Meeting Title: AI Team Standup Date: 2025-05-08 Meeting participants: Uttam Kumaran, Miguel De Veyra, Casie Aviles, Awaish Kumar


WEBVTT

1 00:00:36.890 00:00:38.860 Miguel de Veyra: Hey? Aish morning.

2 00:00:40.860 00:00:41.670 Awaish Kumar: Bye.

3 00:01:25.410 00:01:26.250 Uttam Kumaran: Hey, guys.

4 00:01:28.170 00:01:29.130 Awaish Kumar: Hello!

5 00:01:29.720 00:01:30.570 Miguel de Veyra: Hey? Autumn.

6 00:01:31.030 00:01:31.820 Uttam Kumaran: Hey?

7 00:01:35.210 00:01:39.280 Uttam Kumaran: Like there’s a there’s a squirrel like right in front of my office. Here.

8 00:01:46.790 00:01:49.430 Uttam Kumaran: my dog, my dog here is sleeping.

9 00:01:49.740 00:01:52.820 Miguel de Veyra: Oh, if he wakes up, it’s gonna be a little working.

10 00:01:52.820 00:01:54.619 Uttam Kumaran: A pillow on the ground

11 00:01:56.840 00:02:02.192 Uttam Kumaran: like, what is such a mess in this house? And then, yeah, the squirrels like just ran away.

12 00:02:03.940 00:02:07.169 Uttam Kumaran: All these like animals in my house.

13 00:02:11.510 00:02:12.470 Uttam Kumaran: Oh, man.

14 00:02:15.570 00:02:16.524 Uttam Kumaran: okay.

15 00:02:18.040 00:02:27.169 Uttam Kumaran: yeah. I sent a couple of notes. So yeah, I mean, I think we’ll basically, I think, sort of. We did some talking yesterday. And

16 00:02:27.734 00:02:30.729 Uttam Kumaran: yeah, I think the net of it is

17 00:02:31.110 00:02:39.650 Uttam Kumaran: one. I think this team we can move a lot faster by working on the right thing. But also I wanna certainly help to unblock

18 00:02:40.200 00:02:45.899 Uttam Kumaran: I think probably for me, I have the most context about where automation can be

19 00:02:46.060 00:02:48.290 Uttam Kumaran: most effective for our business.

20 00:02:49.117 00:02:51.240 Uttam Kumaran: But very similarly.

21 00:02:51.470 00:03:06.660 Uttam Kumaran: what I lack is, you know, the actual ability to go execute on that. And so you know, we started this team mainly like it was just me and Miguel really just sort of like cranking through stuff. And so I think some return to roots is good

22 00:03:07.201 00:03:09.659 Uttam Kumaran: I think. Also, you know, for amber

23 00:03:10.302 00:03:12.899 Uttam Kumaran: she’s she’s a rock star. So

24 00:03:13.080 00:03:24.499 Uttam Kumaran: for her I wanted. I told her that. Hey, focus on the client work. So she’s just gonna continue to run standups for matter more and for ABC. And then a couple of other clients, of course.

25 00:03:24.915 00:03:30.149 Uttam Kumaran: But I wanted to make sure that she could go focus on that and nail clients, because without clients

26 00:03:30.310 00:03:37.379 Uttam Kumaran: we don’t really have a business. And then for us, for this crew. We just basically look to scale

27 00:03:37.910 00:03:50.129 Uttam Kumaran: as much as we can and and reduce you know our need to spend money, and either an external tools to do these things or hire more people. And so this team’s directive is to speed every single

28 00:03:50.340 00:03:53.380 Uttam Kumaran: team in the company up, and that hasn’t really changed.

29 00:03:53.980 00:04:20.090 Uttam Kumaran: I think we’re still leveraging the same technologies. I think we’re gonna talk talk about a couple of new concepts and ways of doing things. But I think this crew is the right crew. I know we’re we’re interviewing one or 2 more people. But you know, kind of the the feeling I have is until the 4 of us sort of know that our cadence of working is good. And we’re actually seeing impact. Throwing another person into this is just gonna

30 00:04:20.440 00:04:23.070 Uttam Kumaran: kind of like make things worse.

31 00:04:23.300 00:04:34.409 Uttam Kumaran: And so our job is just to make sure that one couple of things, just one stuff we’re working on is ticketed at minimum. It’s like, What is this? What’s the goal? When is it gonna be done?

32 00:04:34.964 00:04:45.480 Uttam Kumaran: I want to do a lot more show and tell. So like these stand ups are not gonna be so much as like going through like, I expect this crew to update tickets. And I’m not gonna

33 00:04:45.690 00:04:53.140 Uttam Kumaran: sit here and go tell me the due date on this, like we’ll be a lot more candid. I think the 3rd things I want to just

34 00:04:53.250 00:05:01.030 Uttam Kumaran: sort of like, have this be more of demos so like, for example, there’s a couple of things that are in review. Let’s review them in this meeting.

35 00:05:01.526 00:05:12.209 Uttam Kumaran: Let’s sort of all play around and and test out like the stuff that’s that’s coming out the door. Or take a look at S. 3, and then just dive into like, what are problems?

