2025-11-13_brainforge_assistant_project_check-in_f20181b4

Meeting Title: Brainforge Assistant Project Check-in Date: 2025-11-13 Meeting participants: Uttam Kumaran, Gabriel Lam, Samuel Roberts, Rico Rejoso, Mustafa Raja, Casie Aviles, Henry Zhao, Awaish Kumar, Zoran Selinger, Robert Tseng, Demilade Agboola, Amber Lin

---

WEBVTT

1 00:00:14.260 ⇒ 00:00:15.220 Samuel Roberts: Blue.

2 00:00:15.920 ⇒ 00:00:16.890 Uttam Kumaran: Hey, good morning.

3 00:00:17.240 ⇒ 00:00:18.429 Samuel Roberts: Morning, how are you?

4 00:00:21.090 ⇒ 00:00:21.820 Samuel Roberts: Got it.

5 00:00:59.160 ⇒ 00:01:03.359 Uttam Kumaran: Cool. Maybe… Gabe, I’ll let you lead.

6 00:01:06.060 ⇒ 00:01:16.229 Gabriel Lam: Yeah, so, current updates on the Assistant is… Mustafa was able to… Deploy…

7 00:01:16.530 ⇒ 00:01:18.660 Gabriel Lam: Some of the updates we talked about yesterday.

8 00:01:19.660 ⇒ 00:01:27.950 Gabriel Lam: The… we’re able to… test and launch the interviews at a pretty good

9 00:01:28.330 ⇒ 00:01:35.760 Gabriel Lam: level, I believe. I think next steps today would probably be to refine and polish the…

10 00:01:35.930 ⇒ 00:01:38.380 Gabriel Lam: Pre- and post-interview stages.

11 00:01:38.510 ⇒ 00:01:53.670 Gabriel Lam: I think the interview itself is at a pretty good spot. The agents have been updated, and the CRUD UI at the beginning, I think, is at a pretty good spot. We’re able to test, we’re able to have the different statuses, and for the marketing team, they should be able to

12 00:01:54.030 ⇒ 00:02:02.789 Gabriel Lam: see what’s open, what’s not open. I think something that we can also do is all… is to add filters, just to make it easier to differentiate between

13 00:02:03.140 ⇒ 00:02:07.330 Gabriel Lam: Projects, statuses, and…

14 00:02:08.090 ⇒ 00:02:14.449 Gabriel Lam: We can finalize, like, what happens after the copy is done, and whether we can even do some of the diagramming, or…

15 00:02:14.940 ⇒ 00:02:20.460 Gabriel Lam: Where this assistance sits in the entire ecosystem of Case study workflow.

16 00:02:20.720 ⇒ 00:02:22.660 Gabriel Lam: Yeah.

17 00:02:24.710 ⇒ 00:02:25.380 Uttam Kumaran: Okay.

18 00:02:25.680 ⇒ 00:02:26.550 Uttam Kumaran: Cool.

19 00:02:26.790 ⇒ 00:02:31.699 Uttam Kumaran: So I guess, tell me what, like, what is the…

20 00:02:32.960 ⇒ 00:02:42.480 Uttam Kumaran: plan for today, and then, I guess, if we don’t have other updates, maybe we could even just, like, walk through a live demo, and I can record

21 00:02:43.140 ⇒ 00:02:46.189 Uttam Kumaran: You know, a case study that’s just been sitting for a while.

22 00:02:47.730 ⇒ 00:02:50.429 Gabriel Lam: I mean, I’d be happy to hear from Mustafa and Casey about

23 00:02:50.750 ⇒ 00:02:55.269 Gabriel Lam: Maybe more detail about the updates, or if there’s any blockers that they’ve faced so far.

24 00:02:56.810 ⇒ 00:03:04.350 Mustafa Raja: Yeah, so the… so from my side, it’s only the, I was able to add the… add the playback speed.

25 00:03:04.470 ⇒ 00:03:13.729 Mustafa Raja: selector, but the issue is, if we select a higher speed, the pitch, for AI gets higher.

26 00:03:14.470 ⇒ 00:03:17.569 Mustafa Raja: So… so it becomes a higher-pitched voice.

27 00:03:17.970 ⇒ 00:03:21.900 Mustafa Raja: And… For that,

28 00:03:22.340 ⇒ 00:03:29.739 Mustafa Raja: We need to do some sort of time stretching algorithms with some of the libraries that we have.

29 00:03:30.030 ⇒ 00:03:35.250 Mustafa Raja: Available. Other than that, it should be good.

30 00:03:35.900 ⇒ 00:03:36.590 Uttam Kumaran: Okay.

31 00:03:40.450 ⇒ 00:03:47.170 Uttam Kumaran: Cool. So I guess, yeah, what’s the best use… I feel like we’re probably in a good spot. What’s the best use of,

32 00:03:47.840 ⇒ 00:03:50.729 Uttam Kumaran: of… This time.

33 00:03:50.730 ⇒ 00:04:00.000 Gabriel Lam: I think a live demo would be great. I just have one quick question to Mustafa, which is, I saw that there is the additional resources, like the hyperlinks.

34 00:04:03.190 ⇒ 00:04:04.810 Gabriel Lam: Do we feel like the…

35 00:04:05.520 ⇒ 00:04:13.149 Gabriel Lam: as you mentioned, is it better to summarize the additional context and then pass it on, or to just add it on as context? And maybe this is a wider question for everyone.

36 00:04:14.040 ⇒ 00:04:16.980 Mustafa Raja: Yeah, I think that we should,

37 00:04:17.070 ⇒ 00:04:34.009 Mustafa Raja: Firstly, so for now, it takes whatever link we would give it. I think we should, for now, at least we should restrict it to the meetings link from platform, so I can add some sort of regular expression, so that it would only accept

38 00:04:34.040 ⇒ 00:04:40.249 Mustafa Raja: Links that… links to, for the meetings coming from platform.

39 00:04:40.430 ⇒ 00:04:49.839 Mustafa Raja: And then once we have those, we can easily pull the transcript from Superbase, and then maybe we should add a summarization step.

40 00:04:50.020 ⇒ 00:04:53.819 Mustafa Raja: And then add that to the case study architect.

41 00:04:54.710 ⇒ 00:04:56.190 Gabriel Lam: Okay, that sounds good.

42 00:04:56.240 ⇒ 00:04:59.830 Samuel Roberts: I don’t know if we need to limit it, though, for now. I think what we can do…

43 00:05:00.200 ⇒ 00:05:05.139 Samuel Roberts: Is, like, let other links get added, and then filter what we add

44 00:05:05.900 ⇒ 00:05:08.070 Samuel Roberts: Potentially to the context for now.

45 00:05:08.460 ⇒ 00:05:12.409 Samuel Roberts: But I still think we want resources tied to these case studies for the future.

46 00:05:15.950 ⇒ 00:05:21.269 Samuel Roberts: Like, a Notion doc or something, or, you know, any other documentation.

47 00:05:21.980 ⇒ 00:05:23.099 Samuel Roberts: I would say let them.

48 00:05:23.100 ⇒ 00:05:23.890 Mustafa Raja: the foot…

49 00:05:24.180 ⇒ 00:05:24.830 Samuel Roberts: Go ahead.

50 00:05:25.400 ⇒ 00:05:45.169 Mustafa Raja: Yeah, so for now, what’s happening is we… we are only, linking, sorry, we are only adding the resources, right? So, agents right now do not have access to them directly. So, do we want the agents to be able to look into the links?

51 00:05:46.360 ⇒ 00:05:48.930 Mustafa Raja: And how… if so, how do we want to do that?

52 00:05:49.490 ⇒ 00:05:53.229 Samuel Roberts: Yeah, I mean, I think… Having them there is…

53 00:05:53.660 ⇒ 00:06:00.450 Samuel Roberts: the most important thing just for marketing to be able to have access to them. I think depending on what they are.

54 00:06:00.780 ⇒ 00:06:07.650 Samuel Roberts: We may want to try adding them, I mean, I think it… The current state, I don’t…

55 00:06:08.150 ⇒ 00:06:15.359 Samuel Roberts: I don’t know for a fact, but I think the current state is really just using the interview, right? And then marketing might use other resources afterwards?

56 00:06:18.620 ⇒ 00:06:30.230 Samuel Roberts: I’m not sure if that’s gonna add much value to it or not right now. I think having them there is kind of a future-proofing, making sure that we can tie these things together, kind of building that graphic knowledge eventually, but…

57 00:06:31.010 ⇒ 00:06:31.640 Gabriel Lam: Okay.

58 00:06:32.260 ⇒ 00:06:37.759 Samuel Roberts: I don’t know, Utam, do you have any thoughts on whether or not it’s worth feeding those, like, at least the meeting transcripts and stuff that get added?

59 00:06:43.520 ⇒ 00:06:44.880 Uttam Kumaran: I don’t…

60 00:06:45.230 ⇒ 00:06:49.409 Uttam Kumaran: I guess, can you show a visual of it? I guess I’m not following exactly what it is.

61 00:06:49.850 ⇒ 00:06:53.530 Samuel Roberts: So the idea is that we’re adding these, links…

62 00:06:53.770 ⇒ 00:07:00.349 Samuel Roberts: As, like, additional resources. Which, one thing, I’m not sure, how do we add those right now? Is it only at creation time, or…

63 00:07:01.390 ⇒ 00:07:02.250 Mustafa Raja: During the interview.

64 00:07:03.010 ⇒ 00:07:03.390 Samuel Roberts: During…

65 00:07:03.390 ⇒ 00:07:04.700 Mustafa Raja: Excuse me, yeah.

66 00:07:04.700 ⇒ 00:07:08.250 Samuel Roberts: Okay, I think we probably want to be able to add those…

67 00:07:09.890 ⇒ 00:07:12.919 Samuel Roberts: Afterwards, too, so if, like, something needs to get added later.

68 00:07:13.820 ⇒ 00:07:19.789 Samuel Roberts: Like, being able to edit those resources probably is something we want, but… the idea, I think.

69 00:07:19.790 ⇒ 00:07:25.920 Mustafa Raja: Yeah, we can always go back to the interview interface and then be able to edit whatever.

70 00:07:27.830 ⇒ 00:07:31.089 Samuel Roberts: Go back to the in… oh, so we’d have to go to…

71 00:07:31.510 ⇒ 00:07:32.559 Gabriel Lam: So who…

72 00:07:32.560 ⇒ 00:07:33.960 Mustafa Raja: It’s the link to the interview.

73 00:07:33.960 ⇒ 00:07:34.480 Gabriel Lam: link.

74 00:07:35.340 ⇒ 00:07:38.639 Mustafa Raja: Yeah, it’s in the… the interface is in the interview link, yeah.

75 00:07:38.640 ⇒ 00:07:39.370 Samuel Roberts: Okay.

76 00:07:40.520 ⇒ 00:07:45.279 Samuel Roberts: Okay, I guess that can make some sense. I was just thinking if marketing needed to, like, add something or tie something to it.

77 00:07:45.280 ⇒ 00:07:45.900 Mustafa Raja: Hmm.

78 00:07:45.900 ⇒ 00:07:49.950 Samuel Roberts: Do they need to go into the interview? You know, should we be able to do that from the case study page?

79 00:07:51.940 ⇒ 00:07:56.160 Gabriel Lam: I was just thinking that the interviewee probably has a better idea of what is…

80 00:07:56.160 ⇒ 00:07:56.580 Mustafa Raja: Yeah.

81 00:07:56.580 ⇒ 00:07:57.960 Gabriel Lam: So…

82 00:07:57.960 ⇒ 00:08:04.109 Samuel Roberts: No, I think it definitely should be on the interview page, too. I’m just wondering if marketing is like, oh, there’s a Notion page or something later.

83 00:08:04.110 ⇒ 00:08:04.650 Gabriel Lam: Yeah.

84 00:08:05.450 ⇒ 00:08:10.830 Mustafa Raja: I guess then we can also have an interface for marketing people to add that.

85 00:08:12.190 ⇒ 00:08:13.170 Mustafa Raja: Also.

86 00:08:13.170 ⇒ 00:08:13.660 Gabriel Lam: Yeah.

87 00:08:13.660 ⇒ 00:08:16.239 Samuel Roberts: Yeah, but I think, I think for now, this is…

88 00:08:17.090 ⇒ 00:08:19.659 Samuel Roberts: We can… we can add that, that’s just some cred stuff, but…

89 00:08:19.680 ⇒ 00:08:26.130 Gabriel Lam: Yeah, the idea is that if you add these links, especially if they’re meeting links, we could pull those transcripts and use that as part of the generation.

90 00:08:26.130 ⇒ 00:08:27.310 Samuel Roberts: For the case study.

91 00:08:30.080 ⇒ 00:08:35.270 Samuel Roberts: I suppose we could do that with other things, too, including, you know, if we could parse Notion, we could parse…

92 00:08:35.610 ⇒ 00:08:42.899 Samuel Roberts: Other things eventually, but… I’m not… my thought is just, like, I wanted to get this to a, like…

93 00:08:43.140 ⇒ 00:08:45.420 Samuel Roberts: Equal level with the current process.

94 00:08:46.910 ⇒ 00:08:48.210 Gabriel Lam: Yeah.

95 00:08:48.230 ⇒ 00:08:53.109 Samuel Roberts: Rather than worry about… Adding other things?

96 00:08:53.340 ⇒ 00:08:54.599 Gabriel Lam: Which I think it does.

97 00:08:54.990 ⇒ 00:09:03.230 Samuel Roberts: Yeah, no, I agree, I agree. I’m just… I’m… I… we can… we can try testing the transcripts. I just know that’s also going to add, potentially, a ton of context, which might…

98 00:09:05.310 ⇒ 00:09:07.629 Gabriel Lam: Hurt the output overall, to be honest.

99 00:09:07.630 ⇒ 00:09:08.340 Mustafa Raja: Yeah.

100 00:09:11.250 ⇒ 00:09:19.579 Samuel Roberts: Plus, like, if we wanted to add diagrams or something here, then it’s not necessarily something that needs to get added to the context, it might just be something that they’d want for the case study output.

101 00:09:19.580 ⇒ 00:09:20.080 Mustafa Raja: Hmm.

102 00:09:20.080 ⇒ 00:09:21.220 Samuel Roberts: Our final output?

103 00:09:21.530 ⇒ 00:09:29.989 Samuel Roberts: So I would say let’s not worry about that for now, is my initial reaction.

104 00:09:30.160 ⇒ 00:09:32.440 Samuel Roberts: But…

105 00:09:33.350 ⇒ 00:09:38.350 Samuel Roberts: we could definitely add that in the future, or test that at least, because like I said, I’m not sure if it’s going to improve the output or not.

106 00:09:38.550 ⇒ 00:09:43.229 Samuel Roberts: Just because context windows can… when you bloat them too much.

107 00:09:43.530 ⇒ 00:09:44.290 Gabriel Lam: Yeah.

108 00:09:44.290 ⇒ 00:09:46.530 Samuel Roberts: Get bad, so… And the.

109 00:09:46.530 ⇒ 00:09:47.270 Mustafa Raja: Which can be kind of.

110 00:09:47.270 ⇒ 00:09:51.600 Samuel Roberts: Depending on the meeting, so… Okay.

