Meeting Title: All Engineering (AI Uses + Workshop) Date: 2025-08-07 Meeting participants: Sam Roberts, Demilade Agboola, Annie Yu, Uttam Kumaran, Casie Aviles, Mustafa Raja, Awaish Kumar, Vashdev Heerani


WEBVTT

1 00:02:43.040 00:02:43.846 Uttam Kumaran: Hey, guys.

2 00:02:47.120 00:02:50.119 Demilade Agboola: No, Tom, how are you doing.

3 00:02:51.410 00:02:52.440 Uttam Kumaran: I’m good.

4 00:02:53.290 00:02:53.880 Sam Roberts: Thank you.

5 00:04:05.580 00:04:08.960 Casie Aviles: Hey, guys? Sorry I was in a standoff.

6 00:04:10.690 00:04:11.760 Uttam Kumaran: Hello!

7 00:04:12.860 00:04:13.490 Casie Aviles: Dude.

8 00:04:50.580 00:04:55.190 Casie Aviles: Okay, is there anyone else who will be joining?

9 00:04:55.580 00:04:56.380 Casie Aviles: Which is.

10 00:05:13.110 00:05:19.130 Awaish Kumar: I think we can start. And so basically, the agenda for this meeting is to

11 00:05:19.920 00:05:22.880 Awaish Kumar: and like promote the users. AI

12 00:05:23.594 00:05:30.160 Awaish Kumar: in our daily work. So to host the productivity, and, like Casey, is going to host this meeting.

13 00:05:30.670 00:05:36.170 Awaish Kumar: and he will lead the conversation for for this meeting.

14 00:05:38.210 00:05:40.169 Awaish Kumar: I think we can start Kcp.

15 00:05:41.640 00:05:52.840 Casie Aviles: Okay, so yeah, yeah. So basically as always said, I, wanted, to

16 00:05:54.430 00:05:58.259 Casie Aviles: talk about you know how we could use AI further for like

17 00:05:58.930 00:06:02.270 Casie Aviles: focusing on real engineering problems. And

18 00:06:02.700 00:06:06.073 Casie Aviles: you know, I think some of the common pain points that

19 00:06:07.040 00:06:14.809 Casie Aviles: engineering tend to. Engineers tend to experience, or even within the company like

20 00:06:15.562 00:06:20.499 Casie Aviles: dealing with, you know, with code that already exists and trying to understand that

21 00:06:21.853 00:06:26.476 Casie Aviles: and maybe unfamiliar documentation. So I that’s also something

22 00:06:28.107 00:06:31.979 Casie Aviles: I deal with when whenever I try, I read like

23 00:06:32.767 00:06:39.500 Casie Aviles: documentation for for certain, like certain Apis, that I need to integrate with and then

24 00:06:39.830 00:06:42.070 Casie Aviles: having the context switch for a

25 00:06:42.490 00:06:45.659 Casie Aviles: through through a lot of you know. A lot of different

26 00:06:45.800 00:06:51.040 Casie Aviles: like, we, we have different clients. We have different stuff. We have client work, internal work.

27 00:06:51.360 00:06:57.050 Casie Aviles: And I think context context switching is something that can be challenging in the in in this kind of

28 00:06:57.950 00:06:58.555 Casie Aviles: work.

29 00:06:59.900 00:07:06.350 Casie Aviles: yeah. So I guess those are some of the things that maybe we could use AI to help

30 00:07:07.900 00:07:11.210 Casie Aviles: you know, make make things a lot more manageable.

31 00:07:11.780 00:07:18.879 Casie Aviles: So I guess I could start with just showing how I use

32 00:07:19.950 00:07:26.660 Casie Aviles: how I use AI tools in order to yeah, to

33 00:07:27.090 00:07:29.620 Casie Aviles: help with productivity and all of that. So

34 00:07:30.950 00:07:38.640 Casie Aviles: yeah, I can go ahead. I’ll go ahead and share my screen. And so.

35 00:07:40.491 00:07:41.727 Casie Aviles: Yeah, for context.

36 00:07:42.460 00:07:48.999 Casie Aviles: I know, everyone is probably has probably heard of the Zoom Platform that you have already.

37 00:07:49.510 00:07:56.459 Casie Aviles: And for me. One of the challenges for that was, I I’m not

38 00:07:57.150 00:08:03.420 Casie Aviles: like I I don’t have like a front end background. I don’t have like a web development background. I’ve done a few.

39 00:08:03.810 00:08:08.600 Casie Aviles: but I’m not necessarily, you know, the most well versed in that.

40 00:08:09.090 00:08:09.789 Casie Aviles: So

41 00:08:10.700 00:08:18.000 Casie Aviles: also. And then we have this like code base that Miguel has already built. And so what I I guess the challenge for me was.

42 00:08:18.270 00:08:19.539 Casie Aviles: how do I build

43 00:08:20.610 00:08:26.530 Casie Aviles: like something new on top of this, and as you can see here on the left.

44 00:08:27.590 00:08:38.240 Casie Aviles: so this is what for those unfamiliar with courser. It’s like vs code, where there’s like a an AI agent embedded into it. And it’s very helpful because

45 00:08:38.440 00:08:46.729 Casie Aviles: it can help, you understand. And like a code base that’s or that already exists, or it can help you. Basically instantiate

46 00:08:47.670 00:08:48.900 Casie Aviles: the project.