36 00:05:12.750 00:05:14.850 Uttam Kumaran: The last piece I’ll mention is,

37 00:05:16.050 00:05:21.010 Uttam Kumaran: we’re starting to. We’re gonna we’re gonna definitely hit the limit on some of the things that we can throw into context.

38 00:05:21.130 00:05:24.549 Uttam Kumaran: And so we’ll have to learn new strategies for rag

39 00:05:25.056 00:05:33.640 Uttam Kumaran: and we’ll have to take time to do that. I’m happy to help facilitate that. But again, everything will be time box, and will be like discussion or

40 00:05:33.830 00:05:39.269 Uttam Kumaran: written right? So if we’re like, hey? We need to go. Learn about pre filtering, re-ranking

41 00:05:39.560 00:05:50.950 Uttam Kumaran: transaction based rag, cool. Someone. Go take the lead on that. Write up the document. Let’s have a meeting to review hash through how we’re gonna do it. Couple of Demos move forward right? So

42 00:05:51.580 00:06:02.660 Uttam Kumaran: you know, the concept of spikes, I think, are a good name for that. But that doesn’t mean open ended discovery. It means time box. We have a problem. We need to go research and find a solution. So

43 00:06:03.590 00:06:25.010 Uttam Kumaran: I think, for today’s meeting like, Give me a sense of where everything is, and let’s I mean, I know there were a bunch of things that were in review. I mentioned that I’m happy to go through and review those I know I can. I can see that some of those moved forward. So someone give me a sense of if there’s anything we can, even in this meeting, test out together.

44 00:06:26.800 00:06:28.180 Uttam Kumaran: And let’s go from there.

45 00:06:30.217 00:06:32.190 Miguel de Veyra: Yeah. Okay. So

46 00:06:33.220 00:06:44.589 Miguel de Veyra: for the interviews, we actually moved it. So, for example, the guidelines for documentation, we’ve sent that over to both demode and

47 00:06:44.750 00:06:45.910 Miguel de Veyra: sorry. Can you hear me.

48 00:06:47.030 00:06:47.660 Casie Aviles: Yes.

49 00:06:48.540 00:06:49.220 Uttam Kumaran: Yeah, yeah.

50 00:06:50.300 00:06:52.709 Miguel de Veyra: For both Demi Lade and

51 00:06:53.960 00:07:00.240 Miguel de Veyra: Kyle, so that this one, I think, will move to. I moved it to a client review, but I think we can move it to don

52 00:07:00.939 00:07:11.489 Miguel de Veyra: right now. The only in progress is pool parts. I was working on that slack thing just I mean, it’s still the context one, but just to get it initialized, for now.

53 00:07:11.980 00:07:16.620 Uttam Kumaran: Like on these. So like on these ones. So like, let’s say, urban stem slack agent.

54 00:07:18.031 00:07:20.138 Uttam Kumaran: here, actually, let me just

55 00:07:21.580 00:07:28.940 Uttam Kumaran: just gonna go into our cycle so initialize. So so

56 00:07:29.530 00:07:32.300 Uttam Kumaran: I can go into slack right now. And I can see these.

57 00:07:32.300 00:07:33.260 Miguel de Veyra: Yes, yes.

58 00:07:33.590 00:07:33.960 Uttam Kumaran: Okay, so.

59 00:07:33.960 00:07:37.210 Miguel de Veyra: How did the a screenshot or something?

60 00:07:37.650 00:07:38.745 Uttam Kumaran: Well, let’s just

61 00:07:39.230 00:07:40.590 Uttam Kumaran: Let’s just try it out.

62 00:07:40.830 00:07:41.630 Miguel de Veyra: Okay. Sure.

63 00:07:42.330 00:07:43.350 Uttam Kumaran: So.

64 00:07:44.396 00:07:50.980 Uttam Kumaran: We have the Javi, the Javi slack agent, Eden slack agent. So those are in the Eden channels.

65 00:07:51.680 00:07:52.920 Uttam Kumaran: or where are they?

66 00:07:52.920 00:07:54.550 Miguel de Veyra: In AI test.

67 00:07:55.400 00:07:59.180 Uttam Kumaran: Okay, cool.

68 00:08:04.250 00:08:06.430 Uttam Kumaran: Oh, what? The heck? Oh, sorry.

69 00:08:12.130 00:08:13.340 Uttam Kumaran: Okay, nice.

70 00:08:15.080 00:08:17.509 Uttam Kumaran: So let me unmute this.

71 00:08:24.970 00:08:29.019 Miguel de Veyra: But to note with them initialization is only for slack and zoom.

72 00:08:31.910 00:08:43.049 Uttam Kumaran: Okay? So I can say, like, Tell me about the 5 most recent meetings and what discussed.

73 00:08:50.310 00:08:56.160 Uttam Kumaran: So can we add, can we add a gif emoji here, like some sort of moving emoji.

74 00:08:57.597 00:08:58.189 Miguel de Veyra: I think so.

75 00:08:58.190 00:09:00.610 Uttam Kumaran: Do we have these like, for example, this one

76 00:09:02.050 00:09:06.009 Uttam Kumaran: something that indicates that’s like it’s on its way. And then once it’s ready.

77 00:09:07.860 00:09:13.209 Uttam Kumaran: Okay, so there’s 2 most reading meetings. So if I type something here, will it fake? Will it respond.