111 00:09:51.730 ⇒ 00:09:55.690 Samuel Roberts: But yeah, the resources… Maybe a live demo could be great. Yeah, yeah, I think let’s do that.

112 00:09:56.300 ⇒ 00:10:01.519 Samuel Roberts: So, Utam, did you say you wanted to talk through a case study? You wanted to do one?

113 00:10:02.640 ⇒ 00:10:05.799 Uttam Kumaran: Yeah, maybe I can share and we can walk through that.

114 00:10:05.800 ⇒ 00:10:06.690 Samuel Roberts: Yeah, totally.

115 00:10:11.150 ⇒ 00:10:13.119 Uttam Kumaran: And I can just go straight into the platform.

116 00:10:13.480 ⇒ 00:10:18.299 Samuel Roberts: I don’t think it’s merged yet as a PR, I think it’s still that link in the… I can send the link, it’s from the.

117 00:10:18.390 ⇒ 00:10:19.050 Uttam Kumaran: Okay.

118 00:10:19.230 ⇒ 00:10:23.020 Samuel Roberts: It’s on the… Oh, there it is. Okay, good. Thank you, Casey.

119 00:10:38.000 ⇒ 00:10:42.629 Samuel Roberts: Oh, yeah, the actual link. Yeah, I always clicked the wrong one from the Slack update.

120 00:10:55.330 ⇒ 00:10:59.089 Uttam Kumaran: Cool. So, in this situation, I’ll kind of play both…

121 00:10:59.740 ⇒ 00:11:08.250 Uttam Kumaran: folks, right? So let’s just see, I just want to confirm which ones… We’ve done before,

122 00:11:17.420 ⇒ 00:11:18.190 Uttam Kumaran: Okay.

123 00:11:18.890 ⇒ 00:11:22.100 Uttam Kumaran: Cool, so… well, is there anything on shipping?

124 00:11:50.020 ⇒ 00:11:52.129 Samuel Roberts: Oh, they’re not an active one, are they?

125 00:11:53.000 ⇒ 00:11:55.649 Uttam Kumaran: Yeah, so that’s gonna be one. Okay.

126 00:11:55.650 ⇒ 00:12:01.440 Samuel Roberts: Okay, so we can flip that to all the… yeah, because most of the time we’re filtering out active ones, but…

127 00:12:01.950 ⇒ 00:12:04.920 Samuel Roberts: We should keep it whatever is still in there, potentially, that’s a good point.

128 00:12:08.330 ⇒ 00:12:14.319 Samuel Roberts: Yeah, I wonder… Are you the only one that has two?

129 00:12:14.720 ⇒ 00:12:17.240 Uttam Kumaran: There’s… most likely, I’m under… there’s a…

130 00:12:17.240 ⇒ 00:12:19.609 Samuel Roberts: There’s another email probably in Google, I bet.

131 00:12:19.610 ⇒ 00:12:22.110 Uttam Kumaran: Yeah, there’s a few emails where I’m the…

132 00:12:22.370 ⇒ 00:12:25.050 Uttam Kumaran: first and last for, like, sales at Brain… for, like.

133 00:12:25.050 ⇒ 00:12:26.910 Samuel Roberts: Yeah, that’s right, okay.

134 00:12:26.910 ⇒ 00:12:33.919 Uttam Kumaran: You could just, you can actually probably just go in and change the first and last for those to, like, operations team or something like that.

135 00:12:34.450 ⇒ 00:12:35.649 Samuel Roberts: Yeah, that makes sense.

136 00:12:37.250 ⇒ 00:12:39.260 Mustafa Raja: Okay, so I’ll go ahead and hit create.

137 00:12:39.260 ⇒ 00:12:44.960 Uttam Kumaran: And then what is draft? Oh, so, okay, so…

138 00:12:50.370 ⇒ 00:12:51.550 Uttam Kumaran: Oh, okay, so…

139 00:12:51.790 ⇒ 00:12:52.529 Samuel Roberts: That must have been the other one.

140 00:12:53.380 ⇒ 00:12:54.249 Mustafa Raja: Yeah, must be.

141 00:12:54.250 ⇒ 00:12:58.240 Uttam Kumaran: So I guess question here is, like, what is draft versus awaiting interviews?

142 00:12:59.900 ⇒ 00:13:08.099 Gabriel Lam: So our goal was, during draft, it would mean, like, hey, either the description’s not right, or you’re still figuring out who the interviewees are.

143 00:13:08.290 ⇒ 00:13:14.980 Gabriel Lam: I think once you set it to awaiting interviews, it’s like, okay, that’s done, and then on the status, we know

144 00:13:15.310 ⇒ 00:13:18.059 Gabriel Lam: I think the word draft just sounds like it’s incomplete.

145 00:13:18.060 ⇒ 00:13:19.070 Uttam Kumaran: Yeah. Yeah.

146 00:13:19.070 ⇒ 00:13:21.309 Gabriel Lam: So we wanted a different state.

147 00:13:23.040 ⇒ 00:13:24.270 Uttam Kumaran: So I think, yeah.

148 00:13:25.030 ⇒ 00:13:29.310 Samuel Roberts: depending on, like, if these are ready to have the interviews go out, I think is the…

149 00:13:29.860 ⇒ 00:13:34.170 Samuel Roberts: Time it goes from draft to… Waiting for interview.

150 00:13:34.690 ⇒ 00:13:36.020 Samuel Roberts: Are we interview ready?

151 00:13:36.700 ⇒ 00:13:40.419 Samuel Roberts: Yeah, so I think… my thought, I think, was that it would…

152 00:13:40.790 ⇒ 00:13:49.730 Samuel Roberts: when you switch to awaiting interviews is when it would potentially send a Slack message or send a link, but it wouldn’t do that until you’ve flipped that switch over.

153 00:13:50.370 ⇒ 00:13:51.050 Uttam Kumaran: Okay, okay.

154 00:13:51.500 ⇒ 00:13:56.710 Gabriel Lam: We can maybe change… draft might be the… yeah, that might even be something we want to change the phrasing on there, just like…

155 00:13:56.820 ⇒ 00:14:01.189 Samuel Roberts: Or make it clear, have a callout or something that knows, like, interviews haven’t been sent or something.

156 00:14:02.470 ⇒ 00:14:03.120 Uttam Kumaran: Okay.

157 00:14:03.690 ⇒ 00:14:08.960 Samuel Roberts: Also, I just looked in the database, it’s… it’s actually your Tribe Brainforge email.

158 00:14:08.960 ⇒ 00:14:10.370 Uttam Kumaran: Oh, okay, okay.

159 00:14:10.370 ⇒ 00:14:13.130 Samuel Roberts: Which, do we need that in the system at all?

160 00:14:13.570 ⇒ 00:14:19.329 Uttam Kumaran: Yeah, because we sometimes use it as, like, an alt email.

161 00:14:20.610 ⇒ 00:14:22.299 Samuel Roberts: Okay, use that for the porridge, I mean?

162 00:14:23.220 ⇒ 00:14:25.389 Uttam Kumaran: No, I don’t use it to log in here, no.

163 00:14:25.390 ⇒ 00:14:27.029 Samuel Roberts: Okay, that’s what I thought. Yeah, I might just…

164 00:14:28.240 ⇒ 00:14:30.590 Samuel Roberts: I’m trying to think if it… If we don’t.

165 00:14:30.590 ⇒ 00:14:33.349 Uttam Kumaran: But I guess what you… I mean, what you could… yeah, I guess,

166 00:14:35.570 ⇒ 00:14:38.619 Uttam Kumaran: I mean, I’m not using it that often, it just exists.

167 00:14:38.670 ⇒ 00:14:45.250 Samuel Roberts: Yeah, well, it’s gonna sink down from Google, probably, if I delete it here in the table again, so we probably just need to make it…

168 00:14:46.190 ⇒ 00:14:51.590 Samuel Roberts: Well, I don’t know. I can maybe restrict this table to just at BrainForge AI emails, maybe?

169 00:14:52.980 ⇒ 00:14:54.230 Uttam Kumaran: Oh, yeah, that could be.

170 00:14:54.230 ⇒ 00:15:00.960 Samuel Roberts: Rather than… yeah, because I… there’s definitely going to be some in that Google Workspace. Let me… let me take a look at what’s there. I can change that script to only sync

171 00:15:01.440 ⇒ 00:15:07.330 Samuel Roberts: Yeah, that’s probably the only one it’ll catch, but we might as well do it.

172 00:15:08.360 ⇒ 00:15:11.669 Samuel Roberts: to avoid exactly this, so I’ll deal with that.

173 00:15:16.700 ⇒ 00:15:17.810 Uttam Kumaran: Okay…

174 00:15:17.810 ⇒ 00:15:18.750 Gabriel Lam: And then you…

175 00:15:19.200 ⇒ 00:15:23.710 Gabriel Lam: So this would be for the marketing team to see. If you go back to case studies.

176 00:15:24.090 ⇒ 00:15:29.000 Gabriel Lam: There’s an open interview link, which is the link that whoever gets

177 00:15:29.340 ⇒ 00:15:32.329 Gabriel Lam: the Slack notification would be connected to.

178 00:15:32.790 ⇒ 00:15:36.389 Uttam Kumaran: So, but can that… can the person still go into, like, this page?

179 00:15:39.130 ⇒ 00:15:41.419 Samuel Roberts: the person Or the…

180 00:15:41.420 ⇒ 00:15:42.220 Uttam Kumaran: Yeah.

181 00:15:42.220 ⇒ 00:15:42.650 Samuel Roberts: Yeah.

182 00:15:42.650 ⇒ 00:15:44.030 Mustafa Raja: Yeah, for now, yes.

183 00:15:44.030 ⇒ 00:15:45.679 Samuel Roberts: So we should probably have a way to link…

184 00:15:45.900 ⇒ 00:15:50.960 Uttam Kumaran: Yeah, well, I mean, what you’re gonna see is that, like, people are gonna do what I did, they’re just gonna click.

185 00:15:51.210 ⇒ 00:15:59.290 Samuel Roberts: Yeah, so there should be a way to, like, for the… if it’s the person who is the interviewee, there should be a way to say, like, start interview or something on this page.

186 00:15:59.660 ⇒ 00:16:00.200 Samuel Roberts: Or…

187 00:16:00.200 ⇒ 00:16:03.169 Uttam Kumaran: I should say start interview, and they should also be here, too.

188 00:16:04.780 ⇒ 00:16:11.560 Uttam Kumaran: Yeah, I don’t care if people see this, but, like, people may come here and then be like, where do I go to do the interview?

189 00:16:11.560 ⇒ 00:16:12.709 Samuel Roberts: Yeah, that should be able to be okay.

190 00:16:12.710 ⇒ 00:16:15.159 Uttam Kumaran: big-ass button that says, like, start interview, basically.

191 00:16:15.160 ⇒ 00:16:16.610 Samuel Roberts: Yeah. Yeah.

192 00:16:18.710 ⇒ 00:16:22.139 Samuel Roberts: I mean, ideally, they’ll go right from a link from Slack sometimes, too, so it’s…

193 00:16:22.640 ⇒ 00:16:23.440 Uttam Kumaran: Yeah.

194 00:16:23.940 ⇒ 00:16:26.929 Uttam Kumaran: I feel like I probably selected the wrong one again, right?

195 00:16:26.930 ⇒ 00:16:29.930 Samuel Roberts: Go to… go to Settings real quick. Who are you logged in as?

196 00:16:31.470 ⇒ 00:16:32.210 Mustafa Raja: Oh.

197 00:16:33.580 ⇒ 00:16:35.020 Samuel Roberts: Until my brain. No, okay, so.

198 00:16:35.020 ⇒ 00:16:37.999 Uttam Kumaran: Well, I think I may have selected the wrong one again.

199 00:16:38.000 ⇒ 00:16:39.639 Samuel Roberts: Oh, okay, okay.

200 00:16:40.150 ⇒ 00:16:43.949 Samuel Roberts: I can… here, let me just change that in Superbase real quick, so it’ll pull the right,

201 00:16:49.980 ⇒ 00:16:50.660 Samuel Roberts: Okay.

202 00:16:51.530 ⇒ 00:16:54.249 Samuel Roberts: Let’s see if I don’t have any more…

203 00:16:54.980 ⇒ 00:16:56.050 Uttam Kumaran: Okay, cool.

204 00:16:59.450 ⇒ 00:17:03.640 Uttam Kumaran: Okay, cool. So, like, I’ll just do, like, the first, like, 30 seconds of one and see how it goes.

205 00:17:07.589 ⇒ 00:17:10.909 Uttam Kumaran: I guess I’ll make sure to share audio with you guys, hold on.

206 00:17:11.670 ⇒ 00:17:12.510 Samuel Roberts: Oh, yeah.

207 00:17:17.720 ⇒ 00:17:20.790 Audio shared by Uttam Kumaran: Hello and welcome! I’ll be conducting a structured interview with you today.

208 00:17:20.790 ⇒ 00:17:21.349 Gabriel Lam: Yeah.

209 00:17:21.359 ⇒ 00:17:24.509 Audio shared by Uttam Kumaran: A shipping cost project for marketing. I understand the project details.

210 00:17:24.510 ⇒ 00:17:25.550 Mustafa Raja: This is what happens.

211 00:17:25.550 ⇒ 00:17:28.470 Samuel Roberts: Yeah, we’ll figure it out. That’s what he was just commenting about in the…

212 00:17:28.470 ⇒ 00:17:35.119 Uttam Kumaran: Yeah, so that’s definitely not what I meant by, like, speed it up, like, get it high-pitched. More is, like.

213 00:17:35.310 ⇒ 00:17:38.080 Uttam Kumaran: Actual words will come out faster, same pitch.

214 00:17:38.460 ⇒ 00:17:39.270 Mustafa Raja: Yeah.

215 00:17:39.270 ⇒ 00:17:46.379 Uttam Kumaran: try it in ChatGPT, like, talk to the woman or guy, and say, talk faster.

216 00:17:47.490 ⇒ 00:17:52.459 Uttam Kumaran: Yeah, I think what it’s doing is speeding it up, but when you speed it up, the pitch increases, and then they’re…

217 00:17:52.460 ⇒ 00:17:54.830 Samuel Roberts: They’re doing an algorithm to bring that back down.

218 00:17:55.180 ⇒ 00:17:55.800 Gabriel Lam: Yep.

219 00:17:56.980 ⇒ 00:17:57.320 Uttam Kumaran: we had.

220 00:17:57.320 ⇒ 00:17:57.890 Samuel Roberts: tab.

221 00:17:58.160 ⇒ 00:17:58.870 Uttam Kumaran: Yeah.

222 00:17:59.230 ⇒ 00:18:02.549 Uttam Kumaran: Or again, if you can’t figure it out, just leave it, I don’t mind.

223 00:18:02.550 ⇒ 00:18:03.920 Samuel Roberts: Yeah, that’s fair.

224 00:18:06.860 ⇒ 00:18:09.989 Audio shared by Uttam Kumaran: Hello and welcome. I’m glad to have you here.

225 00:18:10.190 ⇒ 00:18:16.410 Audio shared by Uttam Kumaran: We’ll be conducting a structured interview today about the shipping cost project for marketing.