47 00:08:49.860 00:08:55.739 Casie Aviles: So in my case, I had to work with this. And there, there was a lot of things here that I was unfamiliar with. And

48 00:08:55.930 00:09:00.100 Casie Aviles: I’m not very familiar with react or javascript.

49 00:09:01.930 00:09:03.239 Casie Aviles: So yeah, I guess

50 00:09:03.756 00:09:09.839 Casie Aviles: one of the 1st things that I would do is that AI has has helped me is to just ask like.

51 00:09:10.000 00:09:16.300 Casie Aviles: Can you give me a high level overview of the code base.

52 00:09:16.640 00:09:25.579 Casie Aviles: So these are things that I would ask the AI. So, yeah, yeah, let’s go here.

53 00:09:30.970 00:09:37.130 Casie Aviles: Yeah. For example. Let me see if I can think of something to

54 00:09:38.210 00:09:40.559 Casie Aviles: like a goal that we want to get done.

55 00:09:47.600 00:09:52.719 Casie Aviles: Okay, so let’s just say it’s a minor styling thing that we want to do. But okay?

56 00:09:53.229 00:10:00.690 Casie Aviles: So yeah, after the prompt. So you you see that cursor is reading all of the files that we have

57 00:10:02.380 00:10:08.100 Casie Aviles: and this basically makes it, you know, faster for us to understand what’s happening.

58 00:10:09.092 00:10:11.630 Casie Aviles: Because the AI is already reading it for us.

59 00:10:13.320 00:10:16.740 Casie Aviles: And it’s going to create like a like this.

60 00:10:18.640 00:10:21.490 Casie Aviles: yeah, the, this explanation over here.

61 00:10:22.690 00:10:23.980 Casie Aviles: Okay, so

62 00:10:24.210 00:10:30.618 Casie Aviles: yeah, from from here we can get a pretty good sense of what it what this code base is when what it’s

63 00:10:31.130 00:10:32.539 Casie Aviles: yeah, what it’s for.

64 00:10:33.660 00:10:37.330 Casie Aviles: Okay, so let’s say, like

65 00:10:37.650 00:10:45.850 Casie Aviles: how I how I would typically approach this is like, so I don’t know much here, and so, but I want something done like, for example, let’s say

66 00:10:47.520 00:10:48.540 Casie Aviles: I want to.

67 00:10:49.190 00:10:54.505 Casie Aviles: I want to edit this view like, maybe,

68 00:10:55.360 00:10:58.100 Casie Aviles: create like, make this bold or something.

69 00:10:58.250 00:11:06.010 Casie Aviles: So you could ask like, where do we display the meeting? Transcript?

70 00:11:09.240 00:11:13.529 Casie Aviles: So again, cursor is going to search that for us.

71 00:11:14.620 00:11:17.600 Casie Aviles: So instead of having to go through all of this.

72 00:11:19.280 00:11:23.889 Casie Aviles: you don’t need to. And the AI agent, we’ll make it faster.

73 00:11:26.350 00:11:32.850 Casie Aviles: Okay, so it’s it tells us that. The primary display is on the Zoom Meeting detail. Dot. Js.

74 00:11:34.220 00:11:37.659 Casie Aviles: okay, nice. So we can search for that.

75 00:11:38.030 00:11:44.429 Casie Aviles: So right now, we already know where to go, or at least with which

76 00:11:45.417 00:11:47.680 Casie Aviles: Js file is responsible for that.

77 00:11:54.040 00:11:54.970 Casie Aviles: Not here.

78 00:11:55.090 00:11:58.310 Casie Aviles: Yeah, it’s here. You could actually click here and then.

79 00:11:58.640 00:12:03.269 Casie Aviles: okay, so this script, this this, this code here is responsible for it.

80 00:12:05.480 00:12:09.300 Casie Aviles: and then we can say, Okay, boom.

81 00:12:19.590 00:12:22.679 Casie Aviles: So we want to change the styling of the Transcript view.

82 00:12:24.140 00:12:27.630 Casie Aviles: the timestamps to be in bold.

83 00:12:35.770 00:12:43.920 Casie Aviles: Okay. So now that we’ve given it something to do, it’s going to suggest some changes to the

84 00:12:44.970 00:12:49.467 Casie Aviles: file over here, so you can see, and then you’d be able to

85 00:12:52.740 00:12:55.540 Casie Aviles: review them before. Actually,

86 00:12:58.170 00:13:00.640 Casie Aviles: you know, before, before the changes are made.

87 00:13:01.140 00:13:04.499 Casie Aviles: So yeah, let’s say, we just keep this.

88 00:13:06.190 00:13:07.070 Casie Aviles: Okay?

89 00:13:13.850 00:13:19.760 Casie Aviles: And then, okay, nice. So we were able to change the styling so that

90 00:13:20.430 00:13:24.719 Casie Aviles: I know that’s a very small or very minor change. But

91 00:13:25.294 00:13:30.240 Casie Aviles: this could be, you know, we could do pretty much do the same process that I did here.

