you and then we're 248 00:08:29,875 --> 00:08:31,955 back to the same thing of, like, now 249 00:08:31,955 --> 00:08:33,475 it's not just figuring out, like, hey. Why 250 00:08:33,475 --> 00:08:35,715 did that search result surface in my search 251 00:08:35,715 --> 00:08:37,679 index because the permissions were wrong on that 252 00:08:37,679 --> 00:08:39,440 SharePoint site or that list or whatever it 253 00:08:39,440 --> 00:08:41,519 was. But now it's, oh my gosh. Where 254 00:08:41,519 --> 00:08:43,600 else was that thing potentially off the rails? 255 00:08:43,600 --> 00:08:46,240 Oh, over here in Teams, because it was 256 00:08:46,240 --> 00:08:48,159 actually in a Teams message, but it was 257 00:08:48,159 --> 00:08:50,284 in a group Teams message that I had 258 00:08:50,284 --> 00:08:52,044 going with some other folks that were out 259 00:08:52,044 --> 00:08:53,884 there. And because The Graph is able to 260 00:08:53,884 --> 00:08:54,944 surface that information, 261 00:08:55,245 --> 00:08:56,845 all of a sudden, you end up in 262 00:08:56,845 --> 00:08:57,345 this 263 00:08:57,725 --> 00:08:59,404 in this little bit of a a weird 264 00:08:59,404 --> 00:09:00,924 spot. And then you combine that with the 265 00:09:00,924 --> 00:09:02,850 fuzziness of I I don't know if you 266 00:09:02,850 --> 00:09:04,610 ever do this. I I go into teams 267 00:09:04,610 --> 00:09:06,610 all the time just or Copilot the app 268 00:09:06,610 --> 00:09:09,090 or Copilot in teams, and I say, hey. 269 00:09:09,090 --> 00:09:11,809 Search my messages for the last two weeks 270 00:09:11,809 --> 00:09:14,070 or the last six months and 271 00:09:14,625 --> 00:09:16,785 summarize blah blah blah that I discussed with, 272 00:09:16,785 --> 00:09:18,945 you you know, these people. Well, you can 273 00:09:18,945 --> 00:09:21,504 tell Copilot that it doesn't actually know how 274 00:09:21,504 --> 00:09:23,745 to execute a time based search based on 275 00:09:23,745 --> 00:09:25,745 your messages. So, like, what it pulls back 276 00:09:25,745 --> 00:09:27,745 could be completely random, could be outside the 277 00:09:27,745 --> 00:09:29,539 boundaries of that. It was all just what 278 00:09:29,539 --> 00:09:31,539 was available to it in in its grounding 279 00:09:31,539 --> 00:09:33,379 knowledge. Doctor. Right. Right. And I like, the 280 00:09:33,379 --> 00:09:34,579 way I like to the way I explain 281 00:09:34,579 --> 00:09:35,319 this to customers 282 00:09:35,620 --> 00:09:37,220 is that, like, think about if you are 283 00:09:37,220 --> 00:09:39,379 doing your taxes for your business, right? If 284 00:09:39,379 --> 00:09:41,059 you sit down, you have a question for 285 00:09:41,059 --> 00:09:43,514 an accountant, that the accountant is somebody you 286 00:09:43,514 --> 00:09:45,715 just you just hired, they have all of 287 00:09:45,715 --> 00:09:48,195 the knowledge they've learned from being trained as 288 00:09:48,195 --> 00:09:50,215 an accountant when they were in college, right, 289 00:09:50,434 --> 00:09:52,514 and just factual stuff on how QuickBooks works. 290 00:09:52,514 --> 00:09:54,529 QuickBooks works. They're gonna give you an answer 291 00:09:54,529 --> 00:09:56,209 to whatever your problem or whatever your question 292 00:09:56,209 --> 00:09:57,429 is. That's like using ChatGPT. 293 00:09:57,809 --> 00:10:00,370 What's using Microsoft three sixty five Copilot is 294 00:10:00,370 --> 00:10:02,129 he has that information or she has that 295 00:10:02,129 --> 00:10:04,529 information, but they also have all of your 296 00:10:04,529 --> 00:10:07,455 previous tax returns, and there are have access 297 00:10:07,455 --> 00:10:09,615 to your QuickBooks. So now they have context 298 00:10:09,615 --> 00:10:12,414 about you that they can add to the 299 00:10:12,414 --> 00:10:14,575 actual question. That's what the semantic index is. 300 00:10:14,575 --> 00:10:16,754 It's context about you and your organization. 301 00:10:17,375 --> 00:10:19,774 Technically, there's two semantic indexes. There's one for 302 00:10:19,774 --> 00:10:22,549 users and there's one for organizational data. So, 303 00:10:22,549 --> 00:10:24,230 but Copilot figures out which one it's gonna 304 00:10:24,230 --> 00:10:26,070 go through and grab. So, there's a few 305 00:10:26,070 --> 00:10:27,909 things about that though that the semantic index 306 00:10:27,909 --> 00:10:29,049 has some interesting 307 00:10:29,750 --> 00:10:31,750 aspects to it that people need to understand 308 00:10:31,750 --> 00:10:34,070 I think that really helps in understanding why 309 00:10:34,070 --> 00:10:36,149 Copilot can or can't do something. One of 310 00:10:36,149 --> 00:10:38,455 those things is around the billing. So, I 311 00:10:38,455 --> 00:10:40,215 talked about a second ago the difference between 312 00:10:40,215 --> 00:10:41,894 you can start with $30 per user per 313 00:10:41,894 --> 00:10:43,495 month. Doctor. Yep. Doctor. It's like the all 314 00:10:43,495 --> 00:10:45,735 you can eat plan, then there's the 25,000 315 00:10:45,735 --> 00:10:47,495 messages for $200 316 00:10:47,495 --> 00:10:49,980 a month, or it's paid per the message 317 00:10:50,059 --> 00:10:52,220 down on the consumption or metered plan. If 318 00:10:52,220 --> 00:10:55,179 you're a brand new customer to Copilot and 319 00:10:55,179 --> 00:10:57,500 you decide, I'm going to start this a 320 00:10:57,500 --> 00:10:59,100 little bit easy. I don't wanna do the 321 00:10:59,100 --> 00:11:01,019 $30 per user per month with a twelve 322 00:11:01,019 --> 00:11:03,899 month commitment, so one license is a $360 323 00:11:03,899 --> 00:11:06,085 commitment for a year. If you say, I 324 00:11:06,085 --> 00:11:08,245 don't wanna do that. Instead, I wanna do 325 00:11:08,245 --> 00:11:09,924 the consumption model and kind of ease my 326 00:11:09,924 --> 00:11:11,684 way into it. Here's the thing that people 327 00:11:11,684 --> 00:11:14,004 don't realize, and Microsoft does not document very 328 00:11:14,004 --> 00:11:16,804 well, if at all. You don't get a 329 00:11:16,804 --> 00:11:20,440 semantic index until there is at least one 330 00:11:20,500 --> 00:11:21,000 active 331 00:11:21,379 --> 00:11:24,019 Microsoft three sixty five Copilot license, the $30 332 00:11:24,019 --> 00:11:26,340 per user per month. So if you have 333 00:11:26,340 --> 00:11:28,600 10 employees and you want them using Copilot, 334 00:11:28,980 --> 00:11:31,139 if you just get the consumption model, all 335 00:11:31,139 --> 00:11:34,595 Copilot has is the all Copilot has access 336 00:11:34,595 --> 00:11:36,595 to is basically the same stuff that you 337 00:11:36,595 --> 00:11:39,154 get from ChattGPT, just the foundational knowledge from 338 00:11:39,154 --> 00:11:41,315 the training model. If you're a developer and 339 00:11:41,315 --> 00:11:42,595 you go through and extend it and do 340 00:11:42,595 --> 00:11:44,434 