226 00:18:16.800 ⇒ 00:18:20.409 Uttam Kumaran: So, another point is that it always starts back up here.

227 00:18:22.920 ⇒ 00:18:23.440 Mustafa Raja: They’re nice.

228 00:18:23.440 ⇒ 00:18:23.950 Samuel Roberts: Oh, God.

229 00:18:23.950 ⇒ 00:18:26.360 Mustafa Raja: To the, to the bottom.

230 00:18:26.880 ⇒ 00:18:28.109 Mustafa Raja: If he stopped into you.

231 00:18:28.110 ⇒ 00:18:30.100 Audio shared by Uttam Kumaran: Hello, and welcome.

232 00:18:30.290 ⇒ 00:18:31.789 Audio shared by Uttam Kumaran: I’m glad you’re here today.

233 00:18:31.790 ⇒ 00:18:33.129 Samuel Roberts: But it’s, it’s restarting. I’ll be deducting.

234 00:18:33.130 ⇒ 00:18:34.770 Audio shared by Uttam Kumaran: a structured interview with you.

235 00:18:34.770 ⇒ 00:18:40.530 Samuel Roberts: The shipping cost project for marketing. I see we have a blank project description.

236 00:18:40.530 ⇒ 00:18:48.970 Audio shared by Uttam Kumaran: But that’s okay. We’ll focus on gathering your insights and details today. Are you ready to begin with the first section, at a glance?

237 00:18:48.970 ⇒ 00:18:51.660 Uttam Kumaran: So what does that mean, blank project description?

238 00:18:52.020 ⇒ 00:18:54.879 Samuel Roberts: So if you go… Okay, alright.

239 00:18:57.510 ⇒ 00:18:59.579 Mustafa Raja: Oh, we didn’t give it any description, no.

240 00:18:59.770 ⇒ 00:19:00.430 Mustafa Raja: Yep.

241 00:19:02.130 ⇒ 00:19:02.660 Samuel Roberts: So that’s being.

242 00:19:02.660 ⇒ 00:19:03.200 Uttam Kumaran: Oh, but…

243 00:19:03.200 ⇒ 00:19:05.600 Samuel Roberts: So it has context about this project.

244 00:19:05.600 ⇒ 00:19:06.150 Uttam Kumaran: I see.

245 00:19:06.630 ⇒ 00:19:08.410 Uttam Kumaran: So that would be another thing where…

246 00:19:08.590 ⇒ 00:19:13.060 Uttam Kumaran: it should just, like, there’s no way for me to add that description here, right? Like…

247 00:19:13.060 ⇒ 00:19:22.780 Mustafa Raja: We can add, if we go to the… interviewees wouldn’t be able to add the description, so if we go back to the.

248 00:19:22.780 ⇒ 00:19:23.099 Samuel Roberts: But we may.

249 00:19:23.100 ⇒ 00:19:25.999 Uttam Kumaran: But this is where… this is where, like.

250 00:19:26.400 ⇒ 00:19:32.559 Uttam Kumaran: think about me as the user. You’re telling me that I have to go back and go click on it again.

251 00:19:32.560 ⇒ 00:19:37.319 Samuel Roberts: No, no, I think the current assumption was that marketing would create these things.

252 00:19:37.320 ⇒ 00:19:38.030 Uttam Kumaran: Yeah.

253 00:19:38.030 ⇒ 00:19:38.679 Samuel Roberts: And then this page.

254 00:19:38.680 ⇒ 00:19:40.360 Uttam Kumaran: No, no, no, it’s fair, it’s fair, I just…

255 00:19:40.360 ⇒ 00:19:44.089 Samuel Roberts: I think what you’re saying is that, like, we need to just merge those together, probably.

256 00:19:44.730 ⇒ 00:19:48.119 Uttam Kumaran: Yeah, I’m walking through it, like, me, right? So.

257 00:19:48.120 ⇒ 00:19:51.980 Samuel Roberts: But you’re right, yeah, and you’re doing both sides of this, which won’t normally be happening, I think.

258 00:19:51.980 ⇒ 00:19:57.469 Uttam Kumaran: No, but also what I’m gonna… what I’m saying is that marketing is not always gonna have all of the…

259 00:19:57.800 ⇒ 00:19:58.700 Samuel Roberts: That’s true.

260 00:19:58.700 ⇒ 00:20:02.529 Uttam Kumaran: So there’s no… there’s no reason to not just add the description here.

261 00:20:02.530 ⇒ 00:20:03.240 Samuel Roberts: Totally.

262 00:20:03.500 ⇒ 00:20:07.120 Uttam Kumaran: Yeah.

263 00:20:09.530 ⇒ 00:20:11.090 Audio shared by Uttam Kumaran: Hello, and welcome.

264 00:20:11.090 ⇒ 00:20:11.630 Samuel Roberts: We could also.

265 00:20:11.630 ⇒ 00:20:13.039 Audio shared by Uttam Kumaran: Thank you for joining me today.

266 00:20:13.040 ⇒ 00:20:14.170 Samuel Roberts: There’s no description.

267 00:20:14.170 ⇒ 00:20:15.929 Audio shared by Uttam Kumaran: I’ll be conducting a structured interview.

268 00:20:15.930 ⇒ 00:20:18.120 Samuel Roberts: with you about the shipping cost project.

269 00:20:18.520 ⇒ 00:20:19.549 Audio shared by Uttam Kumaran: for marketing.

270 00:20:20.040 ⇒ 00:20:24.290 Audio shared by Uttam Kumaran: We’re going to discuss various aspects of the project in a structured manner.

271 00:20:24.610 ⇒ 00:20:28.519 Audio shared by Uttam Kumaran: Are you ready to begin with the first section? At a glance?

272 00:20:28.960 ⇒ 00:20:32.500 Audio shared by Uttam Kumaran: Great! Let’s start with the basics of the project.

273 00:20:32.720 ⇒ 00:20:44.769 Audio shared by Uttam Kumaran: Could you tell me what type of project the shipping cost initiative was, how long it lasted, and how the team was composed? For example, how many people were involved, and what were their roles?

274 00:20:48.260 ⇒ 00:20:55.299 Uttam Kumaran: So, can I just… okay, so on the shipping cost project, it was me and another analytics engineer, Ryan.

275 00:20:55.410 ⇒ 00:21:03.479 Uttam Kumaran: There… it was probably, like, a three… Or 4-month project.

276 00:21:03.630 ⇒ 00:21:19.189 Uttam Kumaran: And we basically had to ingest all the shipping costs from FedEx and from UPS, all of the historical paid shipping costs from them. We also had to work with FedEx and UPS to get their rate cards.

277 00:21:19.300 ⇒ 00:21:29.239 Uttam Kumaran: In addition, not only the shipping cost, but we have to get the weights of every package and the dimensions, because of additional handling charges.

278 00:21:29.250 ⇒ 00:21:49.109 Uttam Kumaran: And so our first job was just to make sure that we have an accurate sense of past shipping costs, all the way down to the individual item that was shipped. So length, width, height, weight, and additional fees that were tacked on. And that way we sort of compare that to what we were billed.

279 00:21:49.150 ⇒ 00:21:51.250 Uttam Kumaran: Next,

280 00:21:51.640 ⇒ 00:22:07.410 Uttam Kumaran: we then needed to forecast forward. So our goal was to give them a forecast of their shipping costs for the next 2 years. At that point, I could then take that, call FedEx, and then say, hey, this is what we’re estimating to do.

281 00:22:07.700 ⇒ 00:22:20.979 Uttam Kumaran: we’re coming up on renewal, this is way too expensive. Additionally, we have all these fees that you charge us for, which we’d like to get rid of. What can you do? And so I called FedEx and UPS both. UPS gave us,

282 00:22:21.170 ⇒ 00:22:25.249 Uttam Kumaran: A really good offer, because our client was very seasonal.

283 00:22:25.470 ⇒ 00:22:35.150 Uttam Kumaran: They shipped a tremendous amount of volume in just, like, two parts of the year, and so a flat, like, rate

284 00:22:35.270 ⇒ 00:22:46.489 Uttam Kumaran: across the year wasn’t great. Like, we wanted something that scaled with volume, and that’s the deal we cut. So we ended up saving for them, I think, roughly, like, half a million dollars.

285 00:22:46.810 ⇒ 00:22:51.590 Uttam Kumaran: That year, in just shipping costs, just through that renegotiation.

286 00:22:52.870 ⇒ 00:22:56.089 Audio shared by Uttam Kumaran: That’s a comprehensive overview. Thank you.

287 00:22:56.400 ⇒ 00:23:00.110 Audio shared by Uttam Kumaran: It’s clear that the project had a significant impact.

288 00:23:00.300 ⇒ 00:23:08.199 Audio shared by Uttam Kumaran: Before we delve deeper into the context, is there anything else you’d like to add about the team, project duration? Scope.

289 00:23:09.200 ⇒ 00:23:10.060 Uttam Kumaran: No.

290 00:23:11.000 ⇒ 00:23:24.420 Audio shared by Uttam Kumaran: Alright, let’s move on to the next section, context. Can you describe the working environment before the project started? Specifically, what were the challenges or constraints related to shipping costs that may require to get things done?

291 00:23:25.440 ⇒ 00:23:28.630 Audio shared by Uttam Kumaran: Legacy systems, resource limitations, or other complexities.

292 00:23:29.620 ⇒ 00:23:30.929 Uttam Kumaran: It sounds like there was a lack of.

293 00:23:30.930 ⇒ 00:23:36.120 Audio shared by Uttam Kumaran: certain elements. Could you elaborate on what was missing or inadequate in the previous setup?

294 00:23:36.600 ⇒ 00:23:37.650 Uttam Kumaran: Told you what was missing.

295 00:23:37.650 ⇒ 00:23:47.829 Audio shared by Uttam Kumaran: to manage shipping costs effectively before your project. I see. So the key challenge was the absence of historical data on shipping costs, weights, dimensions, and.

296 00:23:47.830 ⇒ 00:23:49.390 Uttam Kumaran: No, there’s nothing else I wanna ask.

297 00:23:49.390 ⇒ 00:23:52.909 Audio shared by Uttam Kumaran: data, it was difficult to manage and forecast costs.

298 00:23:53.190 ⇒ 00:23:55.769 Audio shared by Uttam Kumaran: Were there any previous attempts? Got it.

299 00:23:55.880 ⇒ 00:24:01.789 Audio shared by Uttam Kumaran: The lack of historical data was the main issue, and there hadn’t been previous efforts that successfully.

300 00:24:01.790 ⇒ 00:24:03.770 Uttam Kumaran: Okay, let’s… I already gave you, basically.

301 00:24:03.770 ⇒ 00:24:06.729 Audio shared by Uttam Kumaran: Before we move on to the next section, is there anything else you’d like.

302 00:24:06.730 ⇒ 00:24:07.609 Uttam Kumaran: Go ahead and wrap up.

303 00:24:07.610 ⇒ 00:24:14.080 Audio shared by Uttam Kumaran: context before the project began. Thank you for clarifying. Let’s move on to the challenge section.

304 00:24:14.290 ⇒ 00:24:17.720 Audio shared by Uttam Kumaran: Can you highlight the specific pain points the marketing team…

305 00:24:18.220 ⇒ 00:24:21.920 Samuel Roberts: Still reading the audio coming back from previous ones, it looks like.

306 00:24:21.920 ⇒ 00:24:24.680 Gabriel Lam: Yeah, I think we need to just pause it whenever.

307 00:24:24.970 ⇒ 00:24:26.419 Gabriel Lam: Someone starts talking.

308 00:24:27.570 ⇒ 00:24:36.389 Uttam Kumaran: Well, I guess, like, what’s the… what is the… how is this working? Like, what is the backend for this? Are you guys using, like, you’re just using the real-time voice endpoint for this?

309 00:24:37.200 ⇒ 00:24:37.920 Mustafa Raja: Yes.

310 00:24:39.010 ⇒ 00:24:48.440 Uttam Kumaran: Is there not anything that does this sort of, like, interview-style… back and forth? Or… You’re handling it all.

311 00:24:48.440 ⇒ 00:24:49.639 Samuel Roberts: The speech-to-speech?

312 00:24:50.760 ⇒ 00:24:55.619 Mustafa Raja: I would say, it’s a speech-to-speech endpoint from… From them, yeah.

313 00:25:01.520 ⇒ 00:25:03.750 Uttam Kumaran: Okay, well, that was tough. Like, one…

314 00:25:03.750 ⇒ 00:25:15.869 Samuel Roberts: Yeah, I think… I think there must be… because it’s… it’s streaming back the audio still, but it’s… it’s showing other text that’s coming back, but I think it’s probably just we need to stop that audio ourselves, maybe?

315 00:25:17.150 ⇒ 00:25:17.860 Samuel Roberts: You know what I mean?

316 00:25:17.860 ⇒ 00:25:22.980 Uttam Kumaran: I don’t know, this is where, like, I would do… I would do some research, or go try out some open source…

317 00:25:23.440 ⇒ 00:25:34.370 Uttam Kumaran: repos that… that do ChatGPT-style interviews, and you can look at their code on how they’re doing it, because I don’t think that their people are just, like, kind of going with the raw…

318 00:25:34.570 ⇒ 00:25:37.779 Uttam Kumaran: speech-to-speech. I think there’s something that they’ve tweaked.

319 00:25:38.420 ⇒ 00:25:44.350 Uttam Kumaran: Or we need to kind of think about, like, yeah, how we package this, because…

320 00:25:44.550 ⇒ 00:25:47.260 Uttam Kumaran: One is, I don’t think it’s really…

321 00:25:48.040 ⇒ 00:25:53.830 Uttam Kumaran: I don’t particularly care about seeing… my end?

322 00:25:54.350 ⇒ 00:26:01.469 Uttam Kumaran: I guess it is interesting to see their end, but, like, the AI was, of course, like, way behind.

323 00:26:02.400 ⇒ 00:26:04.129 Samuel Roberts: My voice was way behind, is what I’m saying.

324 00:26:04.130 ⇒ 00:26:05.550 Uttam Kumaran: Yeah, the voice was way behind.

325 00:26:05.550 ⇒ 00:26:11.000 Samuel Roberts: But it was still processing stuff, I think we just probably need to nix the… previous playback.

326 00:26:12.190 ⇒ 00:26:15.289 Uttam Kumaran: Yeah, yeah, I guess, like, on interruption, maybe it should.

327 00:26:15.290 ⇒ 00:26:15.790 Samuel Roberts: Yeah.

328 00:26:15.790 ⇒ 00:26:20.090 Uttam Kumaran: I could do that. And the other thing was… like…

329 00:26:22.700 ⇒ 00:26:28.720 Uttam Kumaran: I just don’t get why I asked this. I’ve already talked about, like, I already gave it this.

330 00:26:29.550 ⇒ 00:26:34.619 Uttam Kumaran: So, I guess my point is there’s probably some prompt tweaking that needs to happen.

331 00:26:34.620 ⇒ 00:26:35.480 Samuel Roberts: Definitely.

332 00:26:35.860 ⇒ 00:26:37.050 Uttam Kumaran: Because, like…

333 00:26:37.490 ⇒ 00:26:48.000 Uttam Kumaran: even this… this is not how typical… like, I just do this a lot, so I typically will just… I know that I’m just gonna give it every single thing, and then I want it to figure out what’s next.

