Meeting Title: Eden Adhoc Tasks - Proactive resolution Date: 2025-08-19 Meeting participants: Awaish Kumar, Uttam Kumaran, Demilade Agboola, Robert Tseng


WEBVTT

1 00:03:28.900 00:03:29.830 Awaish Kumar: Hello.

2 00:03:36.050 00:03:37.040 Uttam Kumaran: Hey, Wish.

3 00:03:38.360 00:03:40.220 Awaish Kumar: Alright, how are you doing?

4 00:03:40.570 00:03:41.800 Uttam Kumaran: Good, how are you?

5 00:03:42.450 00:03:43.599 Awaish Kumar: Think it as well.

6 00:03:44.670 00:03:46.040 Uttam Kumaran: How is it going?

7 00:03:48.380 00:03:51.000 Uttam Kumaran: Good, busy, as usual.

8 00:03:51.170 00:03:56.769 Uttam Kumaran: I’m gonna… sort of take over some of the work for Urban SAM, so… yeah, it’s, …

9 00:03:57.320 00:03:59.000 Uttam Kumaran: It’s good. I’m…

10 00:03:59.230 00:04:05.130 Uttam Kumaran: I’m gonna work on some metaplane stuff today, and then, working on some AI stuff later, so….

11 00:04:06.160 00:04:07.960 Awaish Kumar: Is it a concern, like, …

12 00:04:08.630 00:04:11.630 Awaish Kumar: I wouldn’t sell the data really complex, or….

13 00:04:12.710 00:04:14.340 Uttam Kumaran: It’s kind of like…

14 00:04:14.590 00:04:22.609 Uttam Kumaran: Yeah, it’s more complex data modeling, but we’re not on the hook for, like, analyst work, so it’s a little bit slower in that sense.

15 00:04:23.100 00:04:27.569 Uttam Kumaran: But, yeah, I just have some stuff to catch up on there, so….

16 00:04:28.760 00:04:29.630 Awaish Kumar: Okay.

17 00:04:30.920 00:04:32.920 Awaish Kumar: Hi. Hi, Denver.

18 00:04:33.260 00:04:34.269 Awaish Kumar: That’s not true.

19 00:04:36.370 00:04:38.580 Demilade Agboola: I wish. Hi, Tommy. How’s everyone doing?

20 00:04:39.350 00:04:40.590 Uttam Kumaran: Good, how are you?

21 00:04:41.620 00:04:43.670 Demilade Agboola: Doing alright. It’s you okay?

22 00:04:46.790 00:04:47.910 Awaish Kumar: Alright, great.

23 00:04:51.100 00:04:53.960 Awaish Kumar: Can I quickly start this?

24 00:04:54.210 00:04:57.550 Awaish Kumar: So, like, … We got a…

25 00:04:58.280 00:05:02.459 Awaish Kumar: We get, like, way too many ad hoc requests.

26 00:05:02.860 00:05:04.520 Awaish Kumar: From…

27 00:05:07.920 00:05:10.599 Awaish Kumar: hidden team. So, like, there are two…

28 00:05:10.730 00:05:18.240 Awaish Kumar: two things I want to discuss in this meeting. One is, fact… any new…

29 00:05:18.720 00:05:24.430 Awaish Kumar: Change which comes into our… existing models. For example, you…

30 00:05:24.980 00:05:30.410 Awaish Kumar: Products gets added, we get… our data looks…

31 00:05:30.720 00:05:35.269 Awaish Kumar: weird to… to eating things, because, for example, the…

32 00:05:35.730 00:05:42.279 Awaish Kumar: They launched a new campaign which does not follow the format we look for to figure out product names.

33 00:05:42.870 00:05:43.699 Awaish Kumar: So we….

34 00:05:43.700 00:05:44.480 Uttam Kumaran: Yeah.

35 00:05:44.480 00:05:46.740 Awaish Kumar: So, what happens is, is gets…

36 00:05:46.910 00:05:55.170 Awaish Kumar: Marked as uncategorized. And then, they escalate it as, like, for this product, the ad spend is lower.

37 00:05:55.300 00:06:00.069 Awaish Kumar: But the problem is that we’ve never seen that, value before.

38 00:06:00.350 00:06:02.790 Awaish Kumar: Like, in our regex, we don’t have that.

39 00:06:03.490 00:06:12.039 Demilade Agboola: Scott, can I… can I even add to this? So, the issue… the issues we have are not necessarily… I think we have two main issues. One is…

40 00:06:12.120 00:06:28.719 Demilade Agboola: we don’t always get the best data, and so we… we’re constantly working from a space where we have to be reactive. And two is, even the people who should manage the data are not always on it. So, for instance, the ad spend issues tend to be, …

41 00:06:29.640 00:06:32.719 Demilade Agboola: we have a sheet I could quickly show you, it’s where we…

42 00:06:33.070 00:06:36.919 Demilade Agboola: I have talked to Cutter and Stuart, who are responsible for marketing.