things like what you guys talked about, like 341 00:11:44,434 --> 00:11:46,379 adding actions to it, where you can implement 342 00:11:46,379 --> 00:11:48,059 the the RAG pattern, the retrieval of meta 343 00:11:48,059 --> 00:11:50,700 generation pattern where you where Copilot can look 344 00:11:50,700 --> 00:11:52,540 at your query or your prompt and say, 345 00:11:52,540 --> 00:11:55,100 oh, I need this extra data from this 346 00:11:55,100 --> 00:11:57,420 other endpoint over, like, in Dataverse. I can 347 00:11:57,420 --> 00:11:59,434 pull that stuff in, but it doesn't have 348 00:11:59,434 --> 00:12:01,674 any it won't have any knowledge about your 349 00:12:01,674 --> 00:12:03,774 chats, your your transcripts, 350 00:12:04,154 --> 00:12:06,074 your calendar, your meetings, stuff like that. It 351 00:12:06,074 --> 00:12:08,235 won't have that context because the semantic index 352 00:12:08,235 --> 00:12:10,414 does not exist. It is not until 353 00:12:10,794 --> 00:12:13,195 you decide to spend, I will have one 354 00:12:13,195 --> 00:12:15,809 person get that $30 per user per month 355 00:12:15,809 --> 00:12:17,809 of my 10 employees, and then I'll do 356 00:12:17,809 --> 00:12:20,450 consumption, and it completely changes the experience for 357 00:12:20,450 --> 00:12:23,009 everybody that has that now has that consumption 358 00:12:23,009 --> 00:12:25,809 model in your organization. So now it's like 359 00:12:25,809 --> 00:12:26,850 a lot of people look at this and 360 00:12:26,850 --> 00:12:28,565 go, like, Copilot sucks. It's no better than 361 00:12:28,565 --> 00:12:30,485 ChatGPT. It doesn't have any it's not grounding 362 00:12:30,485 --> 00:12:31,845 any of my any of its knowledge and 363 00:12:31,845 --> 00:12:33,764 stuff that's with my organization. It's because it 364 00:12:33,764 --> 00:12:35,924 doesn't have the semantic index, the trigger to 365 00:12:35,924 --> 00:12:37,784 create it, one act of life. So 366 00:12:38,164 --> 00:12:40,004 that's bizarre. So because I've never tried this 367 00:12:40,004 --> 00:12:41,940 with consumption models. So if I went in, 368 00:12:42,019 --> 00:12:43,539 bought one of the consumption model or the 369 00:12:43,539 --> 00:12:46,279 message pack, went into, like, Teams chat Copilot 370 00:12:46,340 --> 00:12:48,580 and said, go give me this information about 371 00:12:48,580 --> 00:12:50,899 my email or people I email or documents 372 00:12:50,899 --> 00:12:52,659 in SharePoint, it's not gonna return any of 373 00:12:52,659 --> 00:12:54,980 that. I can't speak for certain about the 374 00:12:54,980 --> 00:12:55,779 25,000 375 00:12:55,860 --> 00:12:57,915 the message packs. Okay. But I can say 376 00:12:57,915 --> 00:12:59,995 on the consumption model, yes, that is correct. 377 00:12:59,995 --> 00:13:01,995 The penny per message one. The the penny 378 00:13:01,995 --> 00:13:04,235 per message one does not turn on a 379 00:13:04,235 --> 00:13:06,575 semantic index. The thing that turns it on 380 00:13:06,795 --> 00:13:08,975 is the the $30 per user per month 381 00:13:09,035 --> 00:13:11,990 license. A message pack might also do that. 382 00:13:12,129 --> 00:13:14,289 I think it does, but I can't be 383 00:13:14,289 --> 00:13:15,889 certain on that. Again, Microsoft doesn't do a 384 00:13:15,889 --> 00:13:17,490 good job of documenting this. You gotta find 385 00:13:17,490 --> 00:13:19,009 the right person and catch them at the 386 00:13:19,009 --> 00:13:21,009 right time and and get the answer. I 387 00:13:21,009 --> 00:13:22,610 have not that part I haven't I don't 388 00:13:22,610 --> 00:13:25,144 have a solid answer on. My belief is 389 00:13:25,144 --> 00:13:27,065 that a message pack does also do that 390 00:13:27,065 --> 00:13:28,745 because a message pack is also gonna have, 391 00:13:28,745 --> 00:13:30,024 like they also talk about if you have 392 00:13:30,024 --> 00:13:31,544 a message pack, it's gonna be, like, fifteen 393 00:13:31,544 --> 00:13:33,065 fifteen messages just to go through and ground 394 00:13:33,065 --> 00:13:34,504 data because that's how much a query is 395 00:13:34,504 --> 00:13:37,404 gonna cost to the semantic index. Right. So 396 00:13:37,549 --> 00:13:40,029 I I think so, but I'm not certain. 397 00:13:40,029 --> 00:13:42,990 And it's I'm assuming maybe. Again, this is 398 00:13:42,990 --> 00:13:45,309 not anything that's documented having to do with 399 00:13:45,309 --> 00:13:47,649 that cost. If you're paying $200 a month, 400 00:13:47,789 --> 00:13:49,549 that probably helps to offset the cost of 401 00:13:49,549 --> 00:13:52,029 building the semantic index. $30 a month, same 402 00:13:52,029 --> 00:13:53,615 thing. If you're just paying a penny per 403 00:13:53,615 --> 00:13:55,134 message, you're like, how do we recover the 404 00:13:55,134 --> 00:13:57,934 cost of having the semantic index sitting out 405 00:13:57,934 --> 00:13:59,235 there because 406 00:13:59,615 --> 00:14:01,455 you're only paying a penny per message? I 407 00:14:01,615 --> 00:14:03,794 that would be my speculation as to maybe 408 00:14:03,855 --> 00:14:05,759 the rationale, but I agree with you. Like, 409 00:14:05,840 --> 00:14:06,740 it should be right 410 00:14:07,519 --> 00:14:09,759 in that pricing table of, with this, you 411 00:14:09,759 --> 00:14:11,919 do not get a semantic index or grounded 412 00:14:11,919 --> 00:14:14,320 in your data with these two. Like, just 413 00:14:14,320 --> 00:14:15,940 give us a table and show us that. 414 00:14:16,000 --> 00:14:18,240 I completely agree with you on that point. 415 00:14:18,240 --> 00:14:20,080 That part to me is very frustrating. That 416 00:14:20,080 --> 00:14:21,540 was not it's not as clear. 417 00:14:21,955 --> 00:14:22,995 One of the other reasons why I say 418 00:14:22,995 --> 00:14:24,754 I think it's only limited to the consumption 419 00:14:24,754 --> 00:14:26,995 model is that this wasn't known until the 420 00:14:26,995 --> 00:14:28,294 consumption model was announced 421 00:14:28,754 --> 00:14:31,075 and when it wasn't even really known at 422 00:14:31,075 --> 00:14:33,075 that time, but that'severything kind of points back 423 00:14:33,075 --> 00:14:34,835 to that being like the big deal. So, 424 00:14:34,835 --> 00:14:36,355 Ithat's why I think the message pack does 425 00:14:36,355 --> 00:14:37,899 have it, but again, I'm not a % 426 00:14:37,899 --> 00:14:39,120 certain. Got it. 427 00:14:42,779 --> 00:14:44,860 Do you feel overwhelmed by trying to manage 428 00:14:44,860 --> 00:14:47,180 your Office three sixty five environment? Are you 429 00:14:47,180 --> 00:14:50,480 facing unexpected issues that disrupt your company's productivity? 430 00:14:50,779 --> 00:14:52,584 Intelligink is here to help, much like like 431 00:14:52,584 --> 00:14:54,264 you take your car to the mechanic that 432 00:14:54,264 --> 00:14:56,504 has specialized knowledge on how to best keep 433 00:14:56,504 --> 00:14:59,225 your car running, Intelligink helps you with your 434 00:14:59,225 --> 00:15:01,964 Microsoft cloud environment because that's their expertise. 