334 00:26:48.250 ⇒ 00:26:51.050 Uttam Kumaran: Most people are gonna give even less.

335 00:26:51.850 ⇒ 00:27:00.830 Uttam Kumaran: And if it keeps on every question, if it asks, like, is there anything else, is there anything else, it’s like, no. Just, like, keep going, keep going.

336 00:27:00.830 ⇒ 00:27:03.860 Samuel Roberts: Yeah, I didn’t even seem to take that into the context at first, until you…

337 00:27:03.860 ⇒ 00:27:04.710 Uttam Kumaran: Yeah…

338 00:27:04.710 ⇒ 00:27:06.499 Samuel Roberts: Said, like, wrap it up.

339 00:27:08.170 ⇒ 00:27:15.889 Uttam Kumaran: Yeah, and then also, like, let’s move on to the next section, context. Like, the interviewee should not care about

340 00:27:16.050 ⇒ 00:27:17.689 Uttam Kumaran: the output format.

341 00:27:18.030 ⇒ 00:27:21.460 Uttam Kumaran: Right? So, I think my point would be…

342 00:27:21.690 ⇒ 00:27:26.660 Uttam Kumaran: you have to have some separation. I… you’re overfitting to the output.

343 00:27:27.300 ⇒ 00:27:32.230 Uttam Kumaran: Like, and it’s going very systematically versus it just being a little bit more dynamic.

344 00:27:34.160 ⇒ 00:27:37.770 Uttam Kumaran: like… Let’s move on to the next section, context.

345 00:27:38.480 ⇒ 00:27:40.980 Uttam Kumaran: Alright, well, why? Like, you already have…

346 00:27:42.050 ⇒ 00:27:44.140 Uttam Kumaran: You have a lot of context, so…

347 00:27:44.840 ⇒ 00:27:50.499 Uttam Kumaran: I don’t know, I guess it’s just, like, it’s kind of fixed. So there’s probably something to tweak there.

348 00:27:51.090 ⇒ 00:27:57.120 Uttam Kumaran: Timestamps don’t matter here at all, and totally… Remove those.

349 00:27:57.570 ⇒ 00:28:06.549 Uttam Kumaran: And then… Yeah, I guess another thing is maybe having it be a pause button.

350 00:28:07.570 ⇒ 00:28:08.220 Audio shared by Uttam Kumaran: Underst-.

351 00:28:08.220 ⇒ 00:28:12.700 Uttam Kumaran: Because I almost was like, if I stop this, what happens?

352 00:28:12.700 ⇒ 00:28:14.849 Samuel Roberts: It’s gonna start over like it did before, right?

353 00:28:15.950 ⇒ 00:28:18.440 Audio shared by Uttam Kumaran: Hello, and welcome. I’m glad to have…

354 00:28:18.640 ⇒ 00:28:19.090 Uttam Kumaran: Yeah.

355 00:28:19.090 ⇒ 00:28:20.079 Samuel Roberts: use of X.

356 00:28:20.080 ⇒ 00:28:24.050 Uttam Kumaran: So, probably should just be pause, and then you already have restart.

357 00:28:24.230 ⇒ 00:28:26.480 Uttam Kumaran: So that would be great.

358 00:28:35.300 ⇒ 00:28:44.589 Uttam Kumaran: This makes sense, yeah. Okay. I mean, so I still feel like it’s close.

359 00:28:44.910 ⇒ 00:28:45.600 Samuel Roberts: Yeah.

360 00:28:45.600 ⇒ 00:28:46.790 Uttam Kumaran: Still a bit to do.

361 00:28:52.470 ⇒ 00:28:53.900 Uttam Kumaran: What does everyone else think?

362 00:28:54.080 ⇒ 00:28:59.560 Samuel Roberts: Yeah, no, I mean, I think we’ve seen a few things here. A couple things we knew about, a couple things we didn’t know about.

363 00:28:59.720 ⇒ 00:29:05.960 Samuel Roberts: But there’s definitely a little bit to improve, especially with the speech-to-speech, the speed,

364 00:29:06.210 ⇒ 00:29:10.790 Samuel Roberts: pausing and not restarting. I think there’s probably something we need to figure out with why it’s not…

365 00:29:11.570 ⇒ 00:29:15.059 Samuel Roberts: Getting the context of what you just said and processing that.

366 00:29:15.760 ⇒ 00:29:16.930 Samuel Roberts: Until you…

367 00:29:17.380 ⇒ 00:29:21.989 Samuel Roberts: But I don’t know if that’s just the way it’s working, or if we need to build something a little bigger around it.

368 00:29:22.250 ⇒ 00:29:29.580 Samuel Roberts: It doesn’t seem to just follow the script until it’s… Prompted,

369 00:29:29.780 ⇒ 00:29:34.270 Samuel Roberts: But yeah, you gave it a lot there that it should have been able to… to use.

370 00:29:34.990 ⇒ 00:29:35.990 Uttam Kumaran: Yeah.

371 00:29:36.620 ⇒ 00:29:38.370 Samuel Roberts: And it didn’t seem to at first.

372 00:29:44.780 ⇒ 00:29:45.110 Uttam Kumaran: Okay.

373 00:29:45.110 ⇒ 00:29:47.439 Samuel Roberts: I think it was… We’re cut out for us today.

374 00:29:48.130 ⇒ 00:29:53.619 Uttam Kumaran: Cool. Alright, then if nothing else, I’m gonna probably…

375 00:29:54.190 ⇒ 00:29:57.890 Uttam Kumaran: prepare for the next stand-up.

376 00:29:58.300 ⇒ 00:30:00.519 Uttam Kumaran: Anything else we want to cover?

377 00:30:06.120 ⇒ 00:30:12.100 Uttam Kumaran: I think my only other piece is for, next week,

378 00:30:13.510 ⇒ 00:30:18.769 Uttam Kumaran: Gabe, one… I kinda wanna… I have, like, a couple of ideas, so just let me know when’s…

379 00:30:19.020 ⇒ 00:30:22.019 Uttam Kumaran: best, and I can sort of translate that,

380 00:30:22.280 ⇒ 00:30:25.479 Uttam Kumaran: to you, or brainstorm some of those with you. Yeah.

381 00:30:27.190 ⇒ 00:30:28.560 Uttam Kumaran: Yeah, maybe even…

382 00:30:28.650 ⇒ 00:30:31.630 Gabriel Lam: Tomorrow may be best. Okay.

383 00:30:33.720 ⇒ 00:30:35.749 Gabriel Lam: Yeah, I think that’d be a good idea, I think…

384 00:30:35.890 ⇒ 00:30:42.950 Gabriel Lam: Maybe the team here can spend some time today to really refine and polish the prompt.

385 00:30:46.980 ⇒ 00:30:52.920 Gabriel Lam: But if you have time, I can also do it later today, Utam. And then we can put it on the…

386 00:30:53.480 ⇒ 00:30:55.290 Gabriel Lam: Put it to the side until…

387 00:30:55.440 ⇒ 00:31:00.040 Gabriel Lam: you know, I can keep thinking about it, or we can have two follow-ups.

388 00:31:00.040 ⇒ 00:31:05.210 Uttam Kumaran: Yeah. This afternoon, potentially, but I’m just kind of, like…

389 00:31:05.740 ⇒ 00:31:07.429 Uttam Kumaran: I’m just kind of slammed until then.

390 00:31:07.430 ⇒ 00:31:09.140 Gabriel Lam: Okay, no problem, yeah.

391 00:31:10.460 ⇒ 00:31:11.160 Uttam Kumaran: Okay.

392 00:31:11.790 ⇒ 00:31:13.859 Uttam Kumaran: Alright, thank you guys.

393 00:31:13.860 ⇒ 00:31:14.380 Gabriel Lam: Yep.

394 00:31:14.380 ⇒ 00:31:16.249 Samuel Roberts: Alrighty. Thank you.

395 00:31:16.250 ⇒ 00:31:19.390 Gabriel Lam: Maybe we can stay on just a little bit.

396 00:31:19.390 ⇒ 00:31:22.640 Uttam Kumaran: Sure, I’ll make you… I’ll make you, I’ll escape.

397 00:31:22.640 ⇒ 00:31:23.280 Gabriel Lam: Awesome.

398 00:31:27.210 ⇒ 00:31:33.570 Samuel Roberts: Well, yeah, let’s just, like, summarize a little bit, and… Plan out who’s texting what.

399 00:31:35.260 ⇒ 00:31:39.210 Samuel Roberts: so, we already knew the…

400 00:31:39.790 ⇒ 00:31:41.889 Samuel Roberts: The pitch we gotta figure out.

401 00:31:43.280 ⇒ 00:31:50.760 Gabriel Lam: Yeah. I think there’s a couple, super-based things, mainly to do with inactive projects, the extra emails…

402 00:31:51.070 ⇒ 00:31:55.079 Gabriel Lam: Yes. Yeah, well that, yeah, that was just the one, but I already took care of that, I think. Okay.

403 00:31:55.470 ⇒ 00:31:59.910 Gabriel Lam: then… I don’t know if we want to do Slack notifications.

404 00:32:00.480 ⇒ 00:32:03.929 Gabriel Lam: Automatically? Is that something you want to do for this week?

405 00:32:05.410 ⇒ 00:32:07.600 Samuel Roberts: And then… Oh, yeah, good question.

406 00:32:07.910 ⇒ 00:32:16.600 Gabriel Lam: For the front end, there’s the start interview button on the case study, there’s the pause interview button, and…

407 00:32:18.470 ⇒ 00:32:20.020 Gabriel Lam: the logic there?

408 00:32:20.390 ⇒ 00:32:25.980 Gabriel Lam: For the voice chat itself, there’s… the pitch.

409 00:32:26.130 ⇒ 00:32:33.150 Gabriel Lam: And the velocity… this pitch and speed of the words, there’s the pausing the voice when people are talking, and…

410 00:32:33.860 ⇒ 00:32:37.989 Samuel Roberts: I think there’s something to do with, like, the streaming the text versus streaming the audio, and… Yeah.

411 00:32:37.990 ⇒ 00:32:39.500 Gabriel Lam: Cut it, and when we stop it.

412 00:32:40.490 ⇒ 00:32:41.930 Gabriel Lam: And then for…

413 00:32:41.930 ⇒ 00:32:42.640 Samuel Roberts: Definite.

414 00:32:42.950 ⇒ 00:32:49.600 Gabriel Lam: Front-end, I think, I’m personally okay with the timestamps, like, I get why they’re there, it helps us

415 00:32:50.410 ⇒ 00:32:53.939 Gabriel Lam: I think it’s helpful for the transcript as well. I don’t know…

416 00:32:55.120 ⇒ 00:33:02.369 Gabriel Lam: That’s just me. And then I think on the back end, there’s a prompt tweaking about the output formula. I think Utam speaks

417 00:33:03.050 ⇒ 00:33:12.579 Gabriel Lam: differently to maybe how Hannah would do it, where maybe Hannah is more systematic when she talks with you guys, you can give me feedback as well, instead of, you know.

418 00:33:13.020 ⇒ 00:33:15.920 Gabriel Lam: major… block a text.

419 00:33:16.140 ⇒ 00:33:18.760 Samuel Roberts: Yeah, but we gotta be able to handle both of those.

420 00:33:18.760 ⇒ 00:33:19.560 Gabriel Lam: Yeah.

421 00:33:19.560 ⇒ 00:33:32.469 Samuel Roberts: the real thing, so I’m… yeah, I think there’s probably some prompt stuff there. I’m wondering, Mustafa, what is the… did the prompts follow the structure, or did it follow the questions, or are the questions just following the structure?

422 00:33:32.650 ⇒ 00:33:37.110 Samuel Roberts: for, like… the… current process.

423 00:33:43.010 ⇒ 00:33:56.890 Mustafa Raja: Sorry, I was mute. So the prompt mostly, has the, Notion doc that, that Hannah uses for, the interviews, so how she has structured the interviews.

424 00:33:56.890 ⇒ 00:34:07.069 Mustafa Raja: So… so yeah, it has a systematic flow of how it should go, and what questions in each section it has to ask.

425 00:34:07.880 ⇒ 00:34:11.700 Samuel Roberts: Okay, so we might just want to pull out maybe some sections and make it a little more…

426 00:34:12.110 ⇒ 00:34:17.699 Samuel Roberts: Like, the questions can still be specific, but the flow is more general and less, like…

427 00:34:18.230 ⇒ 00:34:21.069 Samuel Roberts: Moving on to the next section kind of thing.

428 00:34:21.070 ⇒ 00:34:30.209 Gabriel Lam: Yeah, or maybe if there’s, like, a way to flag, like, hey, this has already been answered. If it’s already been answered, then skip the question.

429 00:34:30.289 ⇒ 00:34:33.559 Samuel Roberts: Yeah, that’s what I’m wondering about, because it didn’t seem like it…

430 00:34:33.959 ⇒ 00:34:38.539 Samuel Roberts: listened. It seemed like it just, like, took the text and then ran with it, and kept going.

431 00:34:38.739 ⇒ 00:34:42.059 Samuel Roberts: And then it did understand it later on, so it definitely is…

432 00:34:42.189 ⇒ 00:34:44.889 Samuel Roberts: Getting in there, but it doesn’t seem to…

433 00:34:46.100 ⇒ 00:34:47.360 Mustafa Raja: Hmm.

434 00:34:47.360 ⇒ 00:34:48.359 Samuel Roberts: So maybe we need to…

435 00:34:48.360 ⇒ 00:34:48.920 Mustafa Raja: Yum.

436 00:34:48.929 ⇒ 00:34:49.949 Samuel Roberts: Yeah, go ahead.

437 00:34:50.170 ⇒ 00:34:51.950 Mustafa Raja: Yeah, we need to do something, yeah.

438 00:34:53.690 ⇒ 00:34:55.310 Samuel Roberts: Okay, yeah, got the interviews.

439 00:34:55.310 ⇒ 00:34:56.210 Mustafa Raja: There’s a sections.

440 00:34:56.219 ⇒ 00:34:57.749 Samuel Roberts: Yeah.

441 00:35:00.409 ⇒ 00:35:01.399 Samuel Roberts: Okay.

442 00:35:03.599 ⇒ 00:35:06.069 Samuel Roberts: Yeah, we might just want to break those up.

443 00:35:07.119 ⇒ 00:35:08.919 Samuel Roberts: Or not break them up, but, like…

444 00:35:11.839 ⇒ 00:35:17.359 Samuel Roberts: Maybe it is just even more instructions about, like, don’t worry about explaining the sections or something.

445 00:35:17.769 ⇒ 00:35:19.979 Samuel Roberts: We’ll have to test that out a little bit and see.

446 00:35:19.980 ⇒ 00:35:20.680 Mustafa Raja: Yeah.

447 00:35:20.680 ⇒ 00:35:23.170 Samuel Roberts: But this matches the Notion, Doc, is that…

448 00:35:24.200 ⇒ 00:35:28.349 Mustafa Raja: Yeah, it’s not exactly the Notion Dog, so.