78 00:09:13.210 00:09:15.050 Miguel de Veyra: No, no, you have to tag the agent.

79 00:09:16.039 00:09:20.189 Miguel de Veyra: We’re still basically on that same issue where we can maintain the memory.

80 00:09:31.220 00:09:33.419 Uttam Kumaran: And can we? Can we get rid of this.

81 00:09:35.846 00:09:38.609 Miguel de Veyra: We’ll look into that. Yeah, we’ll take note of that.

82 00:09:48.954 00:09:53.219 Miguel de Veyra: Yeah, I can’t remember the last message you sent. It’s 1 is the one.

83 00:09:54.260 00:09:55.100 Uttam Kumaran: Why not?

84 00:09:57.158 00:10:00.500 Miguel de Veyra: I think this is the same issue we still had since before.

85 00:10:01.940 00:10:04.230 Uttam Kumaran: But you know how you can do chat memory right.

86 00:10:06.400 00:10:07.734 Miguel de Veyra: Yeah. Yeah. Yeah.

87 00:10:08.590 00:10:11.489 Uttam Kumaran: Like there is in the Api. You can do memory.

88 00:10:11.490 00:10:17.069 Miguel de Veyra: Yep, yep, it’s I think it’s okay. See? I think this is with more on slack right then.

89 00:10:17.740 00:10:24.450 Casie Aviles: We do have a window buffer memory attached to this bot, so I’m not sure why it’s

90 00:10:25.230 00:10:26.920 Casie Aviles: not checking the recent.

91 00:10:26.920 00:10:33.260 Uttam Kumaran: Yeah, right? Like, you can do, you basically can. Yeah, like, the, this agent context should

92 00:10:38.540 00:10:43.999 Uttam Kumaran: like this needs these, this whole thing needs to be appended to to this entire interaction.

93 00:10:45.170 00:10:49.090 Miguel de Veyra: Yeah, I think we can. Yeah, I’ll we’ll extend the context window for this one, then.

94 00:10:49.930 00:10:52.159 Miguel de Veyra: and then we’ll wait. Let me check. Actually. Now.

95 00:11:06.110 00:11:08.130 Uttam Kumaran: I mean, it is like buffer memory, right?

96 00:11:08.130 00:11:12.319 Miguel de Veyra: Yep, yep, we actually have that. We are.

97 00:11:12.320 00:11:19.290 Casie Aviles: Yeah, but I’m checking right now, and it only took the last 2 min. Yeah, just not.

98 00:11:19.290 00:11:20.859 Miguel de Veyra: It’s actually remembered.

99 00:11:20.980 00:11:22.119 Miguel de Veyra: Yeah, yeah.

100 00:11:25.570 00:11:29.650 Casie Aviles: We have to fix that and make sure that it’s getting all the.

101 00:11:31.140 00:11:34.547 Uttam Kumaran: Yeah, I’m gonna then I’m what I’m gonna do is I’m just gonna create. Let’s just

102 00:11:35.300 00:11:42.850 Uttam Kumaran: I’m just gonna track on here. So I’ll just put the list a couple of things slack. Bot!

103 00:11:43.490 00:11:49.370 Miguel de Veyra: Hey? See, I think I know it’s because the session id is different. It’s based on the time. Stop.

104 00:11:49.920 00:11:50.480 Miguel de Veyra: But this.

105 00:11:50.480 00:11:51.690 Uttam Kumaran: Maybe on the thread.

106 00:11:52.760 00:11:53.390 Miguel de Veyra: Yeah.

107 00:11:53.390 00:11:54.080 Uttam Kumaran: Right.

108 00:11:54.950 00:11:56.990 Miguel de Veyra: Yeah, I think this is more on the timestamp.

109 00:11:58.290 00:11:59.540 Miguel de Veyra: Let me double check.

110 00:12:10.130 00:12:12.509 Uttam Kumaran: Remove automated by N. 8 n.

111 00:12:12.850 00:12:23.880 Uttam Kumaran: Add loading gif first, st then once done, and check mark.

112 00:12:30.820 00:12:34.150 Miguel de Veyra: Okay, yeah. I think I found the fixer. I’ll discuss it with Casey later.

113 00:12:34.470 00:12:37.440 Casie Aviles: Yeah, you’re right. I. The problem is, it’s only.

114 00:12:37.440 00:12:40.040 Miguel de Veyra: Yeah. Not thread. Ts.

115 00:12:40.040 00:12:42.690 Casie Aviles: So like, each message is going to be unique.

116 00:12:42.890 00:12:43.340 Miguel de Veyra: Yes.

117 00:12:44.920 00:12:49.150 Uttam Kumaran: Okay, so can does can handle attachments.

118 00:12:49.790 00:12:50.360 Miguel de Veyra: Yeah.

119 00:12:51.280 00:12:53.810 Casie Aviles: No, no, we did not build that into it.

120 00:12:54.230 00:12:55.050 Miguel de Veyra: Should we add.

121 00:12:55.050 00:12:56.960 Uttam Kumaran: This one’s up, and you.

122 00:12:57.660 00:12:59.079 Miguel de Veyra: Oh, yeah. Good to ask.

123 00:13:07.520 00:13:13.030 Uttam Kumaran: Yeah, how hard is it for it to take in images and basically like, replicate

124 00:13:13.310 00:13:16.400 Uttam Kumaran: the interaction with that you would have with.