92 00:13:30.941 00:13:35.113 Casie Aviles: For other things also as well like, you know.

93 00:13:36.610 00:13:41.980 Casie Aviles: with all the like. For example, we created this, and we were just adding features on top of this

94 00:13:42.110 00:13:42.740 Casie Aviles: and

95 00:13:43.910 00:13:51.230 Casie Aviles: For example, if someone, another engineer would add, like, for example, most Staffa created like the linear tickets part.

96 00:13:52.250 00:13:54.779 Casie Aviles: they could do the same process where?

97 00:13:55.438 00:14:00.340 Casie Aviles: I want to add a feature to create linear tickets. So

98 00:14:00.520 00:14:04.840 Casie Aviles: that’s also the same thing that cursor will do. It will go through the

99 00:14:05.720 00:14:10.490 Casie Aviles: the code that we have and then suggest changes. And then we could just review it.

100 00:14:11.863 00:14:14.530 Casie Aviles: And that makes things a lot faster. Now.

101 00:14:16.850 00:14:21.160 Casie Aviles: Okay, yeah. So

102 00:14:21.320 00:14:27.699 Casie Aviles: I will pause there, for now does anyone have any questions or anything to share.

103 00:14:33.430 00:14:42.159 Uttam Kumaran: Yeah, I guess I’m curious on how any of the data folks are using cursor for development like, kind of the way I use it. I use it to write models.

104 00:14:42.810 00:14:47.569 Uttam Kumaran: Like most of the time I give it like the contents of the ticket I’m working on.

105 00:14:47.740 00:14:54.749 Uttam Kumaran: or I use whisper speech that’s actually just talking to it. And then I have cursor, kind of build.

106 00:14:55.155 00:15:00.049 Uttam Kumaran: You know the 1st version of of the dpt models that I then take the test.

107 00:15:08.690 00:15:16.950 Casie Aviles: Ye? Yeah, that’s also yeah. That’s also a pretty good way to do it. So we could go to a ticket, for example, and

108 00:15:19.230 00:15:29.099 Casie Aviles: just, you know, get the contents. I I know this is not very data related. Examples, but

109 00:15:29.290 00:15:34.180 Casie Aviles: hopefully, like just the process of what I do

110 00:15:34.580 00:15:42.339 Casie Aviles: is or yeah, getting the context needed from the tickets or the code base. And then just ask AI to

111 00:15:42.970 00:15:46.548 Casie Aviles: actually suggest, like, for example, here.

112 00:15:48.150 00:15:49.761 Casie Aviles: I would just ask a I

113 00:15:50.470 00:15:53.110 Casie Aviles: maybe we could do another example where?

114 00:16:02.390 00:16:10.840 Casie Aviles: yeah. Basically just ask, give the ticket as context and then have a courser. Also take a crack at it and create like the

115 00:16:11.150 00:16:12.210 Casie Aviles: the script.

116 00:16:12.650 00:16:17.060 Casie Aviles: So let me just look for

117 00:16:23.150 00:16:25.990 Casie Aviles: or I’ll just create like a

118 00:16:43.910 00:16:46.725 Casie Aviles: yeah. So let’s say, I’m going to create like a new project.

119 00:16:54.760 00:16:58.629 Casie Aviles: and when we want to. When we instantiate like a project.

120 00:16:59.370 00:17:01.899 Casie Aviles: we could also ask courser to

121 00:17:17.339 00:17:19.999 Casie Aviles: to accomplish this task.

122 00:17:20.750 00:17:25.330 Casie Aviles: So we could do something like this. And we just provide like the ticket as context.

123 00:17:33.120 00:17:37.160 Casie Aviles: So yeah, this way, we could. Also, you know, it’s also

124 00:17:37.810 00:17:39.892 Casie Aviles: much faster to get something out.

125 00:17:42.770 00:17:46.755 Casie Aviles: what we could do is to just review what is being generated. And

126 00:17:48.240 00:17:56.400 Casie Aviles: yeah. So now it’s providing some implementation ideas

127 00:17:58.170 00:18:00.280 Casie Aviles: and also giving us some estimates.

128 00:18:01.530 00:18:03.400 Casie Aviles: The Apis that we need

129 00:18:13.780 00:18:21.380 Casie Aviles: yeah to handle. So let’s say, we want to create like a simple python script to handle the vectorization or the embedding. So

130 00:18:21.490 00:18:23.780 Casie Aviles: for context, the task is to

131 00:18:24.762 00:18:28.200 Casie Aviles: get the meeting transcripts that we have from Zoom.

132 00:18:28.540 00:18:37.570 Casie Aviles: And then we jet, we, we want to create like a modular script, a reusable script that will

133 00:18:38.810 00:18:43.949 Casie Aviles: basically vectorize that for our rug for retrieval. So

134 00:18:44.260 00:18:46.686 Casie Aviles: yeah, it’s going to create something like this.

135 00:18:47.570 00:18:51.022 Casie Aviles: this is and then we, we have the option to just

136 00:18:51.610 00:18:57.499 Casie Aviles: So yeah, this is, this is like the transcript. And then we could review this 1st and then test it.