435 00:15:02,345 --> 00:15:04,679 Intelligink keeps up with the latest updates in 436 00:15:04,679 --> 00:15:06,840 the Microsoft cloud to help keep your business 437 00:15:06,840 --> 00:15:09,080 running smoothly and ahead of the curve. Whether 438 00:15:09,080 --> 00:15:11,080 you are a small organization with just a 439 00:15:11,080 --> 00:15:13,639 few users up to an organization of several 440 00:15:13,639 --> 00:15:14,539 thousand employees, 441 00:15:14,919 --> 00:15:16,919 they want to partner with you to implement 442 00:15:16,919 --> 00:15:17,659 and administer 443 00:15:17,960 --> 00:15:19,659 your Microsoft cloud technology. 444 00:15:20,414 --> 00:15:23,875 Visit them at inteligync.com/podcast. 445 00:15:24,174 --> 00:15:30,914 That's intelligink.com/podcast 446 00:15:31,294 --> 00:15:33,375 for more information or to schedule a thirty 447 00:15:33,375 --> 00:15:35,475 minute call to get started with them today. 448 00:15:35,840 --> 00:15:39,120 Remember, Intelligink focuses on the Microsoft cloud so 449 00:15:39,120 --> 00:15:40,899 you can focus on your business. 450 00:15:43,040 --> 00:15:45,279 I'm curious. This is not a question that 451 00:15:45,279 --> 00:15:47,120 we necessarily talked about. But I know, like, 452 00:15:47,120 --> 00:15:49,120 with sensitivity labels, you're going in and starting 453 00:15:49,120 --> 00:15:52,455 to exclude SharePoint sites from Copilot, that type 454 00:15:52,455 --> 00:15:55,335 of stuff. It also affects SharePoint search, but, 455 00:15:55,335 --> 00:15:57,254 yeah, Copilot and SharePoint search are pulling from 456 00:15:57,254 --> 00:15:58,235 two different indexes. 457 00:15:58,774 --> 00:16:01,254 Is it because the semantic index is still 458 00:16:01,254 --> 00:16:02,475 using the search index 459 00:16:02,774 --> 00:16:04,855 to build it, or is it just because 460 00:16:04,855 --> 00:16:05,914 some of the functionality 461 00:16:06,615 --> 00:16:07,320 is the 462 00:16:07,720 --> 00:16:09,160 same? I think it's the functionality is the 463 00:16:09,160 --> 00:16:11,639 same. I don't know if they're also using 464 00:16:11,639 --> 00:16:13,559 SharePoint as a way to populate the semantic 465 00:16:13,559 --> 00:16:15,559 index. They they may be, but I don't 466 00:16:15,559 --> 00:16:16,519 I don't think I got it. You don't 467 00:16:16,519 --> 00:16:18,929 know or looked into that why that's the 468 00:16:18,929 --> 00:16:20,595 case. Well, no, and I and I also 469 00:16:20,595 --> 00:16:21,715 don't think that they would be using the 470 00:16:21,715 --> 00:16:24,355 keyword indexing for that, for the search index 471 00:16:24,355 --> 00:16:26,274 that they've created for doing like SharePoint search 472 00:16:26,274 --> 00:16:27,955 or three sixty five search. Right. I think 473 00:16:27,955 --> 00:16:29,315 they would actually be going to the content, 474 00:16:29,315 --> 00:16:31,475 but that that actually brings up another interesting 475 00:16:31,475 --> 00:16:33,235 aspect too that that I that I thought 476 00:16:33,235 --> 00:16:35,610 was something that also catches people. And you 477 00:16:35,610 --> 00:16:36,889 so you guys, I think there's a little 478 00:16:36,889 --> 00:16:38,110 bit of a question around 479 00:16:38,570 --> 00:16:41,450 security and permissions. So like if like who 480 00:16:41,450 --> 00:16:43,929 has access to the content? Does Copilot have 481 00:16:43,929 --> 00:16:45,769 access? Does Copilot and the user have access? 482 00:16:45,769 --> 00:16:47,370 Like how does that work? When it comes 483 00:16:47,370 --> 00:16:49,705 to the semantic index, it's the same as 484 00:16:49,705 --> 00:16:51,384 think of the same way as how SharePoint 485 00:16:51,384 --> 00:16:53,865 works. So when something goes into the semantic 486 00:16:53,865 --> 00:16:56,105 index, you have the ability to add an 487 00:16:56,105 --> 00:16:58,184 ACL to it, an access control list, the 488 00:16:58,184 --> 00:17:00,445 same way we do with, like, SharePoint search. 489 00:17:00,504 --> 00:17:03,419 So when something is indexed, it has an 490 00:17:03,419 --> 00:17:03,919 ACL 491 00:17:04,539 --> 00:17:06,559 attached to it or a collection of ACLs, 492 00:17:06,619 --> 00:17:08,319 and that collection says 493 00:17:08,779 --> 00:17:10,960 these people or these groups 494 00:17:11,420 --> 00:17:13,740 have access or do not have access to 495 00:17:13,740 --> 00:17:17,119 this. So when Copilot does a query against 496 00:17:17,259 --> 00:17:18,400 the semantic index, 497 00:17:18,715 --> 00:17:20,955 Copilot's making the query, but it's making that 498 00:17:20,955 --> 00:17:21,455 query 499 00:17:21,835 --> 00:17:24,235 on behalf of the user that submitted the 500 00:17:24,235 --> 00:17:26,235 prompt. Okay. So it's getting so the data 501 00:17:26,235 --> 00:17:28,715 that Copilot gets back from semantic index is 502 00:17:28,715 --> 00:17:30,795 security trimmed for the user who issued the 503 00:17:30,795 --> 00:17:32,619 prompt. The same it Got it. Basically the 504 00:17:32,619 --> 00:17:34,140 same way how SharePoint search works. If I 505 00:17:34,140 --> 00:17:35,579 don't have access to a document and I 506 00:17:35,579 --> 00:17:36,940 do a search, I won't see it. You 507 00:17:36,940 --> 00:17:38,460 won't see it. It's just that the part 508 00:17:38,460 --> 00:17:41,259 that's me is now that is Copilot is 509 00:17:41,259 --> 00:17:43,904 doing it for me. So another question then, 510 00:17:44,065 --> 00:17:45,904 how quick is the semantic index? Because like 511 00:17:45,904 --> 00:17:47,585 SharePoint search index, right, you can have the 512 00:17:47,585 --> 00:17:50,384 continuous index on so that when you change 513 00:17:50,384 --> 00:17:50,884 security, 514 00:17:51,424 --> 00:17:53,265 it's I mean, let's be honest. It's not 515 00:17:53,265 --> 00:17:55,345 always instantaneous because it still has to re 516 00:17:55,345 --> 00:17:57,285 index the content to re pull in ACLs 517 00:17:57,345 --> 00:17:59,799 when you change security to it. Yeah. Semantic 518 00:17:59,799 --> 00:18:00,299 index, 519 00:18:00,759 --> 00:18:02,680 any ideas? I think it's kind of the 520 00:18:02,680 --> 00:18:04,279 same delay as, like, how long it takes 521 00:18:04,279 --> 00:18:05,400 to share point to get in. And the 522 00:18:05,400 --> 00:18:06,680 other part is, like, how long is it 523 00:18:06,759 --> 00:18:08,759 how how fast is it actually querying that 524 00:18:08,759 --> 00:18:10,200 when you issue a prompt to Copilot and 525 00:18:10,200 --> 00:18:12,119 it's like dot dot dot Copilot's thinking. Yeah. 526 00:18:12,119 --> 00:18:13,340 That's how fast it is. 527 00:18:14,315 --> 00:18:15,914 That's one of the reasons why it's like 528 00:18:15,914 --> 00:18:17,515 it's going to the semantic index. It's pulling 529 00:18:17,515 --> 00:18:18,795 some data back. It's going well, first, it 530 00:18:18,795 --> 00:18:20,714 goes to LLM, figuring out, like, what is 531 00:18:20,714 --> 00:18:22,555 the user's intent? And that tells it, Oh, 532 00:18:22,555 --> 00:18:23,755 I need to go over and find out 533 00:18:23,835 --> 00:18:26,575 find calendar message find calendar invites or transcripts 534 00:18:26,714 --> 00:18:27,775 or emails. 