125 00:13:19.052 00:13:22.709 Uttam Kumaran: yeah, so basically rep replicating the interaction you would have with.

126 00:13:22.860 00:13:24.670 Miguel de Veyra: A chat gpt right.

127 00:13:26.060 00:13:30.120 Uttam Kumaran: Where you’re putting in like an image.

128 00:13:30.360 00:13:32.410 Miguel de Veyra: What was the title of that docusign? Sorry.

129 00:13:33.320 00:13:36.250 Uttam Kumaran: Flux, 7. Sow estimation, runbook.

130 00:13:36.830 00:13:43.599 Miguel de Veyra: Okay, so we have the URL private. We could probably, you know, if there’s a file, we could probably run a check if there’s

131 00:13:43.950 00:13:48.149 Miguel de Veyra: a file, or whatever downloaded, run some Ocr or get the text.

132 00:13:48.670 00:13:50.319 Miguel de Veyra: But yeah, there’s a way to do that.

133 00:13:50.520 00:13:54.879 Uttam Kumaran: Yeah, support, for let’s just do files and images.

134 00:13:56.160 00:14:00.749 Uttam Kumaran: I mean, it would be great if you can also do audio. But this is fine, for now.

135 00:14:00.750 00:14:01.380 Miguel de Veyra: Yeah, yeah.

136 00:14:12.440 00:14:15.920 Uttam Kumaran: support for file and images. And then I’m just gonna link this here.

137 00:14:19.940 00:14:26.469 Uttam Kumaran: Okay, it’s a call. Slack, bar quick fixes cool.

138 00:14:33.480 00:14:36.754 Uttam Kumaran: Alright, the the formatting looks fine.

139 00:14:41.170 00:14:48.579 Uttam Kumaran: So you mentioned that there’s a couple of other ones that are done right. So if I so the Eden client Hub. Tell me about

140 00:14:48.740 00:14:52.039 Uttam Kumaran: the last few meetings for Eden.

141 00:15:08.870 00:15:10.849 Uttam Kumaran: and then can I open this and.

142 00:15:15.740 00:15:19.390 Miguel de Veyra: Do you have a preference, by the way, on which Emoji to use? Or Jeff.

143 00:15:22.620 00:15:26.344 Uttam Kumaran: No, I I don’t care. I would prefer

144 00:15:27.140 00:15:30.359 Uttam Kumaran: I mean, you guys can totally decide doesn’t matter to me

145 00:15:30.740 00:15:34.120 Uttam Kumaran: like I’ve seen one other person use like

146 00:15:34.760 00:15:38.239 Uttam Kumaran: this one which is like this loading thing. It’s kind of hard to see.

147 00:15:38.430 00:15:39.460 Miguel de Veyra: Yeah. It’s all good.

148 00:15:40.280 00:15:43.880 Uttam Kumaran: But something that’s moving like, and it’s like sort of circular.

149 00:15:45.070 00:15:47.600 Miguel de Veyra: Or like thinking, or something like that, you know.

150 00:15:47.780 00:15:48.440 Miguel de Veyra: Okay.

151 00:15:49.540 00:15:51.199 Uttam Kumaran: Okay, so this one.

152 00:15:51.200 00:15:52.670 Miguel de Veyra: We need to fix the.

153 00:15:53.050 00:15:53.990 Uttam Kumaran: Formatting.

154 00:15:54.270 00:15:55.210 Miguel de Veyra: Formatting, yet.

155 00:15:55.970 00:15:57.499 Uttam Kumaran: Alright, but not bad.

156 00:16:04.420 00:16:09.829 Uttam Kumaran: so fixed, formatting for Eden.

157 00:16:09.830 00:16:11.920 Miguel de Veyra: And I think it will be the same.

158 00:16:11.920 00:16:14.439 Uttam Kumaran: Is there a formatting tool that you guys are using.

159 00:16:15.260 00:16:16.790 Miguel de Veyra: It’s another step.

160 00:16:19.260 00:16:21.069 Uttam Kumaran: Oh, for every single tool!

161 00:16:21.450 00:16:22.359 Miguel de Veyra: Yeah, yeah.

162 00:16:32.150 00:16:34.039 Uttam Kumaran: So if I go to like Eden

163 00:16:38.990 00:16:45.800 Uttam Kumaran: and like, okay, so let’s just talk through it. So okay.

164 00:16:53.050 00:16:57.060 Uttam Kumaran: and you’re using Google Gemini just for any reason in particular.

165 00:16:57.568 00:17:06.200 Miguel de Veyra: Because it has cause we’re still using for this one, the context, we’re still working on the rag stuff. So it has a million 2 million context. So

166 00:17:06.359 00:17:07.290 Miguel de Veyra: we put it there.

167 00:17:08.380 00:17:12.379 Uttam Kumaran: Okay, so message from slack react set variables. Yeah, this is fine.

168 00:17:12.720 00:17:14.859 Uttam Kumaran: And how much of this can you like?

169 00:17:15.490 00:17:21.589 Uttam Kumaran: Do we need to create an N. 8 n. Workflow for every single one. How much of can we reuse these like

170 00:17:22.599 00:17:25.750 Uttam Kumaran: like this, block or no?