449 00:35:28.350 ⇒ 00:35:28.730 Samuel Roberts: Sure.

450 00:35:28.730 ⇒ 00:35:31.070 Mustafa Raja: GenerationDoc to generate this.

451 00:35:31.700 ⇒ 00:35:32.140 Samuel Roberts: Okay.

452 00:35:32.140 ⇒ 00:35:34.789 Mustafa Raja: So, if I go to the notion…

453 00:35:39.180 ⇒ 00:35:40.809 Gabriel Lam: Something I’m thinking of is, like.

454 00:35:40.980 ⇒ 00:35:48.009 Gabriel Lam: I know Utam is like, hey, it should capture both. I think one thing we could also do is teach the interviewee how to interview.

455 00:35:49.820 ⇒ 00:35:51.100 Gabriel Lam: I don’t know if that makes sense.

456 00:35:51.100 ⇒ 00:35:53.260 Samuel Roberts: Mmm, yeah, I know, I see what you’re saying.

457 00:35:54.210 ⇒ 00:35:56.639 Mustafa Raja: Yeah, this tip. This moves…

458 00:35:57.350 ⇒ 00:35:58.060 Samuel Roberts: Okay.

459 00:35:58.700 ⇒ 00:36:02.940 Mustafa Raja: Yeah, so this mostly has these sections, yeah, it does have sections, but…

460 00:36:03.580 ⇒ 00:36:06.679 Samuel Roberts: Maybe we need to go back and watch an interview.

461 00:36:06.820 ⇒ 00:36:10.109 Samuel Roberts: And see how she did it, and try to match that more.

462 00:36:10.700 ⇒ 00:36:25.210 Mustafa Raja: Yeah, so she follows this flow, but, if, if I were to, explain the challenge in the start, she just say, okay, so we did talk about the challenge in this.

463 00:36:25.210 ⇒ 00:36:26.550 Samuel Roberts: Yeah, you’re right, okay.

464 00:36:26.550 ⇒ 00:36:32.870 Mustafa Raja: has any questions other than what was talked about, she’ll just ask that and move ahead.

465 00:36:33.020 ⇒ 00:36:37.700 Samuel Roberts: Okay, yeah, I think we can tweak the prompt to just… Be a little more,

466 00:36:38.320 ⇒ 00:36:40.720 Samuel Roberts: Inclusive of previous answers, maybe?

467 00:36:40.720 ⇒ 00:36:42.450 Mustafa Raja: Hmm. Yeah, yeah.

468 00:36:42.800 ⇒ 00:36:46.110 Samuel Roberts: Okay. Yeah, we can play with that. Alright,

469 00:36:48.030 ⇒ 00:36:57.340 Samuel Roberts: Yeah, I mean, there’s definitely what Gabe was saying about, like, teaching the interviewee, or guiding the interviewee, like, how much to answer at once.

470 00:36:57.480 ⇒ 00:37:00.410 Samuel Roberts: But I think we also want to try to…

471 00:37:00.510 ⇒ 00:37:02.389 Samuel Roberts: But we can definitely have some things.

472 00:37:02.630 ⇒ 00:37:03.760 Samuel Roberts: There.

473 00:37:04.120 ⇒ 00:37:08.569 Samuel Roberts: like, guidance or something. It’s just, like, answer the questions and, you know, don’t…

474 00:37:08.940 ⇒ 00:37:13.539 Samuel Roberts: I don’t know. But, like, getting all that information in one go shouldn’t be a problem.

475 00:37:14.890 ⇒ 00:37:26.000 Samuel Roberts: I think maybe we just need to tweak the, like, maybe guide the interviewee through the six sections might not be the right phrasing there. You know, maybe explaining that we’re trying to get… collect all this information in whatever

476 00:37:26.120 ⇒ 00:37:29.930 Samuel Roberts: Way we can go about it, make sure all these questions get answered eventually.

477 00:37:34.510 ⇒ 00:37:37.049 Samuel Roberts: But I think that’s just… that’s just gonna be, you know…

478 00:37:37.680 ⇒ 00:37:39.789 Samuel Roberts: Tweaking and testing more than anything.

479 00:37:41.210 ⇒ 00:37:42.610 Gabriel Lam: Casey just asked a question.

480 00:37:43.300 ⇒ 00:37:44.340 Samuel Roberts: Oh, sorry.

481 00:37:45.600 ⇒ 00:37:47.000 Casie Aviles: No…

482 00:37:47.000 ⇒ 00:37:48.469 Samuel Roberts: Who is in the mirror now?

483 00:37:49.110 ⇒ 00:37:50.400 Mustafa Raja: Mmm… nope.

484 00:37:51.460 ⇒ 00:37:52.439 Casie Aviles: Oh, shit.

485 00:37:52.740 ⇒ 00:37:55.359 Casie Aviles: Yeah, that’s also something nice. Hello?

486 00:37:55.360 ⇒ 00:37:58.980 Mustafa Raja: We can… we can get that from, the auth.

487 00:38:01.100 ⇒ 00:38:02.989 Samuel Roberts: Okay, yeah, we can pass that name in.

488 00:38:03.540 ⇒ 00:38:06.900 Mustafa Raja: Yeah, you can get that, get that from auth itself.

489 00:38:08.670 ⇒ 00:38:09.550 Mustafa Raja: Yeah.

490 00:38:10.610 ⇒ 00:38:12.579 Samuel Roberts: Is that… what do we think the…

491 00:38:12.760 ⇒ 00:38:16.639 Samuel Roberts: AI can do with that? Just be friendlier, or is there actually.

492 00:38:16.640 ⇒ 00:38:17.080 Mustafa Raja: Awesome.

493 00:38:17.080 ⇒ 00:38:20.160 Samuel Roberts: That’s helpful there. I mean, Ole might be eventually, but…

494 00:38:20.360 ⇒ 00:38:24.079 Mustafa Raja: So, when I… when it asks… asks about the team.

495 00:38:24.660 ⇒ 00:38:28.909 Mustafa Raja: I just say, I was this… I was handling this, so…

496 00:38:28.910 ⇒ 00:38:29.450 Samuel Roberts: It would…

497 00:38:29.450 ⇒ 00:38:31.500 Mustafa Raja: know who is I, you know?

498 00:38:31.660 ⇒ 00:38:33.790 Samuel Roberts: Yep, no, that makes sense. Good call, good call.

499 00:38:35.310 ⇒ 00:38:36.510 Samuel Roberts: That makes sense, yeah.

500 00:38:37.770 ⇒ 00:38:43.050 Samuel Roberts: Okay, yeah, I guess…

501 00:38:43.930 ⇒ 00:38:47.520 Samuel Roberts: We should figure out how we’re gonna split this up today, then.

502 00:38:48.860 ⇒ 00:38:51.700 Samuel Roberts: For all those pieces Gabe was outlining earlier.

503 00:38:51.930 ⇒ 00:38:55.019 Samuel Roberts: Was there anything else? Because we got a little sidetracked here, but…

504 00:38:57.080 ⇒ 00:38:59.990 Mustafa Raja: The interruption isn’t working, so that…

505 00:38:59.990 ⇒ 00:39:01.969 Samuel Roberts: Okay, yeah, I was gonna say, that was…

506 00:39:01.970 ⇒ 00:39:05.860 Mustafa Raja: But that is actually working on the… on the text side, but…

507 00:39:05.860 ⇒ 00:39:06.350 Samuel Roberts: It does, yeah.

508 00:39:06.350 ⇒ 00:39:08.200 Mustafa Raja: Or… for the voice.

509 00:39:08.390 ⇒ 00:39:09.780 Samuel Roberts: Yeah. Okay, so there’s something.

510 00:39:09.780 ⇒ 00:39:19.090 Mustafa Raja: So, I think, I think that is, that is somewhat easily handleable, I believe. I can just skip the…

511 00:39:19.210 ⇒ 00:39:21.400 Mustafa Raja: what’s it called? The chunks?

512 00:39:21.540 ⇒ 00:39:28.869 Mustafa Raja: If the user speaks, we’ll just move to whatever after the user is speaking, coming towards us.

513 00:39:29.000 ⇒ 00:39:29.690 Samuel Roberts: Right, right, right.

514 00:39:29.690 ⇒ 00:39:31.820 Mustafa Raja: So the audio comes in chunks, right?

515 00:39:32.050 ⇒ 00:39:32.630 Samuel Roberts: Yep.

516 00:39:33.620 ⇒ 00:39:35.410 Mustafa Raja: Because it’s real time.

517 00:39:36.170 ⇒ 00:39:45.260 Mustafa Raja: So, that should be fairly easy to do. I’m also wondering, what if I ask the AI to talk to me in 2x?

518 00:39:45.430 ⇒ 00:39:50.200 Mustafa Raja: what will happen, rather than, you know, programmatically trying to set the…

519 00:39:50.200 ⇒ 00:39:52.650 Samuel Roberts: It’d be worth trying, I don’t know how it handles that, but…

520 00:39:52.650 ⇒ 00:39:59.139 Mustafa Raja: Yeah, I feel that that might… that might work. And then we might not even need to, you know.

521 00:39:59.270 ⇒ 00:40:01.260 Samuel Roberts: Okay, yeah, give that a try.

522 00:40:01.600 ⇒ 00:40:07.110 Samuel Roberts: Sooner rather than later, and then if we need to, we can figure out the, the pitch…

523 00:40:08.650 ⇒ 00:40:10.609 Samuel Roberts: Like, algorithm or whatever.

524 00:40:12.090 ⇒ 00:40:13.820 Samuel Roberts: Excuse me. Okay.

525 00:40:14.000 ⇒ 00:40:14.870 Samuel Roberts: Oh, what?

526 00:40:15.950 ⇒ 00:40:24.770 Mustafa Raja: If it… if it just takes care of it, if we ask it to talk to us in 2X, then I think we should just remove.

527 00:40:26.040 ⇒ 00:40:26.529 Casie Aviles: I think.

528 00:40:26.530 ⇒ 00:40:29.579 Mustafa Raja: Our logic and let OpenAI handle that.

529 00:40:30.680 ⇒ 00:40:33.770 Casie Aviles: Yeah, I think I tried… sorry, can you hear me?

530 00:40:33.770 ⇒ 00:40:34.460 Gabriel Lam: Yes.

531 00:40:34.460 ⇒ 00:40:35.330 Mustafa Raja: Yeah.

532 00:40:35.760 ⇒ 00:40:45.500 Casie Aviles: Oh, okay, yeah. I tried asking it to speak in 2X, I think, yesterday. Looks like it wasn’t able to do it, but you can try again, maybe that’s just me.

533 00:40:51.800 ⇒ 00:40:56.079 Samuel Roberts: Yeah, I don’t know… I bet they’re handling it some other way, if that’s the case.

534 00:41:01.600 ⇒ 00:41:02.890 Samuel Roberts: Okay.

535 00:41:29.430 ⇒ 00:41:30.770 Mustafa Raja: Look into this.

536 00:41:32.610 ⇒ 00:41:33.190 Samuel Roberts: Okay.

537 00:41:39.530 ⇒ 00:41:44.360 Samuel Roberts: Alright, so we have our list of tasks, we divvy that up on Slack then, or do you want to do it now?

538 00:41:45.460 ⇒ 00:41:47.109 Samuel Roberts: Timely at 10 minutes.

539 00:41:48.140 ⇒ 00:41:49.640 Casie Aviles: I don’t mind doing it now.

540 00:41:49.640 ⇒ 00:41:50.200 Gabriel Lam: Yeah.

541 00:41:51.390 ⇒ 00:41:52.160 Samuel Roberts: Say that again?

542 00:41:54.450 ⇒ 00:41:57.339 Casie Aviles: Oh, I mean, I’m saying we can do it now, it’s fine.

543 00:41:57.340 ⇒ 00:42:01.679 Samuel Roberts: Okay, okay, cool. Yeah, we got a few minutes till, I think everyone else will hop on.

544 00:42:02.050 ⇒ 00:42:03.320 Samuel Roberts: Alright, so…

545 00:42:03.770 ⇒ 00:42:11.710 Samuel Roberts: what do we have here? Rico posted a bunch. Is that all of them? Did we cover them? Did he get them all there, Abe?

546 00:42:12.060 ⇒ 00:42:22.350 Gabriel Lam: Ai pause, playback speak, hyperlink to muting link, AI not… I, I can… Hmm.

547 00:42:23.550 ⇒ 00:42:27.900 Gabriel Lam: Yeah, they’re pretty… I can refine it, because I have a bunch of notes there as well.

548 00:42:27.900 ⇒ 00:42:31.439 Samuel Roberts: Okay, yeah, so I want to make sure we’re not missing anything. Okay, so let’s say,

549 00:42:32.910 ⇒ 00:42:39.149 Samuel Roberts: I would say, Mustafa, you can handle the, like, voice and the 2X thing, since you’re most into that right now.

550 00:42:39.490 ⇒ 00:42:43.640 Samuel Roberts: Or…

551 00:42:43.940 ⇒ 00:42:50.230 Samuel Roberts: the refine and Polish case study is… okay, what? Basically the… the prompt stuff, we might wanna…

552 00:42:52.340 ⇒ 00:42:54.769 Samuel Roberts: Test separately from doing that, but…

553 00:42:54.990 ⇒ 00:42:59.930 Samuel Roberts: I can probably take some of the AI… the, front-end stuff and just knock that out real quick.

554 00:43:00.280 ⇒ 00:43:05.140 Samuel Roberts: Start interview button, pause interview…

555 00:43:06.440 ⇒ 00:43:12.390 Samuel Roberts: Filters are actually there, they’re just not obvious, I think, is part of the problem, but… I don’t know.

556 00:43:14.000 ⇒ 00:43:16.290 Mustafa Raja: What is this filter, actually?

557 00:43:16.450 ⇒ 00:43:20.390 Samuel Roberts: On the, table page, the main case study page.

558 00:43:21.240 ⇒ 00:43:22.110 Mustafa Raja: Okay.

559 00:43:22.110 ⇒ 00:43:26.789 Samuel Roberts: If you click, the little three dots next to the name.

560 00:43:26.790 ⇒ 00:43:29.010 Gabriel Lam: I see it.

561 00:43:29.010 ⇒ 00:43:37.839 Samuel Roberts: Yeah, so I… I realized I didn’t make use of that before, but I found that Material UI did have one, it just wasn’t, like, the basic Material UI stuff.

562 00:43:37.950 ⇒ 00:43:44.149 Samuel Roberts: So I added that, and it… you can filter by, like, text.

563 00:43:44.490 ⇒ 00:43:48.600 Samuel Roberts: For, like, the project name or the description, and then for client status.

564 00:43:48.600 ⇒ 00:43:49.220 Mustafa Raja: Hmm.

565 00:43:49.220 ⇒ 00:43:56.699 Samuel Roberts: interviewees… or interviewees you can’t filter by right now, but I can make that. The other ones, it’s filtered by whatever is available.

566 00:43:58.030 ⇒ 00:43:59.410 Samuel Roberts: So…

567 00:44:01.610 ⇒ 00:44:07.640 Samuel Roberts: But it’s, yeah, it’s there. I can maybe make a slightly different UI with it, it’s just this was already baked into the tool I was using there, so…

568 00:44:09.440 ⇒ 00:44:10.200 Gabriel Lam: Okay.