535 00:18:28,119 --> 00:18:30,119 It grabs that. It uses that as part 536 00:18:30,119 --> 00:18:31,480 of like it says that it's grounding that 537 00:18:31,480 --> 00:18:33,960 knowledge, but it it is. But that's actually 538 00:18:33,960 --> 00:18:36,200 pulling data to give it a like a 539 00:18:36,200 --> 00:18:37,799 a implementing the rag pattern on the fly. 540 00:18:37,799 --> 00:18:39,980 So, it's actually CopaD's going to the LLM 541 00:18:40,440 --> 00:18:44,054 multiple times, multiple round trips. So, that's one. 542 00:18:44,054 --> 00:18:46,554 So, another one though that's actually big is 543 00:18:46,615 --> 00:18:47,894 so that's one about the the whole thing 544 00:18:47,894 --> 00:18:50,295 about the the semantic index getting created Doctor. 545 00:18:50,295 --> 00:18:51,815 Yep. Doctor. And then also the content that's 546 00:18:51,815 --> 00:18:53,355 going in. Here's another 547 00:18:53,654 --> 00:18:55,115 I I'm gonna call it a limitation, 548 00:18:55,490 --> 00:18:56,849 but make sure you listen to the whole 549 00:18:56,849 --> 00:18:58,130 part of this because there's like a big 550 00:18:58,130 --> 00:19:00,690 story here. Doctor. Okay. Doctor. So when let's 551 00:19:00,690 --> 00:19:03,730 say I put a 300 page PDF into 552 00:19:03,730 --> 00:19:05,430 a SharePoint document library, 553 00:19:05,809 --> 00:19:09,009 it's getting indexed into the semantic index, but 554 00:19:09,009 --> 00:19:09,509 only 555 00:19:09,970 --> 00:19:11,325 the first percentage 556 00:19:11,705 --> 00:19:14,184 of that file. And now, I'm going to 557 00:19:14,184 --> 00:19:16,505 use some numbers here, but these numbers are 558 00:19:16,505 --> 00:19:19,144 always getting bigger. And so, just take this 559 00:19:19,144 --> 00:19:21,785 as an example, and this was true months 560 00:19:21,785 --> 00:19:24,179 ago, but I'm sure that when the time 561 00:19:24,179 --> 00:19:26,579 this episode comes out, when Build has happened, 562 00:19:26,579 --> 00:19:28,019 they would have made another announcement, so the 563 00:19:28,019 --> 00:19:30,339 numbers may be bigger. So, when if I 564 00:19:30,339 --> 00:19:32,419 have a 300 page PDF in there, and 565 00:19:32,419 --> 00:19:34,980 let's say deep in that PDF, there are 566 00:19:34,980 --> 00:19:37,220 lots of tables and good like documentation stuff 567 00:19:37,220 --> 00:19:38,679 in it and like real statistics, 568 00:19:39,345 --> 00:19:40,085 Only the 569 00:19:40,544 --> 00:19:43,585 first, let's say, 25 pages are getting indexed 570 00:19:43,585 --> 00:19:45,825 in the semantic index. Only the first four 571 00:19:45,825 --> 00:19:47,984 megs, essentially, of a piece of content is 572 00:19:47,984 --> 00:19:49,664 getting indexed. Now, granted, if I got 300 573 00:19:49,664 --> 00:19:52,144 page PDF, usually only about 15% or 20% 574 00:19:52,144 --> 00:19:53,700 of that is actual content, and the rest 575 00:19:53,700 --> 00:19:55,000 of it is all like packaging 576 00:19:55,460 --> 00:19:57,059 and Adobe PDF crap and all that kind 577 00:19:57,059 --> 00:19:58,900 of stuff. Same thing with Word Docs and 578 00:19:58,900 --> 00:20:00,500 PowerPoints. But a lot of people look at 579 00:20:00,500 --> 00:20:02,339 this, they say, I know the answer is 580 00:20:02,339 --> 00:20:04,500 in that PDF. And let's just say it's 581 00:20:04,500 --> 00:20:07,065 on page two seventy four. But when I 582 00:20:07,065 --> 00:20:08,825 ask Copilot, it doesn't know the answer. It's 583 00:20:08,825 --> 00:20:11,325 because the semantic index doesn't get the entire 584 00:20:11,625 --> 00:20:14,284 file. It's limited. And it's just part because 585 00:20:14,505 --> 00:20:16,825 we're early with all this AI stuff, and 586 00:20:16,825 --> 00:20:18,900 Microsoft is still working to get the things 587 00:20:18,900 --> 00:20:20,740 big to make the the the capabilities of 588 00:20:20,740 --> 00:20:22,339 what it can put in the index much 589 00:20:22,339 --> 00:20:24,519 bigger. So I I think it's big today. 590 00:20:24,900 --> 00:20:26,740 You you're already at 512 591 00:20:26,740 --> 00:20:29,619 megabytes today. That four meg has been surpassed. 592 00:20:29,619 --> 00:20:31,539 So if that goes even higher, like, I 593 00:20:31,539 --> 00:20:33,375 don't know. I think about the majority 594 00:20:33,914 --> 00:20:34,575 of organizational 595 00:20:34,875 --> 00:20:37,194 content that you work with is not many 596 00:20:37,194 --> 00:20:39,434 of us are working with 500 meg word 597 00:20:39,434 --> 00:20:42,494 documents, maybe Excel sheets, thing things like that. 598 00:20:42,634 --> 00:20:44,634 But that that number always gets bigger and 599 00:20:44,634 --> 00:20:46,554 bigger, probably up to the limitation of just 600 00:20:46,554 --> 00:20:48,494 file size in SharePoint at some point. 601 00:20:48,859 --> 00:20:51,019 But context windows need to grow. I I 602 00:20:51,019 --> 00:20:53,340 totally agree. I like context there. I actually 603 00:20:53,340 --> 00:20:54,940 have a one of my coaching clients is, 604 00:20:55,100 --> 00:20:58,160 in the, let's just say, the national security 605 00:20:58,380 --> 00:21:00,859 industry in The United States, and they're involved 606 00:21:00,859 --> 00:21:02,855 with the Department of Defense, and 607 00:21:03,154 --> 00:21:04,914 let's just say nuclear is involved in it 608 00:21:04,914 --> 00:21:07,634 as well. And they have some PDFs that 609 00:21:07,634 --> 00:21:09,315 are like three and four gigs of, like, 610 00:21:09,315 --> 00:21:11,154 nuclear safety stuff, and they're like, Why are 611 00:21:11,154 --> 00:21:13,474 we not seeing this stuff? So what we 612 00:21:13,474 --> 00:21:15,750 had to do instead to fix that was 613 00:21:15,750 --> 00:21:17,269 instead of just putting the stuff inside of 614 00:21:17,269 --> 00:21:19,349 a SharePoint document library, which they weren't doing 615 00:21:19,349 --> 00:21:21,349 anyway, but that's a whole head Snowden kind 616 00:21:21,349 --> 00:21:22,250 of like be careful. 