171 00:17:25.750 00:17:28.679 Casie Aviles: Yeah, we we typically just clone this.

172 00:17:28.950 00:17:32.920 Miguel de Veyra: No, I mean, what he’s asking is if we, if we can have it in one place.

173 00:17:32.920 00:17:34.960 Uttam Kumaran: Yeah, have this in one place, and basically like.

174 00:17:34.960 00:17:36.980 Miguel de Veyra: It’s not possible. I don’t think so.

175 00:17:37.520 00:17:39.000 Uttam Kumaran: Cross, project.

176 00:17:39.680 00:17:41.934 Miguel de Veyra: We can look into it. Yeah.

177 00:17:42.830 00:17:47.309 Miguel de Veyra: message from Slack, because I think the bottleneck will be the message from Slack.

178 00:17:48.530 00:17:58.410 Uttam Kumaran: Message from slack is fine. But then these things, basically you can templatize right?

179 00:17:58.660 00:18:03.950 Uttam Kumaran: Well, can you? Well, I mean, you can certainly call. You can. Can you call another

180 00:18:04.100 00:18:05.960 Uttam Kumaran: workflow from in another workflow?

181 00:18:06.370 00:18:07.819 Miguel de Veyra: We are doing that here.

182 00:18:08.290 00:18:09.140 Uttam Kumaran: Where.

183 00:18:09.480 00:18:11.519 Miguel de Veyra: The zoom, the wait, the zoom.

184 00:18:11.520 00:18:12.509 Uttam Kumaran: Oh, yeah.

185 00:18:13.810 00:18:14.380 Miguel de Veyra: Yeah.

186 00:18:19.580 00:18:23.550 Uttam Kumaran: I mean, yeah, dude, because so that’s 1 thing I think to take a look at is like.

187 00:18:24.390 00:18:32.340 Uttam Kumaran: because half I would say, probably what you think like, 50% of of this is gonna be the same per client. Right? Cause? The only thing that’s gonna change is

188 00:18:32.840 00:18:34.310 Uttam Kumaran: the variables.

189 00:18:34.310 00:18:35.629 Miguel de Veyra: The table names. Yeah.

190 00:18:35.630 00:18:36.809 Uttam Kumaran: The table names.

191 00:18:36.960 00:18:39.580 Uttam Kumaran: Every client may have a different table name, but

192 00:18:40.290 00:18:42.349 Uttam Kumaran: the jet that way you can have one.

193 00:18:42.930 00:18:45.749 Uttam Kumaran: You can have basically have, like one thing for all the slack

194 00:18:48.950 00:18:51.179 Uttam Kumaran: Oh, but you do have a slack tool here. Okay.

195 00:18:51.180 00:19:00.600 Miguel de Veyra: Yeah, we separated it because the the thing the good thing about this is, for example, if it’s just the execution workflow, it doesn’t actually have to be on. So it doesn’t count on active workflows.

196 00:19:01.510 00:19:03.369 Uttam Kumaran: So where is this? So this slack.

197 00:19:03.890 00:19:05.180 Miguel de Veyra: Platform is the one below.

198 00:19:06.130 00:19:06.750 Uttam Kumaran: Oh!

199 00:19:06.750 00:19:09.239 Miguel de Veyra: You can have one execution per one workflow.

200 00:19:09.670 00:19:12.550 Miguel de Veyra: So it’s that one zoom tool is another.

201 00:19:12.550 00:19:15.010 Uttam Kumaran: Oh, this is the slack tool, then where’s the zoom tool?

202 00:19:15.304 00:19:19.129 Miguel de Veyra: It’s on another node, cause it’s it can only have one execution per.

203 00:19:20.790 00:19:25.970 Uttam Kumaran: So the Zoom One is this one.

204 00:19:25.970 00:19:26.630 Miguel de Veyra: Yeah.

205 00:19:30.330 00:19:32.599 Uttam Kumaran: Exactly so. What I’m saying is like.

206 00:19:33.040 00:19:35.840 Uttam Kumaran: instead of having one per client

207 00:19:36.550 00:19:38.870 Uttam Kumaran: like, instead of having 4 of these have one.

208 00:19:38.870 00:19:39.290 Miguel de Veyra: Open.

209 00:19:39.290 00:19:42.929 Uttam Kumaran: One of the input. The input variable is the client name.

210 00:19:43.250 00:19:48.489 Miguel de Veyra: Yeah, actually, there’s there’s a plan for that. So if you open the when executed by another workflow node

211 00:19:49.980 00:19:52.440 Miguel de Veyra: like that one, we added the

212 00:19:52.610 00:19:54.749 Miguel de Veyra: we added like a data source.

213 00:19:54.950 00:19:57.679 Miguel de Veyra: So basically, if it’s zoom, if it’s you know what.

214 00:19:57.680 00:19:58.810 Uttam Kumaran: Oh, nice!

215 00:19:58.810 00:20:04.289 Miguel de Veyra: There was. There’s a plan to go with that later down the line. But you know, we just wanted to get stuff out first.st

216 00:20:05.110 00:20:08.540 Uttam Kumaran: Okay, cool. So let me, just, I’m just gonna throw. I’ll just throw this in the backlog.

217 00:20:10.320 00:20:22.250 Uttam Kumaran: Well, one thing I’m gonna create is basically like, modularize, zoom, slack github tools.