137 00:18:58.030 00:19:02.190 Casie Aviles: But yeah, that’s pretty much how the process goes for this.

138 00:19:03.230 00:19:11.600 Casie Aviles: And yeah, these are just some of the ways that we could use. Corsaar. 2, yeah, to

139 00:19:13.280 00:19:18.179 Casie Aviles: handle our tasks. Okay, yeah.

140 00:19:28.893 00:19:35.789 Casie Aviles: Yeah. So any questions. So far, I I think I’ve yeah. Those are the 2

141 00:19:36.170 00:19:38.470 Casie Aviles: main ways. I use cursor.

142 00:19:43.890 00:19:44.600 Awaish Kumar: Great.

143 00:19:45.738 00:19:48.029 Awaish Kumar: Does anyone have any questions.

144 00:19:51.800 00:19:56.040 Awaish Kumar: especially for the people who haven’t used the cursor before?

145 00:19:56.940 00:19:57.610 Casie Aviles: Yes.

146 00:20:08.610 00:20:10.849 Demilade Agboola: So I have. I have used cluster before.

147 00:20:10.970 00:20:18.469 Demilade Agboola: I don’t know if you might want to mention that you can actually add multiple

148 00:20:19.370 00:20:21.409 Demilade Agboola: files into the chat.

149 00:20:22.270 00:20:29.560 Demilade Agboola: And so that allows you to also like, ask very specific questions. To what should I need.

150 00:20:31.970 00:20:33.230 Casie Aviles: Like, if you write.

151 00:20:33.230 00:20:37.179 Demilade Agboola: So if you yeah. So if you right click on the file.

152 00:20:37.610 00:20:39.210 Demilade Agboola: you can add it to a chat.

153 00:20:39.600 00:20:51.230 Demilade Agboola: And then, yeah, so you can actually now add multiple and say, Hey, from this 1st file I need you to extract this information or extract this logic and then use it in this second file or something.

154 00:20:51.540 00:20:54.180 Demilade Agboola: So I know I use that sometimes.

155 00:20:56.030 00:21:02.650 Casie Aviles: Yes, that’s also a a good way to add like context to the chat.

156 00:21:03.456 00:21:07.019 Casie Aviles: Yes, thanks, thanks, Damien Hadi, for sharing that as well.

157 00:21:08.323 00:21:10.989 Casie Aviles: Yeah. And then there are like, also other

158 00:21:12.474 00:21:17.940 Casie Aviles: shortcuts that we could do. I if I yeah, like, for example, going to.

159 00:21:18.150 00:21:19.589 Casie Aviles: if you want just

160 00:21:20.470 00:21:25.509 Casie Aviles: a chunk of the code. Maybe we could also do like control. L, to add it to the chat

161 00:21:25.740 00:21:31.200 Casie Aviles: which adds it to the context, or we could also do.

162 00:21:32.740 00:21:35.420 Casie Aviles: And they can edit here.

163 00:21:35.740 00:21:45.120 Casie Aviles: So you’re you’re just gonna ask the AI to edit this like yeah, that. Those are examples.

164 00:21:46.440 00:21:52.659 Casie Aviles: Maybe that, for example. Let’s change the

165 00:21:58.520 00:22:03.530 Casie Aviles: Count tokens peace.

166 00:22:07.889 00:22:08.729 Casie Aviles: Let me see.

167 00:22:16.000 00:22:17.420 Casie Aviles: Yeah, something like this.

168 00:22:17.710 00:22:21.939 Casie Aviles: You could just ask it to make some edits.

169 00:22:22.950 00:22:27.290 Casie Aviles: That’s pretty. Yeah, just minimal edits. For example, we could do that.

170 00:22:29.624 00:22:33.477 Casie Aviles: But yeah, I think that’s pretty much

171 00:22:34.360 00:22:37.240 Casie Aviles: how I would use courser. And

172 00:22:37.870 00:22:41.840 Casie Aviles: yeah, for for basically making it easier to do all these tasks.

173 00:22:52.900 00:22:53.720 Casie Aviles: Okay.

174 00:22:58.260 00:23:00.860 Casie Aviles: alright. Okay. Just give me a bit.

175 00:23:16.860 00:23:18.899 Casie Aviles: okay, cool. So

176 00:23:19.130 00:23:25.019 Casie Aviles: other things that I would use AI to help me with like, for example, I did mention them.

177 00:23:27.580 00:23:31.670 Casie Aviles: Sometimes I would read like certain documentation, the

178 00:23:33.340 00:23:41.939 Casie Aviles: I would use it to, you know. Basically, instead of going through all of the all of the

179 00:23:43.450 00:23:45.070 Casie Aviles: contents of the Doc.

180 00:23:45.520 00:23:49.770 Casie Aviles: it would be easier for me for me to like, just, you know, query over it.

181 00:23:50.639 00:23:52.839 Casie Aviles: and then make notes.

182 00:23:54.790 00:23:56.110 Casie Aviles: Yeah, for example,

183 00:24:04.920 00:24:11.620 Casie Aviles: or, yeah, let’s just for example, use another one that I was doing a spike on

184 00:24:17.522 00:24:20.180 Casie Aviles: I would just go to the docs and

185 00:24:20.510 00:24:29.460 Casie Aviles: copy paste that basically onto chat. Gpt, yeah. Here, the Api reference.