617 00:21:23,109 --> 00:21:24,470 Instead, what they had to do is they 618 00:21:24,470 --> 00:21:26,970 had to create a custom process that would 619 00:21:27,029 --> 00:21:29,049 take all content out of that PDF. 620 00:21:29,375 --> 00:21:31,475 They would then create the embeddings 621 00:21:31,855 --> 00:21:33,634 and then store it in a vector database, 622 00:21:34,015 --> 00:21:34,674 and then 623 00:21:35,055 --> 00:21:37,055 they could go through and find all the 624 00:21:37,055 --> 00:21:39,215 content based on the user's prompt. Best way 625 00:21:39,215 --> 00:21:41,134 to do that, use Azure AI search. It's 626 00:21:41,134 --> 00:21:43,259 another resource. And you can hook that up 627 00:21:43,259 --> 00:21:45,980 to a Copilot based agent, or you can 628 00:21:45,980 --> 00:21:48,140 write a REST API that a declarative agent 629 00:21:48,140 --> 00:21:49,980 that a developer could write to go get 630 00:21:49,980 --> 00:21:52,240 that data, and now you have full visibility 631 00:21:52,460 --> 00:21:54,620 in the document. The big difference there, you 632 00:21:54,620 --> 00:21:56,620 as the developer, you have to pay for 633 00:21:56,620 --> 00:21:57,920 the Azure AI search 634 00:21:58,275 --> 00:21:59,955 resource and how much which is a a 635 00:21:59,955 --> 00:22:02,115 metered based thing. So how much you're putting 636 00:22:02,115 --> 00:22:03,555 in it and how much you're actually using 637 00:22:03,555 --> 00:22:05,795 it. Whereas, if you're doing a semantic index, 638 00:22:05,795 --> 00:22:07,715 that's part of the $30 per user per 639 00:22:07,715 --> 00:22:09,154 month all you can eat plan. Right? So 640 00:22:09,315 --> 00:22:11,154 Yeah. That's a there's a way around that. 641 00:22:11,154 --> 00:22:13,319 And, also, the Azure AI search doesn't have 642 00:22:13,480 --> 00:22:15,079 the security trimming because they don't have the 643 00:22:15,079 --> 00:22:16,539 capability to do ACLs 644 00:22:16,919 --> 00:22:19,240 on the stuff today. So that's a little 645 00:22:19,399 --> 00:22:21,319 Hey. It's it's a you can't have your 646 00:22:21,319 --> 00:22:23,720 cake. You need it too Right. Depending depending 647 00:22:23,720 --> 00:22:24,759 on where you are. I I think there 648 00:22:24,759 --> 00:22:26,679 is friction there. Like, one of the exciting 649 00:22:26,679 --> 00:22:29,934 things about context windows growing and effectively this 650 00:22:29,934 --> 00:22:32,654 concept of, like, quote, unquote, AI memory, right, 651 00:22:32,654 --> 00:22:34,515 and and chat memory and all these things 652 00:22:34,815 --> 00:22:36,974 is you're not gonna have to do that 653 00:22:36,974 --> 00:22:37,954 extra hop 654 00:22:38,335 --> 00:22:39,315 of potentially 655 00:22:39,775 --> 00:22:41,909 multiple chunks. Like, I I do run into 656 00:22:41,909 --> 00:22:44,409 customers in this situation today to really massive 657 00:22:44,470 --> 00:22:46,869 PDFs, things like that, where they have to 658 00:22:46,869 --> 00:22:48,950 pre chunk, and then they have to go 659 00:22:48,950 --> 00:22:51,109 and vectorize that to generate they have to 660 00:22:51,109 --> 00:22:53,109 go generate embeddings, store all that stuff in 661 00:22:53,109 --> 00:22:56,105 a vector database, be it Pinecone, AI 662 00:22:56,404 --> 00:22:58,265 search, whatever it happens to be, 663 00:22:58,565 --> 00:23:00,565 and manage it through. But if you've got 664 00:23:00,565 --> 00:23:02,404 some maturity and, like, you're there and you 665 00:23:02,404 --> 00:23:04,484 understand that stuff and you're kind of on 666 00:23:04,484 --> 00:23:07,845 the, quote, unquote, cutting edge, like, sure. It's 667 00:23:07,845 --> 00:23:08,505 right there. 668 00:23:08,805 --> 00:23:11,099 It it is available for you. I've always 669 00:23:11,099 --> 00:23:13,980 been surprised, like, even watching, like, PMs on 670 00:23:13,980 --> 00:23:15,259 our team be able to pick some of 671 00:23:15,259 --> 00:23:17,180 that stuff up and and just run with 672 00:23:17,180 --> 00:23:19,579 it, like, out of the box. Well, not 673 00:23:19,579 --> 00:23:21,259 out of the box. Like but the the 674 00:23:21,259 --> 00:23:23,039 number of us now that are sitting here 675 00:23:23,325 --> 00:23:25,404 writing Python scripts on the side to do 676 00:23:25,404 --> 00:23:28,304 weird machinations, like, it it it's ever growing 677 00:23:29,085 --> 00:23:30,924 as as as we get up here and 678 00:23:30,924 --> 00:23:32,065 and we do these things. 679 00:23:32,365 --> 00:23:33,884 We we've been we've been prepping for our 680 00:23:33,884 --> 00:23:35,724 build session all week and thinking about, like, 681 00:23:35,724 --> 00:23:37,420 oh, which model do we show? How are 682 00:23:37,420 --> 00:23:39,440 we gonna demonstrate checkpointing here? 683 00:23:39,820 --> 00:23:41,579 What what are we gonna do in this 684 00:23:41,579 --> 00:23:43,339 scenario, in this thing to to really show 685 00:23:43,339 --> 00:23:45,259 it off where it shines? I'm not at 686 00:23:45,259 --> 00:23:46,619 the point of writing my own Python scripts 687 00:23:46,619 --> 00:23:47,820 yet. I am at the point of asking 688 00:23:47,820 --> 00:23:49,339 chat gbt telling it what I want, and 689 00:23:49,339 --> 00:23:50,700 it writes the Python script for me. But 690 00:23:50,700 --> 00:23:51,900 I'm not I'm not the point of writing 691 00:23:51,900 --> 00:23:54,225 my own. Yeah. Well, nobody should be writing 692 00:23:54,225 --> 00:23:56,225 Python from scratch anyway these days. Let let 693 00:23:56,225 --> 00:23:58,144 me yeah. I do it for you. Yeah. 694 00:23:58,144 --> 00:23:59,664 Those were the really big things that I 695 00:23:59,664 --> 00:24:01,345 saw, and I know they're not exactly dev 696 00:24:01,345 --> 00:24:03,025 stuff, but I I don't know what else 697 00:24:03,025 --> 00:24:04,384 you guys wanna talk about in terms of, 698 00:24:04,384 --> 00:24:06,970 like, dev specific things. So I have another 699 00:24:06,970 --> 00:24:09,369 question that I think you said this in 700 00:24:09,369 --> 00:24:11,369 the video. I think we're okay to talk 701 00:24:11,369 --> 00:24:12,190 about this. 702 00:24:12,490 --> 00:24:14,490 Well, it's something that that This is the 703 00:24:14,490 --> 00:24:16,569 last podcast I'll have my my my MVP 704 00:24:16,569 --> 00:24:18,409 NDA. Well, that's why we can edit. Right? 705 00:24:18,409 --> 00:24:20,250 Todd's the only one listening. Todd, if you 706 00:24:20,250 --> 00:24:22,214 hear this is NDA NDA or anything about 707 00:24:22,214 --> 00:24:24,615 this is NDA, don't I know Todd. Todd 708 00:24:24,615 --> 00:24:27,015 will be good. Like, we talked about SharePoint 709 00:24:27,015 --> 00:24:29,654 index, surfacing content, size of files. One of 710 00:24:29,654 --> 00:24:31,815 the things you used to do or you 711 00:24:31,815 --> 00:24:34,055 still do to help with SharePoint search and 712 00:24:34,055 --> 00:24:36,119 digging through libraries and lists and all that 713 00:24:36,200 --> 00:24:37,579 is metadata on these files 714 00:24:37,880 --> 00:24:38,779 to be able 715 00:24:39,319 --> 00:24:41,160 to do, like, filters on it and query 716 00:24:41,160 --> 00:24:42,359 on it. And then you get a query 717 00:24:42,359 --> 00:24:43,720 and you get your filter down the left 718 00:24:43,720 --> 00:24:45,319 side with your metadata and doing all of 719 00:24:45,319 --> 00:24:47,420 that. How does Copilot handle 720 00:24:47,799 --> 00:24:48,299 metadata 721 00:24:49,375 --> 00:24:50,934 and files? That's a great question. It's a 722 00:24:50,934 --> 00:24:51,634 great answer. 