218 00:20:26.180 00:20:35.629 Uttam Kumaran: So basically say, one tool per source input is client name or whatever you need.

219 00:20:36.740 00:20:37.210 Miguel de Veyra: We can probably.

220 00:20:37.210 00:20:37.560 Uttam Kumaran: Cost.

221 00:20:37.560 00:20:38.450 Miguel de Veyra: Basically.

222 00:20:38.800 00:20:44.130 Uttam Kumaran: Yeah, yeah, yeah, that way, we can cut. We can cut probably like

223 00:20:44.370 00:20:47.880 Uttam Kumaran: half of the stuff that we have in here. But that’s fine. Okay? So.

224 00:20:49.980 00:20:58.630 Miguel de Veyra: Actually, there’s a way. Also, we might want to separate it by project, like, it’s the folder system of anything. Because right now it’s on like overview. Everything.

225 00:21:00.720 00:21:01.680 Uttam Kumaran: What do you mean?

226 00:21:02.310 00:21:07.510 Miguel de Veyra: Like, if you go there under projects like there’s under, you know.

227 00:21:08.030 00:21:10.549 Miguel de Veyra: Yeah, we should. We should probably organize it there.

228 00:21:10.930 00:21:13.850 Uttam Kumaran: Oh, yeah, where is this all now?

229 00:21:15.260 00:21:16.950 Miguel de Veyra: It’s all in overview.

230 00:21:26.010 00:21:27.420 Uttam Kumaran: Oh, you have to move it!

231 00:21:27.420 00:21:28.979 Miguel de Veyra: Yeah, yeah, we’ll. We’ll we’ll do that. We’ll do that.

232 00:21:28.980 00:21:30.120 Uttam Kumaran: Okay, okay, I don’t really.

233 00:21:32.870 00:21:37.270 Miguel de Veyra: Should. Yeah, we’ll do one project per one client or just one, I think just one.

234 00:21:38.500 00:21:40.940 Uttam Kumaran: I think you do a project per client.

235 00:21:41.300 00:21:43.139 Miguel de Veyra: Okay, we’ll do that. Then.

236 00:21:44.340 00:21:46.309 Uttam Kumaran: So have one for ABC have one for us.

237 00:21:48.548 00:21:51.830 Uttam Kumaran: Okay, cool. So what else do we have to review? So

238 00:21:54.520 00:21:58.260 Uttam Kumaran: we tested that the urban stem slack agent. Okay, let’s test that now

239 00:21:58.410 00:22:00.729 Uttam Kumaran: cool part. So let’s just try both. This.

240 00:22:08.330 00:22:12.319 Uttam Kumaran: So initialization includes which sources again.

241 00:22:12.320 00:22:13.620 Miguel de Veyra: Black, and zoom only.

242 00:22:15.250 00:22:17.910 Uttam Kumaran: And then are there tickets for, like the other ones.

243 00:22:18.287 00:22:25.399 Miguel de Veyra: Yeah, yeah, we’re working on that right now. Github is the priority linear will do when we’re done.

244 00:22:26.125 00:22:30.300 Uttam Kumaran: Okay, okay. So pool part. So tell me about

245 00:22:30.460 00:22:43.840 Uttam Kumaran: who was in the last meeting, what was discussed, and rate the conversation in terms of friendliness.

246 00:22:46.830 00:22:48.427 Uttam Kumaran: And then what

247 00:22:55.790 00:22:58.230 Uttam Kumaran: Oh, you’re using 4 0 for everything. Okay.

248 00:22:58.550 00:22:59.100 Miguel de Veyra: Yeah.

249 00:22:59.770 00:23:04.040 Uttam Kumaran: And can you, Casey, can you? Where in azure can I see the usage of these.

250 00:23:06.590 00:23:09.559 Casie Aviles: I think you should call and let me in.

251 00:23:09.720 00:23:10.930 Casie Aviles: Get the link.

252 00:23:23.130 00:23:23.930 Uttam Kumaran: Nice

253 00:23:28.480 00:23:29.380 Uttam Kumaran: peace.

254 00:23:44.990 00:23:50.599 Uttam Kumaran: I mean, for now I feel like with the Github. You could probably toss it all into contacts, I guess, but.

255 00:24:20.620 00:24:23.220 Miguel de Veyra: Let me check Pool price cause that was wrong.

256 00:24:31.890 00:24:37.110 Awaish Kumar: Like for Gpt models right now, like we are the the message.

257 00:24:37.600 00:24:46.880 Awaish Kumar: the the call Api call goes along with context, and we get a reply, or do we? Are we managing customized Gpt models.

258 00:24:47.670 00:24:50.500 Miguel de Veyra: No, no, we’re not managing. We’re not customizing modules.

259 00:24:51.100 00:24:53.359 Casie Aviles: We have them deployed on azure.

260 00:24:53.360 00:24:55.589 Miguel de Veyra: Yeah, yeah. But we didn’t really touch them.

261 00:24:56.050 00:24:58.989 Awaish Kumar: Okay, is it possible, like we customize the models?

262 00:24:59.890 00:25:00.420 Awaish Kumar: Oh.

263 00:25:00.420 00:25:04.760 Miguel de Veyra: We could. We could fine tune it, but we didn’t. We have not looked into that at all.