43 00:06:36.920 00:06:41.899 Awaish Kumar: These are the… these are the metrics, these are the tags we need for.

44 00:06:41.900 00:06:45.920 Demilade Agboola: the art. We agreed on it, it was the naming convention.

45 00:06:46.130 00:06:55.049 Demilade Agboola: If that naming convention is going to change, you need to alert me that the naming convention has changed. This is how we came up with this.

46 00:06:55.690 00:07:02.179 Demilade Agboola: And sometimes they change naming conventions without telling me, and then they come and say ad spend is low.

47 00:07:02.280 00:07:11.240 Demilade Agboola: like, I’m not going to be monitoring this. If you don’t let me know that the conversion has changed, I’m not looking at this. It won’t catch the ads.

48 00:07:11.330 00:07:30.970 Demilade Agboola: The one that happened with HRT recently is this is the agreed naming convention. However, they’re still pushing ads with just HRLT. So we’re looking for FHRT, but the ads themselves are going out with HRT. So it doesn’t catch those ads, it comes back as low, spend, and then there are issues.

49 00:07:30.970 00:07:41.449 Demilade Agboola: So it’s not necessarily that they’re releasing things without us, like, and we’re being… and we’re being reactive, it’s they’re releasing things without a due… any due process.

50 00:07:41.620 00:07:49.560 Demilade Agboola: It’s just, they’re just deleting things, things break, and then they’re like, oh, why, why aren’t we on it, or things are low? And that’s frustrating.

51 00:07:53.890 00:08:01.829 Uttam Kumaran: So I guess, like, my simple, you know, solution here is, like, why don’t just bucket anything uncategorized into uncategorized?

52 00:08:01.950 00:08:05.710 Uttam Kumaran: And then that way the… that way the sums stay the same.

53 00:08:06.450 00:08:09.370 Awaish Kumar: Yeah, we are doing it, right? We have…

54 00:08:10.220 00:08:20.300 Awaish Kumar: the uncategorized category, but the problem is that they escalated for products, like, oh, why are you not able to capture the spend for

55 00:08:20.390 00:08:34.380 Awaish Kumar: like, the semaglitude, then why the ROS or different other matrix are, are, like, not as expected, and that’s because we were not able to categorize it.

56 00:08:35.400 00:08:39.660 Uttam Kumaran: So there’s both new campaigns coming in and new products coming in.

57 00:08:39.870 00:08:40.570 Awaish Kumar: Yep.

58 00:08:40.730 00:08:44.500 Uttam Kumaran: Can you pull up the, SQL logic for….

59 00:08:45.510 00:08:46.530 Awaish Kumar: Yeah, sure.

60 00:08:52.530 00:08:55.269 Uttam Kumaran: I have a solution that you can try.

61 00:08:55.270 00:08:56.170 Awaish Kumar: Yeah, we….

62 00:08:56.520 00:08:57.100 Uttam Kumaran: Yeah.

63 00:08:57.340 00:08:58.110 Uttam Kumaran: Go ahead.

64 00:09:00.510 00:09:02.510 Awaish Kumar: So we have, macros here.

65 00:09:02.860 00:09:05.140 Awaish Kumar: And we have for you.

66 00:09:07.650 00:09:08.680 Awaish Kumar: Yeah.

67 00:09:10.260 00:09:14.869 Awaish Kumar: Here’s the long… macro, and we try to get

68 00:09:15.070 00:09:17.860 Awaish Kumar: The name from, like, in the order of…

69 00:09:18.330 00:09:21.450 Awaish Kumar: as their defining parameters. We first look at name.

70 00:09:21.800 00:09:24.130 Awaish Kumar: Then it’s a theme, and then a campaign name.

71 00:09:24.680 00:09:27.430 Awaish Kumar: And figure out if there’s any….

72 00:09:27.860 00:09:38.539 Demilade Agboola: So, sorry, just to cut… sorry to cut you short, I wanted to add something. So, if you scroll to the very top, you’ll see kind of what I’m talking about. So, if you scroll to the top, you’ll see that

73 00:09:38.700 00:09:39.530 Demilade Agboola: …

74 00:09:40.400 00:09:54.749 Demilade Agboola: you see that, okay, we have MedKit, right? Everything has MedKits, but we can see MedKit 1, MedKit 2, Medicit 3, MedKit 495, right? So we initially had agreed that we would do ads, MedKit 1, 2, 3, 4, 5, as the tags.

75 00:09:55.080 00:10:01.910 Demilade Agboola: And then, all of a sudden, they change it and say, no, what we’re doing now is we’re just doing MedKit ads generally.

76 00:10:02.290 00:10:22.039 Demilade Agboola: And… but then they needed to be spread across each of the medkits. They don’t want to see individual medkits 1, 2, 3, 4, 5. But now, there isn’t any ads going out with medkits 1, 2, 3, 4, 5, and then they’re starting medkits at low, but then they changed it to medkits, ads going out with just MedKit alone, without telling me anything about it.