723 00:24:52,095 --> 00:24:52,674 It doesn't. 724 00:24:53,855 --> 00:24:56,734 So here's the thing. Copilot only okay. So 725 00:24:56,734 --> 00:24:58,494 it's not so much Copilot. Remember, all goes 726 00:24:58,494 --> 00:25:01,054 back to semantic Right. Semantic index. The semantic 727 00:25:01,054 --> 00:25:04,890 index only has the content from files inside 728 00:25:04,890 --> 00:25:07,630 of SharePoint and OneDrive. It does not have 729 00:25:07,769 --> 00:25:10,410 content related to the metadata on those files 730 00:25:10,410 --> 00:25:12,410 in a document library. It does not have 731 00:25:12,410 --> 00:25:14,650 data from SharePoint lists. It does not have 732 00:25:14,650 --> 00:25:16,109 data from SharePoint pages. 733 00:25:16,454 --> 00:25:18,855 Microsoft knows all of this. They that is 734 00:25:18,934 --> 00:25:20,454 might even has changed by the time I'm 735 00:25:20,454 --> 00:25:22,294 saying this. It might even change by the 736 00:25:22,294 --> 00:25:24,694 time you're hearing this, but especially because build 737 00:25:24,694 --> 00:25:26,534 is next week because it's a huge ask 738 00:25:26,534 --> 00:25:28,694 for people. But I know, like, list data 739 00:25:28,694 --> 00:25:30,740 has been a huge blind spot when it 740 00:25:30,740 --> 00:25:32,980 comes to m three sixty five Copilot because 741 00:25:32,980 --> 00:25:35,059 it is that content is not being indexed 742 00:25:35,059 --> 00:25:38,119 into the semantic index. Only the files inside 743 00:25:38,180 --> 00:25:40,500 of a document library in OneDrive are being 744 00:25:40,500 --> 00:25:42,420 indexed. Okay. And not even so not even 745 00:25:42,420 --> 00:25:44,840 ASPX pages because technically, I think of ASPX 746 00:25:44,900 --> 00:25:47,605 pages and, yeah, you look at companies' intranets 747 00:25:47,664 --> 00:25:49,044 or I know companies, 748 00:25:49,505 --> 00:25:51,845 good idea or not, have built, like, whole 749 00:25:51,984 --> 00:25:54,625 wikis and all their documentation in ASPX pages 750 00:25:54,625 --> 00:25:56,565 because they want it in that format. 751 00:25:56,944 --> 00:25:59,190 And then so they could also be very 752 00:25:59,190 --> 00:26:01,769 surprised when they go ask Copilot, and Copilot's, 753 00:26:01,990 --> 00:26:04,390 like, missing data. Maybe it's missing data from 754 00:26:04,390 --> 00:26:06,890 Power Apps that are writing to SharePoint lists 755 00:26:06,950 --> 00:26:09,509 or Yeah. You're a developer or SPFX. I 756 00:26:09,509 --> 00:26:10,329 can do SPFX 757 00:26:10,774 --> 00:26:12,774 writing the SharePoint list. None of that's gonna 758 00:26:12,774 --> 00:26:13,434 come into 759 00:26:13,734 --> 00:26:14,234 Copilot. 760 00:26:14,615 --> 00:26:16,294 That's correct. But, again, I just wanna stress 761 00:26:16,294 --> 00:26:18,294 one thing, and that is that is the 762 00:26:18,294 --> 00:26:20,554 way it at least was 763 00:26:21,014 --> 00:26:21,514 recently. 764 00:26:21,974 --> 00:26:24,134 And I would whenever something like this comes 765 00:26:24,134 --> 00:26:25,755 up and cope with with, like, a limitation, 766 00:26:25,894 --> 00:26:27,950 it's always, I'm filing this back in my 767 00:26:27,950 --> 00:26:29,470 head. When I do a search, something's not 768 00:26:29,470 --> 00:26:31,789 happening. It's like, I remember that guy told 769 00:26:31,789 --> 00:26:33,390 me this thing at one point. Yeah. I 770 00:26:33,390 --> 00:26:34,590 wonder if that's going on. Let me go 771 00:26:34,590 --> 00:26:36,830 research or go to your research and be 772 00:26:36,830 --> 00:26:38,670 like, well, no. Andrew's wrong. I just saw 773 00:26:38,670 --> 00:26:41,115 this build this week, and they said that 774 00:26:41,115 --> 00:26:42,715 now we're doing list content, like, going, yeah, 775 00:26:42,715 --> 00:26:44,394 well, that's probably that's a couple days after 776 00:26:44,394 --> 00:26:45,535 we recorded this. So 777 00:26:45,994 --> 00:26:48,715 this is changing. Microsoft is is like they're 778 00:26:48,715 --> 00:26:51,674 feverishly working Right. On everything with this to 779 00:26:51,674 --> 00:26:54,234 make it more powerful. So this stuff could 780 00:26:54,394 --> 00:26:56,210 is I would expect it to change and 781 00:26:56,210 --> 00:26:57,730 get better, but it's just one of the 782 00:26:57,730 --> 00:26:59,490 things you wanna be, like, let's double check 783 00:26:59,490 --> 00:27:00,769 and make sure this is the case. That's 784 00:27:00,769 --> 00:27:02,690 okay. Right. That's turned into my default answer. 785 00:27:02,690 --> 00:27:04,369 And I'm doing demos now or something, and 786 00:27:04,369 --> 00:27:06,210 I can't find something. I'm like, must have 787 00:27:06,210 --> 00:27:08,494 changed because it does. It feels like this 788 00:27:08,494 --> 00:27:11,875 stuff is changing, like, daily or hourly or 789 00:27:12,095 --> 00:27:12,994 by the minute. 790 00:27:13,375 --> 00:27:14,734 I was teaching a class this week. I 791 00:27:14,734 --> 00:27:16,494 recorded a demo at 10:30 at night. I 792 00:27:16,494 --> 00:27:17,855 taught the class at nine the next morning, 793 00:27:17,855 --> 00:27:19,710 and it literally was different at nine the 794 00:27:19,710 --> 00:27:21,549 next morning in the exact same tenet. And 795 00:27:21,549 --> 00:27:23,230 so I was like, Yeah, stuff is flighting 796 00:27:23,230 --> 00:27:25,970 for build, so my bad. Welcome to Evergreen 797 00:27:26,029 --> 00:27:26,929 SaaS products. 798 00:27:27,230 --> 00:27:29,869 Exactly. Right. And this time of year is 799 00:27:29,869 --> 00:27:31,970 always fun too because the documentation 800 00:27:32,269 --> 00:27:34,704 often freezes before these events. So sometimes you 801 00:27:34,704 --> 00:27:36,484 start seeing new things come out before 802 00:27:37,105 --> 00:27:40,065 documentation's even there or or anything else along 803 00:27:40,065 --> 00:27:41,184 the way. So it's it's a little bit 804 00:27:41,184 --> 00:27:41,764 of a 805 00:27:42,065 --> 00:27:44,065 a challenge to to keep up with and 806 00:27:44,065 --> 00:27:46,144 balance. But I think, like, we've been talking 807 00:27:46,144 --> 00:27:47,744 about this for years and years and years 808 00:27:47,744 --> 00:27:50,319 with just Office March, m three sixty five 809 00:27:50,319 --> 00:27:52,559 in general. Like, the only constant is change 810 00:27:52,559 --> 00:27:55,359 there anyway. So you gotta kinda be ready 811 00:27:55,359 --> 00:27:57,279 ready to pivot and and deal with those 812 00:27:57,279 --> 00:27:57,779 things 813 00:27:58,400 --> 00:27:59,839 as they do come up. And I think, 814 00:27:59,839 --> 00:28:02,365 like, back to earlier about those customers who 815 00:28:02,365 --> 00:28:03,664 are mature enough maybe 816 00:28:04,525 --> 00:28:05,345 to go ahead 817 00:28:06,444 --> 00:28:09,085 and chunk and and generate embeddings for for 818 00:28:09,085 --> 00:28:09,904 their own stuff, 819 00:28:10,365 --> 00:28:12,125 they also know that they're kind of on 820 00:28:12,125 --> 00:28:14,525 the bleeding edge and need to keep up 821 00:28:14,525 --> 00:28:16,065 with some of this stuff. 