264 00:25:05.820 00:25:10.099 Awaish Kumar: Okay. So okay, and we don’t know the cost of the doing that as well. Right.

265 00:25:10.100 00:25:15.180 Uttam Kumaran: Yeah, I mean, I also don’t know like what it would help with yet, because we certainly haven’t like

266 00:25:15.570 00:25:18.920 Uttam Kumaran: isolated the types of questions we typically get, you know.

267 00:25:19.699 00:25:34.979 Awaish Kumar: Like. What I was thinking is that like, like, for example, the slack, slack thing we are discussing. So we are adding, Fill, we have to filter the messages, then provide the context and get a message. But if we have a customized one, we

268 00:25:35.120 00:25:41.690 Awaish Kumar: we don’t need to send all the message messages again, like we just add additional slack messages and get a

269 00:25:42.000 00:25:44.790 Awaish Kumar: for the specific client and get the response.

270 00:25:45.356 00:25:52.289 Awaish Kumar: I like, that’s something which we have been doing in the previous company. I work so they were creating customized

271 00:25:52.430 00:25:57.760 Awaish Kumar: Gpt models for every client, and add only the additional data in

272 00:25:58.260 00:26:01.830 Awaish Kumar: or to fine, tune it and get the answer.

273 00:26:03.330 00:26:08.679 Uttam Kumaran: Hmm, interesting. Okay.

274 00:26:15.430 00:26:17.620 Uttam Kumaran: yeah. Casey, where do I even go?

275 00:26:18.750 00:26:20.320 Casie Aviles: Yeah. I’m not sure where.

276 00:26:20.320 00:26:20.810 Miguel de Veyra: You know.

277 00:26:20.810 00:26:23.460 Casie Aviles: Cost management is.

278 00:26:30.380 00:26:32.780 Uttam Kumaran: The azure AI foundry portal.

279 00:26:44.100 00:26:45.819 Uttam Kumaran: This is the one we’re using right.

280 00:26:46.030 00:26:47.950 Miguel de Veyra: Yes, that. That plan.

281 00:26:49.420 00:26:50.710 Casie Aviles: There’s like.

282 00:26:51.300 00:26:52.529 Uttam Kumaran: Okay, here we go.

283 00:26:53.620 00:26:54.500 Miguel de Veyra: Oh, there you go!

284 00:26:56.130 00:27:00.140 Uttam Kumaran: So I’m just gonna basically save this somewhere here.

285 00:27:00.690 00:27:02.460 Miguel de Veyra: What browser is this with? I’m sorry.

286 00:27:02.460 00:27:03.210 Uttam Kumaran: This is ark.

287 00:27:04.560 00:27:06.459 Miguel de Veyra: Dude you should use. Ark. It’s great.

288 00:27:06.460 00:27:07.260 Miguel de Veyra: Yeah, yeah.

289 00:27:08.100 00:27:15.100 Uttam Kumaran: Like. For example, if I want to go like, if I want to say cool, I want to do this. I can hit Alt, open it in the in the side panel.

290 00:27:17.310 00:27:18.950 Miguel de Veyra: Is it paid, or is it free?

291 00:27:18.950 00:27:19.840 Uttam Kumaran: No, it’s free.

292 00:27:19.970 00:27:23.550 Uttam Kumaran: It’s basically chrome, except you can have profiles.

293 00:27:23.970 00:27:26.570 Uttam Kumaran: So I have one profile for all of our clients.

294 00:27:26.890 00:27:28.550 Uttam Kumaran: Amber uses this, too.

295 00:27:29.460 00:27:30.190 Uttam Kumaran: Okay.

296 00:27:30.190 00:27:31.580 Miguel de Veyra: Yeah, I’ll start using that.

297 00:27:31.930 00:27:32.790 Uttam Kumaran: Yeah.

298 00:27:33.071 00:27:39.260 Miguel de Veyra: Going back to that slack thing. I adjusted some stuff and then resent your message. Your message. It’s a bit better now.

299 00:27:39.590 00:27:41.589 Uttam Kumaran: Okay. Can I delete some of these.

300 00:27:42.198 00:27:46.310 Miguel de Veyra: No, no, no, please don’t not yet cause we don’t know which one’s active and which one’s not.

301 00:27:49.075 00:27:49.470 Miguel de Veyra: Oh.

302 00:27:50.210 00:27:52.340 Casie Aviles: Development’s not active, though.

303 00:27:53.100 00:27:59.800 Uttam Kumaran: Okay, let me delete the vellum one clean up auto deployments.

304 00:28:00.420 00:28:05.830 Uttam Kumaran: Because I also wanna I’m gonna create one, for I’m gonna create O 3 and.

305 00:28:06.570 00:28:08.879 Miguel de Veyra: Oh, yeah, because our Openai doesn’t have O, 3.

306 00:28:10.830 00:28:13.610 Uttam Kumaran: Yeah. So dude O, 3 is way better.

307 00:28:14.790 00:28:16.479 Uttam Kumaran: I’m gonna delete this one

308 00:28:23.210 00:28:32.080 Uttam Kumaran: So I think for us a way to figure out how to get the the usage into

309 00:28:33.280 00:28:34.520 Uttam Kumaran: S. 3. Somehow.