186 00:24:39.370 00:24:47.961 Casie Aviles: Okay, so so I would just there, there, yeah, this is one way to do it.

187 00:24:49.790 00:24:51.850 Casie Aviles: I would just get like the contents.

188 00:24:54.100 00:24:54.780 Casie Aviles: And

189 00:24:58.332 00:25:06.020 Casie Aviles: given this. Doc, how do I perform a search? Query.

190 00:25:16.880 00:25:20.639 Casie Aviles: okay, so this this way. I don’t have to like

191 00:25:21.679 00:25:25.299 Casie Aviles: Read individually, and I would just ask like the AI to.

192 00:25:26.170 00:25:29.759 Casie Aviles: you know, to use AI kind of like as a

193 00:25:30.190 00:25:34.792 Casie Aviles: like an expert on the docs that we’re reading already. And then you could just

194 00:25:36.105 00:25:43.810 Casie Aviles: ask questions. So me personally, it’s also I’ve also find it a lot easier to just ask questions immediately on something.

195 00:25:44.640 00:25:47.687 Casie Aviles: And this is one of the ways that I

196 00:25:49.266 00:26:02.090 Casie Aviles: Streamline, like the the way I work when it comes to having to deal with a lot of different, or a lot of new Apis or tools with you know all their documentation and whatnot

197 00:26:09.160 00:26:10.530 Casie Aviles: yeah, okay.

198 00:26:13.580 00:26:20.000 Casie Aviles: so I think that’s that’s pretty much how I

199 00:26:20.480 00:26:26.540 Casie Aviles: get, you know I get help from AI.

200 00:26:27.206 00:26:30.689 Casie Aviles: I guess other other things, of course, like debugging

201 00:26:32.226 00:26:34.760 Casie Aviles: like I have like a.

202 00:26:36.040 00:26:45.210 Casie Aviles: For example, we get like errors like this. I would just also paste them into chat. Gpt.

203 00:26:48.601 00:26:52.030 Casie Aviles: You know, especially if you’re not too familiar with

204 00:26:53.250 00:26:58.430 Casie Aviles: with other. The reasons of why this why errors have, why this kind of error happens so

205 00:27:03.510 00:27:05.430 Casie Aviles: something like this. So

206 00:27:05.600 00:27:11.079 Casie Aviles: when when I saw this error like I’m not entirely sure what are the steps that I needed to take?

207 00:27:11.550 00:27:16.687 Casie Aviles: And you know, Chat gpt when I just asked it to give me a lot of

208 00:27:17.920 00:27:25.789 Casie Aviles: contacts already on like oh, next steps and suggestions. And why it happens so. These are the things that I could look out for

209 00:27:28.660 00:27:29.320 Casie Aviles: fine.

210 00:27:29.420 00:27:30.680 Casie Aviles: Yes. So

211 00:27:31.320 00:27:40.300 Casie Aviles: yeah, these are the things that it’s suggested, and these are what I could do, what steps I could do next in order to resolve the problem.

212 00:27:41.260 00:27:48.520 Casie Aviles: So yeah, I think that’s pretty much those are some of the examples. The top of my head

213 00:27:49.170 00:27:51.920 Casie Aviles: where I would use AI to help me with engineering.

214 00:28:02.300 00:28:03.830 Casie Aviles: Yeah, okay.

215 00:28:06.910 00:28:13.630 Casie Aviles: I guess another thing that I could share is with our platform. I know that we have

216 00:28:14.290 00:28:16.680 Casie Aviles: our AI agents here.

217 00:28:16.960 00:28:19.339 Casie Aviles: So I I believe they they work

218 00:28:19.550 00:28:22.599 Casie Aviles: very similarly to like custom gpts.

219 00:28:23.300 00:28:28.260 Casie Aviles: So if we’re worried about like not using the right prompt.

220 00:28:29.250 00:28:31.590 Casie Aviles: we have a lot here that we could use.

221 00:28:34.130 00:28:39.139 Casie Aviles: you know. We could build our prompts using this prompt improver or optimizer.

222 00:28:42.980 00:28:47.640 Casie Aviles: You can see, there are a lot of different other other custom prompts here. And

223 00:28:48.150 00:28:52.200 Casie Aviles: we could also create our own based on our specific needs. So

224 00:28:53.734 00:29:00.180 Casie Aviles: yeah, over here to other options, we also have the manage agents feature, and

225 00:29:01.090 00:29:04.226 Casie Aviles: we could create like a an agent that helps us.

226 00:29:05.650 00:29:10.650 Casie Aviles: yeah, that helps us with specific tasks. And we could create like a prompt for that.

227 00:29:11.290 00:29:15.000 Casie Aviles: And we can even use the the prompt improver

228 00:29:15.540 00:29:19.600 Casie Aviles: so like, if we have, like a very rough idea of what the

229 00:29:21.164 00:29:27.420 Casie Aviles: prompt or like the instructions need to accomplish. Then we could definitely use this.