822 00:28:16,420 --> 00:28:18,820 Yep. I I do worry sometimes about the 823 00:28:19,220 --> 00:28:21,140 not that's not like some, like, heavy hearted, 824 00:28:21,140 --> 00:28:22,759 like, the world's gonna end kind of thing, 825 00:28:22,820 --> 00:28:24,900 but just about the friction for customers there, 826 00:28:24,900 --> 00:28:26,900 just as the buttons are moving every day, 827 00:28:26,900 --> 00:28:29,000 the functionality is changing. Like, 828 00:28:29,460 --> 00:28:31,000 oh, I I I can do 829 00:28:31,384 --> 00:28:34,585 deep research today in in Copilot consumer, but 830 00:28:34,585 --> 00:28:36,744 I can't do it yet in m three 831 00:28:36,744 --> 00:28:40,444 sixty five Copilot. Like, those disparate experience experiences 832 00:28:40,504 --> 00:28:42,365 and things, they they do tend to 833 00:28:43,144 --> 00:28:44,904 add up. So you still gotta do the 834 00:28:44,904 --> 00:28:47,269 TCO and figure out, like, where value is 835 00:28:47,269 --> 00:28:48,490 inflicted in my org. 836 00:28:48,789 --> 00:28:50,950 I completely I completely agree with that. I 837 00:28:50,950 --> 00:28:52,150 mean, it's like it's one of the things 838 00:28:52,150 --> 00:28:54,230 that I I find is like a service 839 00:28:54,230 --> 00:28:55,750 offer for my customers. I I I do, 840 00:28:55,750 --> 00:28:57,910 like, this coaching service, and it's like, hey. 841 00:28:57,910 --> 00:28:59,190 Look. You can't stay on top of this 842 00:28:59,190 --> 00:29:00,630 stuff. I'm the one that's my job. I'm 843 00:29:00,630 --> 00:29:02,355 tango staying on top of it. So it's 844 00:29:02,355 --> 00:29:03,474 pretty much we jump on a call. It's, 845 00:29:03,474 --> 00:29:04,595 like, going, what do you know that's different 846 00:29:04,595 --> 00:29:05,714 based on what we're doing? And I'm, like, 847 00:29:05,714 --> 00:29:07,154 going, okay, this is no longer the case, 848 00:29:07,154 --> 00:29:08,434 or this is the case, or be careful 849 00:29:08,434 --> 00:29:09,795 with this, or I know that this is 850 00:29:09,795 --> 00:29:11,315 gonna change, so I would not invest time 851 00:29:11,315 --> 00:29:13,154 here because stuff is gonna change in the 852 00:29:13,154 --> 00:29:14,755 next few months. So it's it's just the 853 00:29:14,755 --> 00:29:16,529 only thing that's different about what you said 854 00:29:16,529 --> 00:29:18,130 to me, Scott, is that it's the change 855 00:29:18,130 --> 00:29:19,730 is just going so much faster. Like, we 856 00:29:19,730 --> 00:29:21,890 are on a much steeper downhill than we 857 00:29:21,890 --> 00:29:23,730 were on the mountain before. For sure. I 858 00:29:23,730 --> 00:29:25,730 I think pace to change is is very 859 00:29:25,730 --> 00:29:28,289 real there. So should you get into this 860 00:29:28,289 --> 00:29:30,244 world, all of a sudden, you end up 861 00:29:30,244 --> 00:29:31,464 in a place where 862 00:29:32,005 --> 00:29:33,845 you're kind of joining the space race of 863 00:29:33,845 --> 00:29:36,505 everybody else. And and you gotta be prepared 864 00:29:36,724 --> 00:29:39,125 for that. So the solution that you build 865 00:29:39,125 --> 00:29:40,825 today, should you take a dependency 866 00:29:41,125 --> 00:29:42,505 on any kind of 867 00:29:43,000 --> 00:29:45,480 SaaS surface or anything like that, it's or 868 00:29:45,480 --> 00:29:47,799 or a PaaS service even, you have to 869 00:29:47,799 --> 00:29:49,880 know that it's gonna change. And it's probably 870 00:29:49,880 --> 00:29:52,679 gonna change more rapidly than you're used to 871 00:29:52,679 --> 00:29:54,839 in the past. That said, if you're still 872 00:29:54,839 --> 00:29:56,200 doing some of this stuff maybe in, like, 873 00:29:56,200 --> 00:29:58,335 an on prem context, like, it's one thing 874 00:29:58,335 --> 00:30:00,335 to say, hey. I'm using Azure AI studio 875 00:30:00,575 --> 00:30:02,734 or Azure AI search, and it's moving at 876 00:30:02,734 --> 00:30:03,954 this pace. Or 877 00:30:04,335 --> 00:30:06,494 I'm using AI Foundry, and it's going this 878 00:30:06,494 --> 00:30:07,394 fast versus 879 00:30:07,934 --> 00:30:08,994 I I took a dependency 880 00:30:09,375 --> 00:30:11,535 on langchain and Pinecone here, and and that 881 00:30:11,535 --> 00:30:13,820 was locally for me to do. So you 882 00:30:13,820 --> 00:30:15,420 you can have a little bit of that 883 00:30:15,420 --> 00:30:17,599 stability along the way if you need it. 884 00:30:17,740 --> 00:30:19,660 Yep. It's not gonna be super stable. Like 885 00:30:19,660 --> 00:30:21,099 you said, the hill is still a hill. 886 00:30:21,099 --> 00:30:22,859 And and it's and it's a very it's 887 00:30:22,859 --> 00:30:25,019 it's a very steep one. Yeah. That we're 888 00:30:25,019 --> 00:30:26,619 all falling down or trying to climb up, 889 00:30:26,619 --> 00:30:27,839 I can't figure it out. 890 00:30:28,140 --> 00:30:29,474 It's I think it's a little bit of 891 00:30:29,474 --> 00:30:30,434 both. I can I can make a little 892 00:30:30,434 --> 00:30:32,115 bit of progress, but I fall four rings 893 00:30:32,115 --> 00:30:32,615 down? 894 00:30:33,875 --> 00:30:36,994 Yes. Yeah. %. So I know, Scott, you 895 00:30:36,994 --> 00:30:39,154 have meetings coming up. I have meetings coming 896 00:30:39,154 --> 00:30:40,375 up. AC has 897 00:30:40,769 --> 00:30:42,710 coaching and meetings coming 898 00:30:43,170 --> 00:30:46,049 up. But anything else, AC, any last thoughts, 899 00:30:46,049 --> 00:30:48,470 other things people should be aware of or 900 00:30:48,690 --> 00:30:50,849 even if people wanna learn more about Copilot 901 00:30:50,849 --> 00:30:52,609 dev? I know we didn't get into all 902 00:30:52,609 --> 00:30:54,450 the developing your own agents. Maybe we'll have 903 00:30:54,450 --> 00:30:56,444 to have you on for the fifth or 904 00:30:56,444 --> 00:30:58,444 sixth time or whatever and talk more about 905 00:30:58,444 --> 00:31:00,845 the custom dev story, but, yeah, people wanna 906 00:31:00,845 --> 00:31:02,845 get in touch with you, any of that. 907 00:31:02,845 --> 00:31:04,684 Yeah. I I appreciate that. I guess it 908 00:31:04,765 --> 00:31:06,044 I would love to come back on happy 909 00:31:06,044 --> 00:31:07,085 to talk about the dev story. In fact, 910 00:31:07,085 --> 00:31:08,204 it'd probably be better if we did it 911 00:31:08,204 --> 00:31:09,724 after build too so that I can Alright. 