310 00:28:35.440 00:28:36.050 Awaish Kumar: Okay.

311 00:28:36.050 00:28:42.115 Uttam Kumaran: I don’t know. I don’t know where the heck this is even coming from, but

312 00:28:43.950 00:28:46.310 Awaish Kumar: What we want to get into. S. 3.

313 00:28:46.810 00:28:49.810 Uttam Kumaran: I wanna get I wanna I basically wanna get all this like.

314 00:28:50.000 00:28:50.390 Awaish Kumar: Okay.

315 00:28:50.390 00:28:55.400 Uttam Kumaran: Number of requests. I want to get the token counts cause.

316 00:28:55.400 00:28:56.880 Miguel de Veyra: Be an Api for that.

317 00:28:57.860 00:28:58.610 Uttam Kumaran: Yeah.

318 00:29:01.480 00:29:04.239 Awaish Kumar: Okay? And I don’t have access to azure.

319 00:29:05.090 00:29:10.450 Uttam Kumaran: Oh, Casey, do you know how to do that?

320 00:29:11.790 00:29:13.239 Uttam Kumaran: Did a while ago?

321 00:29:14.200 00:29:15.880 Uttam Kumaran: Oh.

322 00:29:15.880 00:29:16.480 Miguel de Veyra: Oh!

323 00:29:19.530 00:29:21.030 Uttam Kumaran: Okay, let me check.

324 00:29:28.135 00:29:29.290 Uttam Kumaran: Okay, hold on.

325 00:29:29.630 00:29:31.170 Miguel de Veyra: Probably some sort of IM no.

326 00:29:32.160 00:29:36.515 Uttam Kumaran: Yeah, I just don’t even know I can do it from here. Okay,

327 00:29:37.595 00:29:38.239 Miguel de Veyra: There you go!

328 00:29:47.000 00:29:49.129 Uttam Kumaran: Is there some sort of global Iam?

329 00:29:51.070 00:29:53.920 Uttam Kumaran: Oh, wait! Should we use azure like I am for.

330 00:29:58.470 00:30:03.720 Awaish Kumar: Okay. I’ve used. I am, but not the 0 ion. Right?

331 00:30:04.190 00:30:09.010 Uttam Kumaran: It’s similar. Yeah, I just, I just want to give someone else ownership of this, because

332 00:30:09.300 00:30:12.549 Uttam Kumaran: this is like, not even I hate this.

333 00:30:20.540 00:30:21.986 Uttam Kumaran: Oh, God,

334 00:30:23.600 00:30:24.520 Uttam Kumaran: But.

335 00:30:40.100 00:30:42.669 Awaish Kumar: Is the azure slow, or or the Internet.

336 00:30:44.680 00:30:46.409 Uttam Kumaran: I saw azure’s. Yeah.

337 00:30:51.140 00:30:56.469 Uttam Kumaran: I mean, I guess, do I have to like just add you here literally, or what the heck.

338 00:31:00.440 00:31:03.450 Awaish Kumar: Yeah, on the top, it says, add right.

339 00:31:04.180 00:31:04.775 Uttam Kumaran: Oh.

340 00:31:13.360 00:31:14.040 Uttam Kumaran: oh.

341 00:31:21.490 00:31:23.590 Uttam Kumaran: this is like insane

342 00:31:30.610 00:31:33.189 Uttam Kumaran: like how do I just like Clone?

343 00:31:35.480 00:31:36.365 Uttam Kumaran: Horrible!

344 00:31:50.520 00:31:54.200 Uttam Kumaran: Oh, see, it’s not even. You’re not even like a member on here.

345 00:32:02.040 00:32:03.620 Uttam Kumaran: Let’s try this again.

346 00:32:06.100 00:32:19.120 Uttam Kumaran: Cost management plus billing. Nope users basis. Okay, this is it new user create new user?

347 00:32:19.800 00:32:21.710 Uttam Kumaran: And what did I call before.

348 00:32:22.000 00:32:26.639 Uttam Kumaran: What was the naming Convention? Oh, let me shh.

349 00:32:38.530 00:32:39.190 Awaish Kumar: Oh!

350 00:32:46.740 00:32:49.869 Uttam Kumaran: this is where, like, I don’t even know what? Roles? I did.

351 00:32:57.670 00:33:03.780 Uttam Kumaran: Okay, well, i’ll just give you AI i’ll give you billing,

352 00:33:24.470 00:33:26.709 Miguel de Veyra: Oh! Might want to add reports, either

353 00:33:30.610 00:33:31.320 Miguel de Veyra: under.

354 00:33:31.320 00:33:37.460 Uttam Kumaran: Oh, nice good call! Oh, but this is something else there’s like for sign in.

355 00:33:37.860 00:33:38.610 Miguel de Veyra: Okay.

356 00:33:42.490 00:33:43.150 Miguel de Veyra: yeah.

357 00:33:44.798 00:33:45.950 Miguel de Veyra: Global admin. Okay, that’s.

358 00:33:45.950 00:33:47.150 Uttam Kumaran: Alright cool.

359 00:33:53.574 00:33:57.740 Uttam Kumaran: Okay, I’m gonna just create this.

360 00:34:02.340 00:34:04.340 Uttam Kumaran: Well, alright, hold on.