230 00:29:28.740 00:29:35.340 Casie Aviles: like, it’s like a meta prompt for responsible for improving what? We right? So

231 00:29:35.730 00:29:42.259 Casie Aviles: yeah, those are some of the options that we could definitely leverage for AI,

232 00:29:48.780 00:29:55.659 Casie Aviles: yeah, I think, yeah, I I think that’s it. For for, like my demo.

233 00:29:59.400 00:30:03.009 Awaish Kumar: And like, is there any difference like

234 00:30:03.710 00:30:07.230 Awaish Kumar: like if I if I’m using culture, and there are different like agents.

235 00:30:08.010 00:30:12.609 Awaish Kumar: option to select different agents like how they is there any any

236 00:30:14.720 00:30:18.919 Awaish Kumar: like like? If you open the concussor, there’s like option to select

237 00:30:19.140 00:30:22.309 Awaish Kumar: different 4, 3, or 4 different.

238 00:30:22.310 00:30:23.130 Casie Aviles: Yes.

239 00:30:23.560 00:30:25.280 Awaish Kumar: And things like that.

240 00:30:25.380 00:30:30.010 Awaish Kumar: So does that make any difference in like if I’m asking.

241 00:30:30.320 00:30:33.680 Awaish Kumar: I have to write. Ask you, Kerry, so how

242 00:30:34.931 00:30:42.449 Awaish Kumar: like, do we have anything like which which one be the better to use? For which use cases? Do you know anything.

243 00:30:44.708 00:30:50.119 Casie Aviles: As far as I know, I’ve heard like good things about, you know. Claude, for

244 00:30:50.710 00:30:54.640 Casie Aviles: for coding. Yeah. So I I would say like.

245 00:30:55.050 00:31:01.560 Casie Aviles: that’s a good option for if you want to do like a lot of coding, then

246 00:31:01.670 00:31:03.609 Casie Aviles: blood’s a good option as well.

247 00:31:04.630 00:31:11.099 Casie Aviles: For you know smaller tasks, but for speed, then I guess the smaller models would also be.

248 00:31:11.430 00:31:14.940 Casie Aviles: And then, for reasoning, we have O 3.

249 00:31:15.120 00:31:19.050 Casie Aviles: So Audrey is going to be slower.

250 00:31:19.620 00:31:22.201 Casie Aviles: It’s going to like like it does.

251 00:31:22.910 00:31:25.590 Casie Aviles: I think a chain of thought prompting where?

252 00:31:26.831 00:31:29.290 Casie Aviles: It tries to think about

253 00:31:30.438 00:31:37.949 Casie Aviles: or like mimic. How we would think in terms of steps. So that’s why it’s going to take a lot longer. But it can provide like

254 00:31:38.270 00:31:44.340 Casie Aviles: more insightful results in that way. Yeah.

255 00:31:47.830 00:31:48.560 Awaish Kumar: Nothing.

256 00:31:48.670 00:31:51.790 Awaish Kumar: Does anyone have any questions?

257 00:31:53.310 00:31:55.040 Awaish Kumar: The last day of any.

258 00:32:05.360 00:32:08.730 Vashdev Heerani: Nope, I don’t have any question right now.

259 00:32:16.810 00:32:17.230 Casie Aviles: Okay.

260 00:32:17.230 00:32:20.020 Annie Yu: A quick question. So I don’t work with

261 00:32:22.840 00:32:48.889 Annie Yu: Dbt, a lot compared to real. And when working with real a lot of times, we have to get the source from like snowflake or bigquery, and then build the models there. So in order to get cursor to read all those fields, do we have to build like a raw model based on those sources so it can

262 00:32:49.000 00:32:51.169 Annie Yu: know what fields there are.

263 00:32:53.655 00:32:59.360 Casie Aviles: Like. What? What do you mean by building a role model like creating our own model from?

264 00:32:59.680 00:33:00.370 Casie Aviles: I’m crazy.

265 00:33:00.650 00:33:04.799 Annie Yu: 1st file, we would just do like select star from

266 00:33:05.370 00:33:09.299 Annie Yu: a data table from Snowflake. So

267 00:33:10.690 00:33:14.420 Annie Yu: so that with that source file.

268 00:33:15.320 00:33:18.239 Annie Yu: does it know what fields there are.

269 00:33:25.330 00:33:32.460 Casie Aviles: Think we can go. Let me see if I have like a real, if I can, access really.

270 00:33:32.460 00:33:37.180 Demilade Agboola: I’m sorry just to be clear. It’s a question that if you use the let’s start.

271 00:33:37.950 00:33:41.420 Demilade Agboola: We didn’t know all the the list of every single column in there.

272 00:33:41.420 00:33:45.850 Annie Yu: Yeah, pretty much. Yeah, without. Yeah.

273 00:33:46.830 00:33:56.810 Demilade Agboola: In my experience, unless there’s documentation that you can read or you give it. Maybe a Csv like an output of it. It wouldn’t know

274 00:33:57.870 00:34:03.790 Demilade Agboola: in my experience without your question, I’m not usually real, and maybe has access to some of the actual data.

275 00:34:04.030 00:34:08.129 Demilade Agboola: But if there’s no access to the data. It doesn’t know what the star represents.