569 00:44:10.550 ⇒ 00:44:13.830 Gabriel Lam: I mean, if it’s in, I think we can fix it later.

570 00:44:13.830 ⇒ 00:44:21.399 Samuel Roberts: Exactly, exactly. It’s there, the functionality is there if it needs to move to something. Like, we can maybe add buttons that, like, auto-filter stuff, or whatever.

571 00:44:21.520 ⇒ 00:44:23.210 Samuel Roberts: Okay.

572 00:44:23.460 ⇒ 00:44:27.400 Samuel Roberts: Yeah, so I will add, let’s see, I will add the…

573 00:44:29.590 ⇒ 00:44:33.910 Samuel Roberts: The links, so that it doesn’t matter who’s doing it, they’re still seeing it all.

574 00:44:34.180 ⇒ 00:44:36.909 Samuel Roberts: For, starting the interview.

575 00:44:37.270 ⇒ 00:44:43.779 Samuel Roberts: More of the other UI stuff, because I don’t really…

576 00:44:44.670 ⇒ 00:44:46.300 Gabriel Lam: I… I think I can…

577 00:44:46.300 ⇒ 00:44:46.800 Samuel Roberts: Okay.

578 00:44:46.800 ⇒ 00:44:47.750 Gabriel Lam: Shit out a little more.

579 00:44:47.750 ⇒ 00:44:50.819 Samuel Roberts: Okay, yeah, let’s do that, because I think people are hopping on, so…

580 00:44:50.820 ⇒ 00:44:51.480 Gabriel Lam: Alright.

581 00:44:51.840 ⇒ 00:44:52.450 Samuel Roberts: Okay.

582 00:44:55.010 ⇒ 00:44:56.970 Samuel Roberts: Alright, sounds good. Thank you, Gabe.

583 00:44:56.970 ⇒ 00:45:00.150 Gabriel Lam: And Casey just said he’s gonna work on the problem refinement.

584 00:45:00.150 ⇒ 00:45:02.149 Samuel Roberts: Perfect. Okay, yeah, that would be perfect.

585 00:45:03.400 ⇒ 00:45:04.830 Samuel Roberts: Alright, sounds good, guys.

586 00:45:06.710 ⇒ 00:45:08.759 Samuel Roberts: I’m gonna hop off and get working on that.

587 00:45:09.360 ⇒ 00:45:14.190 Gabriel Lam: Awesome, sounds good. I’m gonna wait for people to hop on, and then I will give away the hosting.

588 00:45:14.190 ⇒ 00:45:17.519 Samuel Roberts: Oliver, you’re the host, okay, of course, alright. Alrighty.

589 00:45:17.520 ⇒ 00:45:18.879 Gabriel Lam: Alright, catch you guys later.

590 00:45:18.880 ⇒ 00:45:19.750 Samuel Roberts: See you later.

591 00:45:20.580 ⇒ 00:45:21.569 Mustafa Raja: Thank you.

592 00:45:21.570 ⇒ 00:45:22.390 Gabriel Lam: Thank you.

593 00:45:22.650 ⇒ 00:45:23.190 Samuel Roberts: Yep.

594 00:46:01.760 ⇒ 00:46:03.500 Robert Tseng: Hello, everyone!

595 00:46:08.160 ⇒ 00:46:08.950 Rico Rejoso: Nice.

596 00:46:26.670 ⇒ 00:46:28.609 Zoran Selinger: We’re rocking some merch, I see.

597 00:46:29.740 ⇒ 00:46:33.329 Robert Tseng: Oh, yes. Yes, we’ll, we’ll get you one of these, too.

598 00:46:33.690 ⇒ 00:46:37.979 Robert Tseng: That’s usually what I wear it on a bad hair day.

599 00:46:41.750 ⇒ 00:46:44.669 Zoran Selinger: I don’t have those, ever.

600 00:46:47.480 ⇒ 00:47:01.720 Robert Tseng: But yeah, that might be me in a couple years. I just haven’t… my hairline keeps going back, I haven’t fully committed to the balm look yet, but it’s good. Once you’re a dad, like, I feel like that’s the move, so maybe that’ll be me in a couple years.

601 00:47:06.190 ⇒ 00:47:06.840 Uttam Kumaran: That’s funny.

602 00:47:06.840 ⇒ 00:47:09.770 Robert Tseng: Makes you… makes you look more friendly, is what I hear.

603 00:47:15.590 ⇒ 00:47:17.930 Uttam Kumaran: Cool, guys. Should we talk about,

604 00:47:18.840 ⇒ 00:47:22.309 Uttam Kumaran: Do we reverse order, you wanna talk about Eden first, as usual?

605 00:47:22.830 ⇒ 00:47:24.880 Robert Tseng: No, no, let’s reverse it, yeah.

606 00:47:25.150 ⇒ 00:47:41.699 Uttam Kumaran: Okay, I want to talk… I want to talk about, sorry, Zora, I want to talk about maybe Honey Stinger briefly. So, Henry, I think, I’ve said to you the sort of stitching piece, and let me actually ask Sam if he can join this,

607 00:47:42.420 ⇒ 00:47:57.740 Uttam Kumaran: But basically, there’s two pieces. So, one, we’re waiting on all the message body-related stuff to come back from Polytomic, so we won’t be able to do that until then. Instead, I think, Henry, I kind of tasked you with kind of probably, I think, two things. One is, like, just a rough

608 00:47:58.070 ⇒ 00:48:06.509 Uttam Kumaran: look at, klaviyo, and then second is the identity stitching. Thanks, Sam.

609 00:48:06.800 ⇒ 00:48:24.500 Uttam Kumaran: Sure. I… I think, Sam, it’d be great if you and Henry could collaborate on applying that script I sent yesterday. Basically, for another client, we developed a script that does matching, across a number of factors, between customer data sets from various

610 00:48:24.830 ⇒ 00:48:29.619 Uttam Kumaran: selling platforms, like… so, basically, the objective here is to look at

611 00:48:29.770 ⇒ 00:48:34.429 Uttam Kumaran: who… what overlap is there across Amazon, Shopify, Walmart?

612 00:48:35.120 ⇒ 00:48:48.860 Uttam Kumaran: And, yeah, so I haven’t looked at that script in a few months, so I know I kind of just, like, threw that over the hill, but I think if you guys can collaborate on that and kind of let me know what you find, ideally, like, what we’re trying to produce is

613 00:48:49.020 ⇒ 00:49:03.809 Uttam Kumaran: a data set that shows the list of users and which platform they bought from, and then they could use that for retargeting. Ideally, of course, they want to be able to move people from one platform to the

614 00:49:03.840 ⇒ 00:49:09.469 Uttam Kumaran: to the more owned channel. So ideally, it will end up some process of, like, hey.

615 00:49:09.960 ⇒ 00:49:13.550 Uttam Kumaran: ongoing, we’re gonna move clients from Amazon to…

616 00:49:13.780 ⇒ 00:49:19.190 Uttam Kumaran: another platform, for example. But I think it would be great for you guys to kind of become

617 00:49:19.390 ⇒ 00:49:24.619 Uttam Kumaran: subject matter experts on that script. I haven’t looked at it in quite a while, but yeah.

618 00:49:25.060 ⇒ 00:49:29.579 Henry Zhao: Do you remember why two of your files in there are archived? They seem pretty useful.

619 00:49:30.100 ⇒ 00:49:35.360 Henry Zhao: Were they archived because they’re not correct, or should we be looking at those? Do you remember?

620 00:49:35.850 ⇒ 00:49:37.640 Uttam Kumaran: I don’t know.

621 00:49:37.640 ⇒ 00:49:38.749 Henry Zhao: and address match.

622 00:49:39.090 ⇒ 00:49:41.690 Uttam Kumaran: I… I have no clue.

623 00:49:41.690 ⇒ 00:49:42.150 Henry Zhao: Okay.

624 00:49:42.150 ⇒ 00:49:44.929 Samuel Roberts: What script was this? I don’t know if I saw it.

625 00:49:44.930 ⇒ 00:49:48.040 Henry Zhao: Data… data science script that you…

626 00:49:48.460 ⇒ 00:49:51.700 Henry Zhao: I think it’s in the public channel, Sam, if you wanna…

627 00:49:51.880 ⇒ 00:49:52.380 Robert Tseng: Didn’t a.

628 00:49:52.380 ⇒ 00:49:52.920 Uttam Kumaran: Yeah, I don’t know.

629 00:49:52.920 ⇒ 00:49:56.979 Robert Tseng: Annie run this, so, like, he might have been the last one to have touched this.

630 00:49:57.820 ⇒ 00:49:58.240 Uttam Kumaran: Yeah, wait.

631 00:49:58.240 ⇒ 00:49:58.920 Awaish Kumar: Did you touch this?

632 00:49:58.920 ⇒ 00:49:59.660 Uttam Kumaran: last?

633 00:50:00.630 ⇒ 00:50:05.639 Awaish Kumar: I just executed these. I haven’t looked into the code. It was created by Pius.

634 00:50:06.960 ⇒ 00:50:14.819 Uttam Kumaran: I mean, of course I can call pies. I just want you guys to… you guys to do a little bit of discovery. Let me know if it’s on… let me know if there’s, like… yeah.

635 00:50:15.130 ⇒ 00:50:15.820 Uttam Kumaran: But…

636 00:50:17.750 ⇒ 00:50:22.630 Awaish Kumar: Yeah, if you need my help, like, I can go into the scripts and look how they’re working.

637 00:50:22.630 ⇒ 00:50:30.979 Henry Zhao: No, we don’t need help, I was just asking in case… we don’t have to spend too much time on this, I was just asking if you had it on the top of the head. Like, if you’re like, that’s old, it’s not right, I would just not even look at it.

638 00:50:31.300 ⇒ 00:50:32.350 Henry Zhao: But I’ll look at it.

639 00:50:32.550 ⇒ 00:50:33.540 Henry Zhao: If you don’t remember.

640 00:50:34.600 ⇒ 00:50:41.640 Uttam Kumaran: Yeah, take a look. If you guys can profile it, and then if there’s any outstanding questions, I’ll… I can connect us with the guy who wrote it, but…

641 00:50:41.640 ⇒ 00:50:42.230 Henry Zhao: Got it.

642 00:50:42.400 ⇒ 00:50:43.949 Uttam Kumaran: Take a look first.

643 00:50:43.950 ⇒ 00:50:56.170 Henry Zhao: Yeah, in the meantime, in the Klaviyo table, there’s 138… 128 different event tables, so I was just going through the list and seeing, like, which ones would I need to clean up right now in order to do my analysis.

644 00:50:56.240 ⇒ 00:51:16.159 Henry Zhao: that was where my question was coming from, is should I be looking at stuff like Unveiled and Recharged? But probably not. For now, for by tomorrow, I just want to look at 3 basic things. One is email subscriber health. So, who’s subscribing? Who’s bouncing? Who’s unsubscribing? Secondly is the, email to intent, so who…

645 00:51:16.440 ⇒ 00:51:32.280 Henry Zhao: are the ones that opened an email, and then added stuff to cart, and then purchased. And then the last is conversion and logistics. So, out of the people that added to cart, who purchased, who actually got shipped, who canceled, who got retained. So that’s what I want to get cleaned up, and hopefully have insights for by tomorrow.

646 00:51:32.770 ⇒ 00:51:33.360 Uttam Kumaran: Okay.

647 00:51:35.940 ⇒ 00:51:39.680 Uttam Kumaran: Cool, so yeah, I think for me, like, kind of checkpoint would be, like.

648 00:51:40.730 ⇒ 00:51:47.240 Uttam Kumaran: This afternoon or so, would love to just get a little bit of, like, a… what’d you guys find?

649 00:51:47.360 ⇒ 00:51:52.499 Uttam Kumaran: Our meeting is tomorrow at… At, 2?

650 00:51:52.770 ⇒ 00:52:02.549 Uttam Kumaran: Last week I was very, like, last minute, so I’d love to at least try to, like, get a version of the deck prepared tonight, so that I can present it to y’all

651 00:52:02.660 ⇒ 00:52:04.160 Uttam Kumaran: And stand up tomorrow?

652 00:52:06.080 ⇒ 00:52:12.719 Uttam Kumaran: And then, of course, like, whatever feedback there is, we can make before the meeting. So, I’ll just put a little, like, checkpoint meeting

653 00:52:13.210 ⇒ 00:52:17.910 Uttam Kumaran: Or just an async thing for this afternoon. Okay.

654 00:52:17.910 ⇒ 00:52:28.730 Henry Zhao: In an ideal world, what would you like in that deck? Would you like, this is what the data discovery we did, this is how the data is, or do you want just insights, or do you want, here’s a roadmap for insights?

655 00:52:28.730 ⇒ 00:52:30.899 Uttam Kumaran: Yeah, I think it depends on what you find.

656 00:52:31.860 ⇒ 00:52:41.959 Uttam Kumaran: And I also told Amber that if she could also help, you know, put that together, but I would say don’t worry about the deck, but worry about, like, what the story is.

657 00:52:41.960 ⇒ 00:52:45.640 Henry Zhao: So part of this is just, like, we will go and confirm.

658 00:52:45.700 ⇒ 00:52:53.440 Uttam Kumaran: hey, we looked at this, this is what we found, just confirming. And then it’s up to us to think about, okay, like, what are the stories we want to tell?

659 00:52:54.470 ⇒ 00:53:00.589 Uttam Kumaran: So… I wouldn’t worry… I would say don’t worry, the deck format is…

660 00:53:00.720 ⇒ 00:53:17.659 Uttam Kumaran: what we can figure out later. More is, like, look at the data and tell me, like, kind of, like, what you’re seeing in Slack, and then I can give you the direction. So as long as… if you just livestream me the stuff in Slack today, I’ll give you the answers, or, like, kind of, like, what maybe… how to shape it. And then the deck meet… we can all work on together, probably later today.

661 00:53:18.050 ⇒ 00:53:23.970 Henry Zhao: So I’ll livestream it, and you can just add, kind of, like, your context of, like, what the client would care about, and what they would think about.

662 00:53:23.970 ⇒ 00:53:24.570 Uttam Kumaran: Perfect.

663 00:53:24.570 ⇒ 00:53:27.410 Henry Zhao: That we’re trying to tell, and then we can frame the story that way.

664 00:53:27.740 ⇒ 00:53:31.319 Uttam Kumaran: Yeah, the other thing is, like, I can give you access to their Klaviyo if you want.

665 00:53:32.280 ⇒ 00:53:33.690 Uttam Kumaran: I don’t think it is. Like, AI.

666 00:53:34.240 ⇒ 00:53:37.040 Henry Zhao: I think from Mother Duck, I can already tell things.

667 00:53:37.490 ⇒ 00:53:41.390 Henry Zhao: the questions I’m asking today are just, like, would they care about this? Like, is this one of.

668 00:53:41.390 ⇒ 00:53:56.690 Uttam Kumaran: Okay, okay. Yeah, probably the biggest thing is, like, of course, I think they’re gonna have aggregates, like, on open rates and stuff. I think we want to start to do, like, what we did for Insomnia, which is just, like, identify the outlier campaigns, ask, like, hey, was there anything done about this? And, like.