912 00:31:09,804 --> 00:31:11,380 I know some stuff that's coming that we'll 913 00:31:11,380 --> 00:31:12,980 talk about that Microsoft will talk about next 914 00:31:12,980 --> 00:31:15,059 week. But, yeah, I have a company called 915 00:31:15,059 --> 00:31:18,359 Voitanos. I focus on making you the best 916 00:31:18,419 --> 00:31:20,819 Microsoft three sixty five developer in the Microsoft 917 00:31:20,819 --> 00:31:22,039 three sixty five ecosystem 918 00:31:22,659 --> 00:31:24,980 and indispensable in your organization. So I do 919 00:31:24,980 --> 00:31:27,434 that through, like, SharePoint framework training, Teams app 920 00:31:27,434 --> 00:31:30,174 dev training, and Copilot training. I have a 921 00:31:30,394 --> 00:31:31,214 a six hour, 922 00:31:31,595 --> 00:31:33,994 two day virtual workshop that I do on 923 00:31:33,994 --> 00:31:36,634 building declarative agents for the developer audience, so 924 00:31:36,634 --> 00:31:39,515 not using Copilot Studio, but using Visual Studio 925 00:31:39,515 --> 00:31:41,980 Code. And I'm due to refresh that. I'm 926 00:31:41,980 --> 00:31:44,059 intentionally waiting until after build because I know 927 00:31:44,059 --> 00:31:45,820 things are gonna be changing. But I'm gonna 928 00:31:45,820 --> 00:31:47,339 I plan on I do that at conferences 929 00:31:47,339 --> 00:31:48,620 for a full day workshop. I did it 930 00:31:48,620 --> 00:31:50,539 at the Microsoft three sixty five community conference. 931 00:31:50,539 --> 00:31:51,980 I'm gonna do it again at the TechComm 932 00:31:51,980 --> 00:31:54,105 conference in Seattle in June. But I also 933 00:31:54,105 --> 00:31:56,184 have that as a, as a live workshop 934 00:31:56,184 --> 00:31:58,924 that I'll offer on my site at voitonos.io. 935 00:31:58,984 --> 00:32:01,224 And if you, if you if you're interested 936 00:32:01,224 --> 00:32:02,585 too, I mean, I've got, like, the recordings 937 00:32:02,585 --> 00:32:04,265 from when I last delivered that in March 938 00:32:04,265 --> 00:32:05,244 are also available 939 00:32:05,799 --> 00:32:07,559 for, to enroll in on my course where 940 00:32:07,559 --> 00:32:09,400 you'll actually get access to the the updated 941 00:32:09,400 --> 00:32:11,480 ones once those are done. So, yeah, I 942 00:32:11,480 --> 00:32:12,680 mean, the best place to find me is 943 00:32:12,680 --> 00:32:14,220 just voitanos.io 944 00:32:14,519 --> 00:32:16,119 or search for I I'll put give you 945 00:32:16,119 --> 00:32:17,240 guys my link so you can put them 946 00:32:17,240 --> 00:32:19,240 in the show notes, but Perfect. LinkedIn, Blue 947 00:32:19,240 --> 00:32:20,299 Sky, YouTube, 948 00:32:20,644 --> 00:32:21,845 I'm easy to find. If you just search 949 00:32:21,845 --> 00:32:24,244 for Andrew Connell in SharePoint, it's I'm pretty 950 00:32:24,244 --> 00:32:25,605 much the first one that's gonna come up. 951 00:32:25,605 --> 00:32:28,565 Perfect. Sounds good. Well, thanks, AC. Thanks for 952 00:32:28,565 --> 00:32:30,884 jumping on with us, filling in some of 953 00:32:30,884 --> 00:32:34,325 those additional details around CoPilot agents. And, appear 954 00:32:34,470 --> 00:32:36,630 I was gonna say the message takeaway from 955 00:32:36,630 --> 00:32:39,029 this is go pay attention at build, but 956 00:32:39,029 --> 00:32:41,029 this is gonna come out after build. So 957 00:32:41,029 --> 00:32:42,230 if you're listening to this Go back and 958 00:32:42,230 --> 00:32:44,390 watch build. Go back and watch build and 959 00:32:44,390 --> 00:32:48,069 look for maybe certain announcements around Copilot agents 960 00:32:48,069 --> 00:32:49,429 and some of the stuff you may have 961 00:32:49,429 --> 00:32:51,505 heard here. Watch for my recap. I will 962 00:32:51,505 --> 00:32:53,265 have a recap for eight when it comes 963 00:32:53,265 --> 00:32:55,505 to SharePoint, Teams app dev, and agents that 964 00:32:55,505 --> 00:32:57,345 is, like, the build stuff, like, this is 965 00:32:57,345 --> 00:32:58,785 what you wanna focus on. I did it 966 00:32:58,785 --> 00:33:00,785 for Ignite. I did it for Build. It's 967 00:33:00,785 --> 00:33:02,065 a great way for me to get get 968 00:33:02,065 --> 00:33:03,809 up to speed on everything, and I plan 969 00:33:03,809 --> 00:33:05,569 on doing it again as Build goes on 970 00:33:05,569 --> 00:33:07,409 next week as well. Okay. Perfect. And if 971 00:33:07,409 --> 00:33:08,309 you have that 972 00:33:08,609 --> 00:33:11,569 before this show goes live, send it over, 973 00:33:11,569 --> 00:33:12,929 and we'll include that in the show notes 974 00:33:12,929 --> 00:33:14,450 as well for anybody listening. If you wanna 975 00:33:14,450 --> 00:33:16,769 get AC's recap, and if it's live or 976 00:33:16,769 --> 00:33:18,724 going to be live, we'll add that as 977 00:33:18,724 --> 00:33:21,284 well. Awesome. Sounds good, man. Awesome. Well, thanks, 978 00:33:21,284 --> 00:33:23,444 guys. Appreciate it. Yeah. Enjoy your weekend, and 979 00:33:23,444 --> 00:33:24,804 we'll talk to you guys later. Thanks. Alright. 980 00:33:24,804 --> 00:33:26,804 Thanks, Ben, Scott. And congrats again on episode 981 00:33:26,804 --> 00:33:29,444 400. That's awesome. Alright. Thanks, AC. Appreciate it. 982 00:33:29,444 --> 00:33:29,910 Bye. 983 00:33:31,910 --> 00:33:34,150 If you enjoyed the podcast, go leave us 984 00:33:34,150 --> 00:33:36,390 a five star rating in iTunes. It helps 985 00:33:36,390 --> 00:33:38,070 to get the word out so more IT 986 00:33:38,070 --> 00:33:40,230 pros can learn about Office three sixty five 987 00:33:40,230 --> 00:33:40,890 and Azure. 988 00:33:41,430 --> 00:33:43,109 If you have any questions you want us 989 00:33:43,109 --> 00:33:45,269 to address on the show or feedback about 990 00:33:45,269 --> 00:33:47,644 the show, feel free to reach out via 991 00:33:47,644 --> 00:33:49,825 our website, Twitter, or Facebook. 992 00:33:50,125 --> 00:33:51,964 Thanks again for listening, and have a great 993 00:33:51,964 --> 00:33:52,464 day.

Digital Dispatch Podcast Podcast Artwork Image

Microsoft Cloud IT Pro Podcast

Ben Stegink, Scott Hoag

On the MS Cloud IT Pro Podcast, Scott and Ben discuss the Microsoft Cloud with a focus on IT Pros. They'll discuss the latest in Microsoft 365 and Office 365 News, Azure news and talk about their experiences with managing the Microsoft Cloud as well as interview industry experts on various cloud technology. They'll cover things such as SharePoint, Exchange, Microsoft Teams, PowerShell, Azure, Azure AD, Security, Networking, Storage, and the many other technologies and products that have made their way into the Microsoft 365 suite and Azure. To stay up-to-date on the latest in Microsoft Cloud news and gain some valuable knowledge as you deploy it within your own organization, make sure to tune in every week! Find out more at msclouditpropodcast.com.

Contact Show