276 00:34:09.179 00:34:09.789 Awaish Kumar: So.

277 00:34:10.102 00:34:11.667 Casie Aviles: Yeah, yeah, that makes sense.

278 00:34:11.989 00:34:14.729 Casie Aviles: But is it possible to provide a.

279 00:34:14.979 00:34:17.209 Awaish Kumar: The CC file in context.

280 00:34:19.109 00:34:19.669 Casie Aviles: I think.

281 00:34:19.670 00:34:23.210 Demilade Agboola: Yes, yes, you can. I believe you cannot. I’m not sure

282 00:34:23.880 00:34:28.889 Demilade Agboola: if I use G. If I use Gpt or crystal. But let’s try. But I think you can upload

283 00:34:29.730 00:34:32.770 Demilade Agboola: documents into this into cursor.

284 00:34:33.790 00:34:36.340 Demilade Agboola: Or it might just be images. I’m not sure. Okay.

285 00:34:38.330 00:34:41.636 Casie Aviles: Yeah, we could use images. For sure.

286 00:34:46.480 00:34:51.409 Casie Aviles: And then we could reference like documents here as well files.

287 00:34:52.880 00:34:57.110 Casie Aviles: So if we had like a Csv file. I think we could be. We would be able to

288 00:34:57.380 00:34:58.980 Casie Aviles: read from that as well.

289 00:35:05.200 00:35:06.179 Awaish Kumar: Yeah. So

290 00:35:06.860 00:35:13.470 Awaish Kumar: the one of the problem which I, I would face in a data while building the data model is

291 00:35:13.690 00:35:20.150 Awaish Kumar: like, if we ingest a new source, for example, like Amazon, which has hundreds of tables.

292 00:35:20.330 00:35:27.019 Awaish Kumar: and it’s like impossible to get the simple schemas and things like that doing it

293 00:35:27.300 00:35:31.069 Awaish Kumar: to chip, chat, gpt, or concern, and then ask for modeling work.

294 00:35:34.290 00:35:42.200 Awaish Kumar: So how would you want? How? How can we like better handle one of the way we normally do is like, identify the table ourselves.

295 00:35:42.550 00:35:47.789 Awaish Kumar: What is needed for for a specific use case and then

296 00:35:48.150 00:35:57.020 Awaish Kumar: give the sample data, then ask activity. But are there any like tools which can integrate with, for example, database and

297 00:35:57.650 00:35:59.299 Awaish Kumar: help write some queries.

298 00:36:07.817 00:36:08.380 Casie Aviles: so

299 00:36:09.150 00:36:16.750 Casie Aviles: going going back to like just a Csv, I believe that that was that the question like.

300 00:36:17.600 00:36:20.740 Casie Aviles: how do we ingest, like the

301 00:36:21.430 00:36:26.570 Casie Aviles: the the files, in order to get a better sense of like what what we need to do is that

302 00:36:32.660 00:36:34.820 Casie Aviles: is that like the the question.

303 00:36:37.520 00:36:42.100 Awaish Kumar: Yeah, like that like Demon Demo said. Like, if if we have a single

304 00:36:42.620 00:36:46.930 Awaish Kumar: table, I can download it from bigquery or sample Csv and load it

305 00:36:47.450 00:36:49.839 Awaish Kumar: in the context on the cursor, and ask

306 00:36:50.413 00:36:53.820 Awaish Kumar: ask us to write some queries for me. My question was like.

307 00:36:54.980 00:37:01.610 Awaish Kumar: If I have a list of tables like not just one, that maybe 50 tables, which

308 00:37:02.220 00:37:09.000 Awaish Kumar: which is like basically hard like I. I don’t think like any any like the

309 00:37:09.260 00:37:12.649 Awaish Kumar: yeah. I told, I know, like Christmas, Jd. Hack.

310 00:37:12.860 00:37:15.829 Awaish Kumar: I can accept that many files.

311 00:37:16.790 00:37:18.739 Awaish Kumar: So is there any file like.

312 00:37:19.020 00:37:24.249 Awaish Kumar: for example, is there any way in the cursor or any other tools I can connect?

313 00:37:24.560 00:37:28.890 Awaish Kumar: Get to a database which has 50 different tables.

314 00:37:29.090 00:37:31.349 Awaish Kumar: and it can read the schema

315 00:37:31.730 00:37:37.670 Awaish Kumar: automatically, like without me writing some script or tool to that, sure.

316 00:37:37.820 00:37:43.689 Awaish Kumar: And then this is, get a prompt and then create something.

317 00:37:45.564 00:37:48.605 Casie Aviles: For that I have not explored, like, you know, directly

318 00:37:50.010 00:37:55.849 Casie Aviles: Connecting it with a like a database like at least on courser, or even chat. Gpt.

319 00:37:56.110 00:37:59.279 Casie Aviles: What I think

320 00:38:00.070 00:38:05.390 Casie Aviles: the what we could do is to just, you know, add as much as we can into the context like.

321 00:38:05.850 00:38:09.710 Casie Aviles: if it’s a schema. Then I guess we could paste that as well into.

322 00:38:11.110 00:38:13.680 Casie Aviles: you know. Add that to the chat, to the context.