669 00:53:57.070 ⇒ 00:54:03.410 Uttam Kumaran: I don’t mind… I don’t mind if they say, like, oh, these are all interesting, but we kind of know all these. What I don’t want is, like.

670 00:54:03.520 ⇒ 00:54:06.909 Uttam Kumaran: Well, I don’t want to put in front of those, hey, your open rate is 30%.

671 00:54:07.230 ⇒ 00:54:13.529 Uttam Kumaran: You know what I mean? So, like, even if they’re like, oh, we know all that, that’s fine, because I’ll drag out the next pieces.

672 00:54:13.540 ⇒ 00:54:29.870 Uttam Kumaran: But I would say… so that… and that just also gives us an understanding of, like, their data, so that’s just some discovery. So that, plus this, like, customer stitching. For the customer stitching side, one is we want to produce an analysis of, like, what the overlap is, if the script runs. Second is…

673 00:54:29.930 ⇒ 00:54:41.929 Uttam Kumaran: based on those results, we can kind of dive into there on, like, okay, what things are able to stitch, and, like, what factors are causing the stitching to happen. If that number is good enough, then what I want to then…

674 00:54:42.170 ⇒ 00:54:44.140 Uttam Kumaran: talked to Byron about is, hey.

675 00:54:44.260 ⇒ 00:55:00.059 Uttam Kumaran: like, which customers can we then move… can we activate in Klaviyo for a platform switch campaign, basically, or, like, whatever, you know? So, again, I’ll kind of leave you guys to poke at that and see what you find. But I think, Sam, I wanted to involve you, because I don’t know exactly

676 00:55:00.330 ⇒ 00:55:08.739 Uttam Kumaran: I don’t know whether we can run that in Mother Duck, or where we’re gonna run that, so I’ll let you… but then Henry has, I think, the business context on the use case, so…

677 00:55:10.560 ⇒ 00:55:11.160 Samuel Roberts: Sounds good.

678 00:55:11.260 ⇒ 00:55:13.140 Uttam Kumaran: Team, I’ll look into the code and see.

679 00:55:13.230 ⇒ 00:55:13.760 Samuel Roberts: what I can…

680 00:55:13.760 ⇒ 00:55:14.140 Uttam Kumaran: Okay.

681 00:55:14.140 ⇒ 00:55:16.479 Samuel Roberts: And I can parse from it and figure out where it can run.

682 00:55:16.820 ⇒ 00:55:17.440 Uttam Kumaran: Okay.

683 00:55:18.430 ⇒ 00:55:19.230 Uttam Kumaran: Cool.

684 00:55:19.590 ⇒ 00:55:24.509 Uttam Kumaran: Great, let’s talk about, insomnia. So,

685 00:55:25.680 ⇒ 00:55:31.560 Uttam Kumaran: Yeah, Amber, the insights are great. I think maybe we should, do we want to just chat through them?

686 00:55:31.840 ⇒ 00:55:35.200 Uttam Kumaran: While we’re on… the call?

687 00:55:35.970 ⇒ 00:55:39.240 Amber Lin: I can do that. I also have a one-on-one with Robert today.

688 00:55:39.240 ⇒ 00:55:40.560 Uttam Kumaran: Oh, okay, so you guys…

689 00:55:40.560 ⇒ 00:55:42.370 Amber Lin: So, talk band.

690 00:55:42.980 ⇒ 00:55:44.329 Uttam Kumaran: Yeah, Robert, what do you think?

691 00:55:44.800 ⇒ 00:55:45.730 Uttam Kumaran: Okay, yeah, then…

692 00:55:45.730 ⇒ 00:55:47.720 Robert Tseng: Yeah, we can, we can, we can fund it, yeah.

693 00:55:47.870 ⇒ 00:55:48.330 Amber Lin: books.

694 00:55:48.330 ⇒ 00:55:52.140 Uttam Kumaran: Then, yeah, then do it then. I’m getting us time with Matt on Tuesday.

695 00:55:52.270 ⇒ 00:55:53.830 Amber Lin: Awesome, okay, I’ll…

696 00:55:53.830 ⇒ 00:56:00.780 Uttam Kumaran: You guys can also, yeah, also talk about rewards and stuff like that, like, get from Robert whatever questions we have.

697 00:56:01.250 ⇒ 00:56:02.309 Uttam Kumaran: Cool.

698 00:56:02.960 ⇒ 00:56:04.930 Uttam Kumaran: Cool. On,

699 00:56:06.870 ⇒ 00:56:14.420 Uttam Kumaran: I’m… I added you to the meeting with default today, Amber, so we can just walk her through other findings, and then I’m gonna continue to get…

700 00:56:14.550 ⇒ 00:56:16.739 Uttam Kumaran: A few other pieces for scope.

701 00:56:21.030 ⇒ 00:56:31.770 Uttam Kumaran: For Eden, so I think, Robert, maybe, if you have the meeting, I can review that, or if you want to just give the lowdown on, like, priorities while…

702 00:56:32.250 ⇒ 00:56:33.599 Uttam Kumaran: like, Eden.

703 00:56:33.890 ⇒ 00:56:38.210 Uttam Kumaran: team is here, then I can work to sort of, like, Yeah.

704 00:56:38.790 ⇒ 00:56:46.300 Robert Tseng: Yeah, I mean, I think, like, what we had in the deck, decks, were fine.

705 00:56:46.630 ⇒ 00:56:49.639 Robert Tseng: I mean, I’m just gonna pull it up.

706 00:56:52.100 ⇒ 00:56:59.830 Henry Zhao: While you pull that up, Robert, Rebecca’s gonna probably ask me for an update on, like, the Pharmetica data. What should I tell her in the meantime, since…

707 00:57:00.550 ⇒ 00:57:03.699 Henry Zhao: We probably won’t be able to get that stuff from the API immediately, right?

708 00:57:04.420 ⇒ 00:57:08.809 Robert Tseng: Yeah, yeah, just tell her that, like, they haven’t… they haven’t sent over the docs, right? So…

709 00:57:08.810 ⇒ 00:57:10.900 Henry Zhao: They did send over the docs, right? They sent it over yesterday.

710 00:57:11.090 ⇒ 00:57:12.540 Robert Tseng: But don’t they want…

711 00:57:12.710 ⇒ 00:57:15.969 Henry Zhao: Don’t they want to visit? I’d like to have us have a visit before they give us the data?

712 00:57:15.970 ⇒ 00:57:34.359 Robert Tseng: No, no, I mean, we should still look into it, but they are trying to schedule… I mean, they do want a two-day on-site when… I’m just like, I mean, I wouldn’t go to that. I… I might… yeah, that’s… I’m not going to Wisconsin for two days, it just doesn’t make sense. So, yeah, like, I… we…

713 00:57:34.540 ⇒ 00:57:46.199 Robert Tseng: the pharmacy… we should be able to get some stuff out of the pharmacy API just by reviewing their docs, but, like, they do want their solutions architect to kind of, like, do a walkthrough with us. Like, I don’t know what the logistics of that, I think…

714 00:57:47.420 ⇒ 00:57:48.259 Henry Zhao: Can it not be done online?

715 00:57:48.260 ⇒ 00:57:48.810 Robert Tseng: Yeah.

716 00:57:49.450 ⇒ 00:57:53.380 Robert Tseng: I… I… I do not. I don’t… I don’t know. They… they just…

717 00:57:55.280 ⇒ 00:57:58.109 Robert Tseng: Yeah, I mean, I expect you to kind of,

718 00:57:58.310 ⇒ 00:58:03.280 Robert Tseng: I mean, you should… you should befriend Michelle, like, I just… I just like…

719 00:58:03.280 ⇒ 00:58:03.910 Henry Zhao: Okay.

720 00:58:03.910 ⇒ 00:58:21.650 Robert Tseng: on the call yesterday, I just called her after you told me, and I introduced myself, and we chatted for, like, 15 minutes, and she gave me all the stuff, so I feel like they’re pretty… I’m sure Eden’s their biggest customer, like, they’ll do… they’ll bend over backwards for you, like, you do… you can… I feel like we have more leverage than you think, like…

721 00:58:21.650 ⇒ 00:58:23.460 Henry Zhao: Okay, cool. Yeah, it’s a look.

722 00:58:23.560 ⇒ 00:58:24.300 Robert Tseng: Yeah.

723 00:58:24.620 ⇒ 00:58:25.870 Henry Zhao: Alright, sounds good, I’ll do that.

724 00:58:26.030 ⇒ 00:58:26.720 Robert Tseng: Yeah.

725 00:58:26.720 ⇒ 00:58:27.680 Henry Zhao: After Honey Stinger.

726 00:58:30.300 ⇒ 00:58:30.830 Robert Tseng: Okay.

727 00:58:30.830 ⇒ 00:58:31.340 Uttam Kumaran: Yes.

728 00:58:32.030 ⇒ 00:58:34.230 Robert Tseng: Yeah, so if I just…

729 00:58:35.590 ⇒ 00:58:38.540 Robert Tseng: Let’s do this real quick. So…

730 00:58:39.110 ⇒ 00:58:48.169 Robert Tseng: we went through… yeah, I guess, like, this roadmap stuff is… is, like, they’re… yeah, I think these are the right… these are the right eye…

731 00:58:48.360 ⇒ 00:58:50.880 Robert Tseng: These are the right items.

732 00:58:51.190 ⇒ 00:59:08.570 Robert Tseng: I do think, like, with our time, we can… we can manage this, right? So, like, this, to me, this ends up becoming, like, an analysis motion, where we’re submitting something, like, bi-weekly, and like, kind of what we were doing for, insomnia, but we have a lot more data at our fingertips.

733 00:59:08.570 ⇒ 00:59:19.900 Robert Tseng: the pharmacy ops stuff, this is kind of the pharmetica work that Henry’s already doing, so that’s that. And then this is less us-driven, this is more like giving their team the tools that they need.

734 00:59:19.900 ⇒ 00:59:31.550 Robert Tseng: bringing the Judd and Ryan into Mixpanel, making sure that they’re… that they’re enabled in… to be able to… to run CRO more effectively. So, I think they’re… they’re all good… they’re good with these as, like, the…

735 00:59:31.760 ⇒ 00:59:44.819 Robert Tseng: higher-level objectives. And then, as far as, like, opportunities, I kind of presented it more as, like, how these opportunities roll up into those objectives, right? So, Pharmetica, that’s pretty straightforward.

736 00:59:44.940 ⇒ 00:59:51.070 Robert Tseng: financing… finance and fulfillment forecasting. Tom, I think this is kind of where, like.

737 00:59:51.200 ⇒ 01:00:08.220 Robert Tseng: I mean, I don’t know if… it’s either gonna be me going back into it, or I haven’t seen anyone else on this team make any, like, real moves on the finance… on the forecasting side. So, if we’re trying to bring in financial analysts, like, I think this is kind of where they would be able to help.

738 01:00:08.350 ⇒ 01:00:20.619 Robert Tseng: And I think that this is where the AI team can really start helping with the insight generation. So, I think this is pretty low-hanging fruit. They want to be able to take their most common, like, questions, and which I kind of…

739 01:00:21.300 ⇒ 01:00:30.729 Robert Tseng: I mean, I kind of riffed, like, I don’t think we’ve actually measured the num… like, but I… anecdotally, I think that these are the most common types of requests.

740 01:00:30.820 ⇒ 01:00:40.579 Robert Tseng: you know, sales by pharmacy, like, period-over-period analysis, orders and sales, and then, you know, some of the most common segments by product and channel that we talk about.

741 01:00:40.580 ⇒ 01:00:51.850 Robert Tseng: if we just kind of set up something that allows them to self-serve and asking those questions directly, through… with, like, some AI chat feature, I think that would…

742 01:00:51.980 ⇒ 01:01:02.180 Robert Tseng: you know, hopefully reduce the volume of those types of questions that our team has to take on, and Adam’s excited about that, just because he thinks that that’s… that would be helpful.

743 01:01:02.180 ⇒ 01:01:12.639 Robert Tseng: And then I think this one is kind of the longer term. We need to kind of build out, like, a voice of customer motion, but, like, I’m calling it an AI-powered product manager just because it’s…

744 01:01:12.650 ⇒ 01:01:27.729 Robert Tseng: you know, whatever, I’m… it’s just the… it’s just the marketing… marketing term. So, I’m most excited about this work. I feel like I have a lot of ideas of how this could be run, but I also think that this is kind of the AI team helping. So, I… I kind of foresee that, like.

745 01:01:27.770 ⇒ 01:01:46.290 Robert Tseng: our Eden team shifting a bit? Like, if these are gonna be our priorities, like, you know, we’re not gonna need so much data engineering, like, all the time, and I, you know, there’s room to bring the AI team into this client. So, that’s kind of what I see coming ahead.

746 01:01:46.880 ⇒ 01:01:47.470 Uttam Kumaran: Okay.

747 01:01:47.840 ⇒ 01:01:48.400 Robert Tseng: Yeah.

748 01:01:50.690 ⇒ 01:01:56.209 Uttam Kumaran: Cool, so I think, maybe, Zoran, like, you wanna… maybe we can meet tomorrow morning.

749 01:01:56.360 ⇒ 01:02:01.069 Uttam Kumaran: Or some time to kind of, like, map out a larger roadmap, and then I can work

750 01:02:01.260 ⇒ 01:02:06.349 Uttam Kumaran: With either Amber… have Amber or Rico help us sort of, like, get that into linear.

751 01:02:06.470 ⇒ 01:02:08.209 Uttam Kumaran: And then, yeah, I think,

752 01:02:08.690 ⇒ 01:02:12.990 Uttam Kumaran: We can also put together a bit of a plan on the AI piece.

753 01:02:13.160 ⇒ 01:02:17.489 Uttam Kumaran: Whether that’s considering a new tool or implementing something ourselves.

754 01:02:19.380 ⇒ 01:02:26.010 Uttam Kumaran: And then same with the financial analysis, I think we’ll just… basically, I’m gonna create a few outlines, and then whoever has capacity

755 01:02:26.150 ⇒ 01:02:28.900 Uttam Kumaran: On the analysis side can run at those.

756 01:02:31.180 ⇒ 01:02:35.860 Zoran Selinger: Okay, we’ll, we’ll have to do it… At the latest, at 10.

757 01:02:36.590 ⇒ 01:02:37.130 Uttam Kumaran: Okay.

758 01:02:38.560 ⇒ 01:02:39.509 Uttam Kumaran: Or, I mean…

759 01:02:39.510 ⇒ 01:02:40.850 Zoran Selinger: our catch-up,

760 01:02:40.850 ⇒ 01:02:41.470 Uttam Kumaran: Okay.

761 01:02:41.470 ⇒ 01:02:52.440 Zoran Selinger: Or, way later, because I’m gonna be away, for an event. I’ll come back, kind of, 9, 10 PM. I can log in then, again.

762 01:02:52.550 ⇒ 01:02:54.560 Zoran Selinger: When I come back.

763 01:02:54.560 ⇒ 01:03:11.280 Uttam Kumaran: Yeah, I mean, also what we can do is, like, we can do stuff async or with Loom, like, even if you want to record a Loom with, like, high-level roadmap, or even using that other document that we started, that would at least help me, and I can structure that, and then we can

764 01:03:11.520 ⇒ 01:03:13.199 Uttam Kumaran: We could discuss next week.