323 00:38:14.450 00:38:20.359 Casie Aviles: and and for 1st like, for example, for like structured data, sometimes on what courser does is

324 00:38:20.970 00:38:23.770 Casie Aviles: it doesn’t read like the entire file.

325 00:38:24.110 00:38:27.480 Casie Aviles: What it does is it just gets like up to

326 00:38:28.490 00:38:31.539 Casie Aviles: like. Let’s say, 20 records from that file.

327 00:38:31.830 00:38:36.269 Casie Aviles: It’s going to use that in order to just get a sense of what’s happening.

328 00:38:40.990 00:38:47.710 Sam Roberts: I also add that you can add some Mcp tools to cursor, so, depending on

329 00:38:48.090 00:38:53.319 Sam Roberts: what the providers are, and if they have an Mcp. Server set up.

330 00:38:53.530 00:38:56.359 Sam Roberts: Do you have the settings open still? Yeah, that’s a bad world.

331 00:38:57.700 00:39:01.350 Sam Roberts: You can add things here that will give it. Basically, tool calls

332 00:39:01.500 00:39:08.500 Sam Roberts: that the you know. Lm, that’s being prompted by cursor could potentially use to query things on its own.

333 00:39:09.430 00:39:15.139 Sam Roberts: or do other other things like, I have one set up that knows to search documentation, and it checks out.

334 00:39:15.290 00:39:22.400 Sam Roberts: you know, does a search on Github for certain libraries, and pulls down the readines and things like that all on its own. So, depending on

335 00:39:22.640 00:39:26.889 Sam Roberts: where you’re trying to pull stuff from, the Mcp servers might be the way to go for certain things.

336 00:39:30.913 00:39:31.800 Awaish Kumar: It’s okay.

337 00:39:31.950 00:39:32.620 Casie Aviles: Yes.

338 00:39:44.650 00:39:45.035 Casie Aviles: yeah.

339 00:39:45.820 00:39:58.109 Casie Aviles: cool. Yeah, that. That’s I agree there. It’s just a matter of getting like the context, really. And Mcp is just. It’s another way that makes things more flexible.

340 00:39:58.890 00:40:04.790 Awaish Kumar: Just one more question like, like says, we can add a custom. Mcp. Server.

341 00:40:05.550 00:40:09.170 Awaish Kumar: So is it possible, for example, I had Mcb. Server

342 00:40:09.330 00:40:12.000 Awaish Kumar: to connect it with with a

343 00:40:12.260 00:40:15.469 Awaish Kumar: database and then share it with the team right?

344 00:40:15.850 00:40:24.209 Awaish Kumar: Like everybody that is able to utilize what I is it possible in cursor.

345 00:40:27.280 00:40:34.010 Casie Aviles: Yeah, I believe it should be possible, like, you’re just gonna have to. Give them the configuration something like this.

346 00:40:34.910 00:40:36.940 Casie Aviles: And then they just have to turn it on.

347 00:40:37.350 00:40:37.990 Awaish Kumar: Okay.

348 00:40:47.750 00:40:53.099 Casie Aviles: Hmm, cool. Yeah, I think that’s that’s pretty much it from

349 00:40:53.980 00:40:58.369 Casie Aviles: me, unless there are any other things you want to share or.

350 00:41:09.170 00:41:12.199 Awaish Kumar: Okay, I think we can close this.

351 00:41:14.430 00:41:14.899 Casie Aviles: Cool

352 00:41:15.370 00:41:18.279 Awaish Kumar: Thank you, Casey, like I thought it was really nice

353 00:41:19.440 00:41:23.970 Awaish Kumar: and like, Thank you for taking the 1st pause, like it was a 1st

354 00:41:24.150 00:41:27.870 Awaish Kumar: such kind of session for thanks for volunteering.

355 00:41:29.090 00:41:32.472 Awaish Kumar: I think. For the next we are going to have

356 00:41:33.480 00:41:37.530 Awaish Kumar: have it much better, and like maybe, like, add new.

357 00:41:37.670 00:41:41.040 Awaish Kumar: maybe kind of workshop type, where everybody tries something

358 00:41:42.050 00:41:49.214 Awaish Kumar: like so that we have much more engagement instead of just use

359 00:41:50.810 00:41:58.040 Awaish Kumar: showing the demo like we. Everybody will be trying some something, some hands off breakfast.

360 00:41:58.350 00:42:00.820 Awaish Kumar: So yeah, thank you.

361 00:42:02.870 00:42:09.090 Casie Aviles: Yeah, definitely. It would be great if people could get their hands on the tool as well

362 00:42:09.790 00:42:14.480 Casie Aviles: and work on like, yeah, in the future. In the next

363 00:42:15.718 00:42:17.750 Casie Aviles: sessions. We could do that definitely.

364 00:42:21.200 00:42:23.010 Awaish Kumar: Okay, thank you. Everybody.

365 00:42:23.620 00:42:24.490 Awaish Kumar: Thanks for joining.

366 00:42:24.490 00:42:25.319 Sam Roberts: Thanks, Casey.

367 00:42:25.320 00:42:25.990 Annie Yu: You.

368 00:42:27.210 00:42:27.950 Casie Aviles: Thank you.