765 01:03:13.200 ⇒ 01:03:15.520 Zoran Selinger: Okay, I’ll do that. I’ll do that, that’s fine.

766 01:03:15.520 ⇒ 01:03:16.729 Uttam Kumaran: Okay, cool, yeah.

767 01:03:17.080 ⇒ 01:03:17.760 Uttam Kumaran: Great.

768 01:03:19.810 ⇒ 01:03:25.710 Uttam Kumaran: Cool. So we talked about those. On Urban Stems, yeah, we’re just continuing to…

769 01:03:25.860 ⇒ 01:03:31.450 Uttam Kumaran: sort of push on Looker pieces. I guess, Robert, did you end up finding anything or getting into stuff?

770 01:03:32.040 ⇒ 01:03:38.950 Robert Tseng: Yeah, I’m looking at it, I mean, obviously there’s… just looking at Looker doesn’t really tell me much, that’s why I’m kind of like,

771 01:03:39.160 ⇒ 01:03:43.510 Robert Tseng: I mean, I have some general benchmarks now. I know that they do…

772 01:03:43.680 ⇒ 01:03:55.870 Robert Tseng: something between 30 to 50 million a year in sales. Like, I know what their… what the… what the seasonality looks like. I know that their current forecasting sucks. It’s off by, like, 25 to 30%, so…

773 01:03:55.930 ⇒ 01:04:05.070 Robert Tseng: I mean, yeah, one of the first directives I had in the previous company was bring forecasting to within 10%, and I don’t think it’s that hard to do it, so…

774 01:04:05.070 ⇒ 01:04:15.940 Robert Tseng: I mean, I have a lot of stuff that I could say about how we could bring the forecast accuracy, down to a, like, a less than 10% delta. But once again, I think I would need another… I mean, I would…

775 01:04:16.010 ⇒ 01:04:22.650 Robert Tseng: either I’m gonna have to do it, or I would need to pull in a financial analyst to do so. So… but I think… I think it’s clear. It’s just…

776 01:04:22.650 ⇒ 01:04:36.889 Robert Tseng: I haven’t looked at the SQL. I think I understand how they’re modeling their data for… they probably only built, like, one, like, budget-based financial model. They’re not… they don’t have any driver-based modeling, which is the easiest thing that you can do to make it,

777 01:04:36.890 ⇒ 01:04:47.930 Robert Tseng: to bring it closer. So, if they’re consistently off their forecast by 25-30%, that’s very significant. Like, I think that’s kind of where I would probably hit the hammer from what I’ve seen so far.

778 01:04:47.930 ⇒ 01:04:54.760 Robert Tseng: But, I do feel like I need to poke around in Redshift and actually run some queries to continue to build that out.

779 01:04:54.880 ⇒ 01:05:07.000 Robert Tseng: Yeah, it’ll probably take me another, like, one or two hours, so I don’t really know when we exactly have to give them an answer, so maybe just let me know when that discussion is gonna be, and I’ll try to work backwards from that.

780 01:05:07.250 ⇒ 01:05:13.390 Uttam Kumaran: Yeah, the discussion is at 2 o’clock, at 3 o’clock Eastern, on… no, no, no, on Tuesday.

781 01:05:13.530 ⇒ 01:05:15.610 Robert Tseng: Oh. Oh, okay. Well, then.

782 01:05:15.610 ⇒ 01:05:20.230 Uttam Kumaran: Yeah, I’ll be fine. But the other thing is, I can get you on the phone with the…

783 01:05:20.720 ⇒ 01:05:29.429 Uttam Kumaran: subject matter experts on forecasting and on shipping. And if you want, I can also… I can handle the shipping piece if you’re, like, just want to go deeper on forecasting.

784 01:05:29.810 ⇒ 01:05:39.370 Robert Tseng: Yeah, I mean, I haven’t poked around the shipping stuff yet. I mean, I think I just… I just need to run some… I just need to look at, kind of, how the…

785 01:05:40.370 ⇒ 01:05:48.240 Robert Tseng: SQL is written for some of those models. Like, I know what the models are now. Actually, like.

786 01:05:48.710 ⇒ 01:05:54.859 Robert Tseng: Half those dashboards that you set me in the folders, like, are not functional, like.

787 01:05:54.860 ⇒ 01:05:55.630 Uttam Kumaran: Yes.

788 01:05:55.630 ⇒ 01:06:02.369 Robert Tseng: Yeah, so… there wasn’t that much to look through, is what I’m saying. I guess, like, I do think I need to just…

789 01:06:02.680 ⇒ 01:06:05.329 Robert Tseng: look at this… I just need to actually look at the SQL.

790 01:06:06.100 ⇒ 01:06:15.300 Uttam Kumaran: Well, like, I can… I can get us and Emily, like, on the phone pretty easily, or if you want to come, we have a stand-up meeting in, like, 10 minutes.

791 01:06:15.550 ⇒ 01:06:19.079 Uttam Kumaran: And you can just grill her with, like, where to go find things.

792 01:06:19.440 ⇒ 01:06:21.590 Robert Tseng: Okay. It may just save you, like, an hour.

793 01:06:22.150 ⇒ 01:06:22.850 Robert Tseng: Okay.

794 01:06:24.280 ⇒ 01:06:39.439 Uttam Kumaran: she knows, like, we’re poking at this. I told her I’d call her this… I told her we’d chat about it this week. You can just ask her where to find these things, and then she’ll tell you, oh, you should… I don’t know, you should go talk to so-and-so. There’s two people that run… one runs forecasting, one runs this.

795 01:06:39.750 ⇒ 01:06:45.570 Uttam Kumaran: they’ll pick up our phone whenever. So I just want to make sure that we chat with them, or we get all that by the end of this week.

796 01:06:46.160 ⇒ 01:06:46.800 Robert Tseng: Okay.

797 01:06:47.030 ⇒ 01:06:48.310 Uttam Kumaran: So then, yeah.

798 01:06:48.850 ⇒ 01:06:52.259 Amber Lin: Yeah, Robert, I’ll invite you to stand up, just in case you want to come.

799 01:06:52.570 ⇒ 01:06:53.170 Robert Tseng: Sure.

800 01:06:54.270 ⇒ 01:06:58.460 Uttam Kumaran: Okay, cool. Otherwise,

801 01:07:03.990 ⇒ 01:07:08.970 Uttam Kumaran: The ABC, proposal I will get out, Amber.

802 01:07:09.220 ⇒ 01:07:16.660 Uttam Kumaran: I just keep running out of time for that. And then, yeah, so CES signed yesterday.

803 01:07:16.890 ⇒ 01:07:19.340 Uttam Kumaran: So I will…

804 01:07:19.810 ⇒ 01:07:26.410 Uttam Kumaran: I’m gonna email and kick off. I don’t want to… I think I’m gonna continue to do…

805 01:07:26.850 ⇒ 01:07:40.139 Uttam Kumaran: the slow roll, which is just, like, I’ll meet, and I’ll kind of figure it out, and then kind of see, like, what the scope is. I’m not… I know what we talked about before, but I don’t know exactly yet. I think in the spirit of, like.

806 01:07:40.700 ⇒ 01:07:44.720 Uttam Kumaran: enterprise, like, we just don’t have to move so fast.

807 01:07:44.850 ⇒ 01:07:57.130 Uttam Kumaran: So I will figure… give me a week to kind of figure out what’s the deal, and a little bit of a roadmap. That way, when I loop folks in, it’s a pretty easy handoff.

808 01:07:57.240 ⇒ 01:08:07.169 Uttam Kumaran: I think it’ll probably be… depending on the scope, I think, Sam, it’ll be either you or Zara, and I don’t know yet if there’s anything on the marketing side.

809 01:08:07.540 ⇒ 01:08:10.760 Uttam Kumaran: But I, I, I’ll… I’ll find out.

810 01:08:12.600 ⇒ 01:08:15.669 Uttam Kumaran: Cool, okay. That’s all I had.

811 01:08:15.670 ⇒ 01:08:20.129 Amber Lin: I, I have two questions, one for Remo. What did they say about Remo?

812 01:08:20.850 ⇒ 01:08:24.209 Amber Lin: Robert, when we met on Wednesday with them.

813 01:08:24.569 ⇒ 01:08:30.489 Robert Tseng: Yeah, so basically, surf and…

814 01:08:30.789 ⇒ 01:08:37.599 Robert Tseng: Cameron are gonna duke it out tomorrow. I think they’ve… yeah, so…

815 01:08:38.509 ⇒ 01:08:46.589 Robert Tseng: I think there… there’s, like, some confusion over, like, Cameron thinks the job is done, he wants to get paid now, but, like.

816 01:08:46.959 ⇒ 01:08:53.799 Robert Tseng: I don’t know, the contract’s just not very clear, so it’s not really anything for us to do at this point. It’s between the two of them, like, I…

817 01:08:53.799 ⇒ 01:08:55.489 Uttam Kumaran: I’m gonna…

818 01:08:55.489 ⇒ 01:09:00.789 Robert Tseng: help surf out a little bit, but after that, I’m like, dude, you need to be clear on, like, what your…

819 01:09:03.139 ⇒ 01:09:08.979 Robert Tseng: like, the acceptance criteria was for him to own. Like, it wasn’t… like, they shouldn’t be asking for, like.

820 01:09:09.259 ⇒ 01:09:22.349 Robert Tseng: well, is this ready to accept or not? Like, I mean, whatever. So, like, we… we can only do so much to help… help him. Like, I don’t want us to go and, and make a decision. I think they… they need to talk and, like, they’ll… they’ll figure it out.

821 01:09:22.810 ⇒ 01:09:36.349 Amber Lin: Okay, sounds good. Okay, then I’ll keep the meeting for them. I guess, Utam, question. I have a meeting for Urban Stems that’s with all the analysts. I know we’re starting to do… Oh, yeah, I just, yeah, I just…

822 01:09:36.410 ⇒ 01:09:39.359 Uttam Kumaran: Yeah, I said we can cancel it. No, we can cancel it.

823 01:09:39.359 ⇒ 01:09:40.079 Amber Lin: Okay, okay.

824 01:09:40.329 ⇒ 01:09:45.709 Amber Lin: Should I cancel all of, like, this type of event, and then just book it as needed?

825 01:09:46.889 ⇒ 01:09:51.300 Uttam Kumaran: I feel like let’s just cancel it recurring for now, and then we’ll come back to it, yeah.

826 01:09:51.300 ⇒ 01:09:54.450 Amber Lin: Great. Sounds good. Okay, that’s all my questions.

827 01:09:57.290 ⇒ 01:09:57.840 Uttam Kumaran: Okay.

828 01:09:58.450 ⇒ 01:10:03.280 Robert Tseng: Yeah, README, I guess you were poking around in Mongo.

829 01:10:03.280 ⇒ 01:10:06.660 Uttam Kumaran: Yeah, I’m trying… I’m just like… they just haven’t given it.

830 01:10:06.660 ⇒ 01:10:13.099 Robert Tseng: I can tell Alicia, like, I don’t think we’re gonna be able to give you anything today.

831 01:10:13.770 ⇒ 01:10:19.690 Uttam Kumaran: Well, like, or she… if I get access, like, I can… we can do it in the afternoon, I mean, they’re Pacific, so…

832 01:10:19.970 ⇒ 01:10:25.670 Uttam Kumaran: I’m gonna try… I’m… yeah… They gave me, like, the most baby out.

833 01:10:25.670 ⇒ 01:10:30.379 Robert Tseng: That’s the blocker, can I just, like… I can just tell her, like, hey, we didn’t actually get everything we needed.

834 01:10:30.380 ⇒ 01:10:40.450 Uttam Kumaran: It’s all… it’s all… it’s all in Slack. It’s literally in our Slack, where I’m discussing with Mark, like, hey, I didn’t… I need this extended ad, because I can only run queries in Mongo. I need to export

835 01:10:40.780 ⇒ 01:10:43.040 Uttam Kumaran: all of their projects and plans.

836 01:10:43.420 ⇒ 01:10:51.610 Uttam Kumaran: I see. To, like, a database where I can run queries in SQL, which is, like, that was their original scope anyways. They just didn’t give me the access

837 01:10:51.800 ⇒ 01:10:53.560 Uttam Kumaran: Necessary to do that.

838 01:10:54.610 ⇒ 01:10:55.420 Uttam Kumaran: Yeah.

839 01:10:56.170 ⇒ 01:11:01.220 Uttam Kumaran: And we only got ac- we only got the access on, like, Late Monday.

840 01:11:01.670 ⇒ 01:11:02.830 Uttam Kumaran: And…

841 01:11:03.570 ⇒ 01:11:10.249 Uttam Kumaran: Mark basically said he could give it to me. I think if you can ping… if you ping her and just say, hey, can you nudge this?

842 01:11:12.410 ⇒ 01:11:13.080 Robert Tseng: Okay.

843 01:11:15.110 ⇒ 01:11:19.000 Uttam Kumaran: I will try between now and then to get something.

844 01:11:19.180 ⇒ 01:11:25.729 Robert Tseng: Okay. Well, if it’s… I mean, it’s okay, I’ll push her out if we’re not ready, like, I don’t want to just…

845 01:11:25.950 ⇒ 01:11:30.919 Uttam Kumaran: Well, I don’t know, I feel like if it’s worth getting FaceTime, we should at least get FaceTime, and I can show her behind the scenes.

846 01:11:30.920 ⇒ 01:11:35.650 Robert Tseng: I mean, I just talked to her, like, a couple days ago, so I’m not, like, that worried about it, it’s fine.

847 01:11:37.200 ⇒ 01:11:39.210 Uttam Kumaran: Yeah, I mean, I would prefer…

848 01:11:39.370 ⇒ 01:11:46.709 Uttam Kumaran: I would prefer to do it tomorrow, or we could send a loom with everything tomorrow. As soon as I get access, I’ll have all that into a database for you, so…

849 01:11:46.710 ⇒ 01:11:47.850 Robert Tseng: Yeah, okay.

850 01:11:47.910 ⇒ 01:11:52.519 Uttam Kumaran: Sucks, it’s just, like, two steps away, I tried a bunch of workarounds, didn’t work.

851 01:11:52.520 ⇒ 01:11:53.609 Robert Tseng: Yeah, no worries.

852 01:11:55.850 ⇒ 01:12:03.880 Uttam Kumaran: Okay, cool. Yeah, let’s just jump… let’s just jump to the Urban Stems call, and then, Amber, I’ll just tee it up, and then I’ll have,

853 01:12:04.090 ⇒ 01:12:05.799 Uttam Kumaran: Robert worked directly with Emily on.

854 01:12:05.800 ⇒ 01:12:06.210 Amber Lin: support.

855 01:12:06.750 ⇒ 01:12:07.410 Amber Lin: Cool. Okay.

856 01:12:07.410 ⇒ 01:12:08.050 Robert Tseng: Okay.

857 01:12:08.870 ⇒ 01:12:09.290 Uttam Kumaran: Thanks, guys.

858 01:12:10.250 ⇒ 01:12:10.890 Robert Tseng: Thanks.