77 00:10:22.040 00:10:22.950 Uttam Kumaran: Oh, yeah.

78 00:10:23.260 00:10:25.899 Demilade Agboola: There’s things like that that start to cause issues.

79 00:10:26.250 00:10:34.720 Uttam Kumaran: So again, so I think I’ll… I’m gonna kind of talk through, like, two parts of this. One is finding out, right? So can you scroll down to the bottom of this?

80 00:10:35.300 00:10:38.960 Uttam Kumaran: So at the end of the case, when? Where does it hit? It hits, like, something else?

81 00:10:42.800 00:10:43.390 Awaish Kumar: Nope.

82 00:10:43.880 00:10:47.850 Uttam Kumaran: So, one of the things we should do

83 00:10:49.230 00:10:55.079 Uttam Kumaran: And this would be something that gets put into their face, so… as well, either it’s, like, through Slack.

84 00:10:55.450 00:11:01.570 Uttam Kumaran: And we can think about this, is an alert that says that there is something new and uncategorized.

85 00:11:02.030 00:11:08.809 Uttam Kumaran: The best way to do this is to just have a table that runs this case when.

86 00:11:09.210 00:11:18.639 Uttam Kumaran: where uncategorized, and then says, there are now, like… and flag when there’s more than zero, right? Like…

87 00:11:18.880 00:11:32.610 Uttam Kumaran: We can do that in Metaplain if we want. So there’s so… so I guess I want to just start with that. So the first thing is to… the first way to get empathy for this problem is to really show them how often this is happening.

88 00:11:32.790 00:11:48.729 Uttam Kumaran: So, ideally, my proposal is you create an alert into the marketing channel, or wherever these folks are, which says, there is a new something in uncategorized. That way, the triage happens at that point.

89 00:11:49.170 00:11:56.050 Uttam Kumaran: So at least we’ve now solved one part of this problem, which is being reactive. We’re a little bit more…

90 00:11:56.320 00:12:01.479 Uttam Kumaran: we’re still reactive, but at least we’re, like, closer to the… to the problem. Does that make sense?

91 00:12:02.900 00:12:04.800 Awaish Kumar: Yes, yeah, thanks.

92 00:12:08.610 00:12:09.970 Uttam Kumaran: What do you think, Demolade?

93 00:12:10.110 00:12:13.709 Uttam Kumaran: I mean, you guys know the people, so I can’t comment on, like.

94 00:12:14.540 00:12:24.620 Uttam Kumaran: the culture, but I… I think I want to get at least the very basic thing we can do is, like, just get alerts faster for when something ends in here… ends up in here.

95 00:12:26.670 00:12:28.350 Uttam Kumaran: Absolutely. Plan something.

96 00:12:30.590 00:12:34.610 Demilade Agboola: Yeah, I mean, definitely is something, that we can do.

97 00:12:34.950 00:12:37.840 Demilade Agboola: … I mean, it’s kind of what we did.

98 00:12:37.840 00:12:41.240 Uttam Kumaran: You can immediately have the discussion at that point, right?

99 00:12:41.340 00:12:44.349 Uttam Kumaran: Hey, somebody came in, you tag them and say, what is this?

100 00:12:45.040 00:12:45.690 Uttam Kumaran: And then….

101 00:12:45.690 00:12:46.630 Demilade Agboola: Yeah.

102 00:12:46.630 00:12:47.180 Uttam Kumaran: What, what?

103 00:12:47.180 00:12:48.050 Demilade Agboola: And I….

104 00:12:48.050 00:12:49.100 Uttam Kumaran: Yeah, yeah.

105 00:12:49.270 00:13:04.670 Demilade Agboola: I mean, I think my issue, or where it becomes frustrating is, or where it becomes, like, a bit annoying is, I mean, we had to be set up that process for, like, orders, which is kind of how we’re able to go, hey, there are new orders coming in, we have no idea what’s going on here.

106 00:13:04.980 00:13:09.239 Demilade Agboola: … I think it’s just a pin of…

107 00:13:09.350 00:13:13.249 Demilade Agboola: if there are naming conventions, I think the reason why

108 00:13:13.550 00:13:27.399 Demilade Agboola: For the orders, I think there’s a bit more, like, clarity in that, because we understand why we’re not… why uncategorized things can come in. New products get launched, Basque isn’t giving us product sheets, so we’re kind of, like, looking out for that.

109 00:13:27.640 00:13:44.509 Demilade Agboola: But then, when we already have naming conventions, and I get at the end of the day, you want… this is the solution, but I think that’s where the… if we already have naming conventions, and they are just going about doing whatever they want to do, yeah, I guess this solution of trying to, like, flag them, categorize that we do for

110 00:13:44.650 00:13:51.020 Demilade Agboola: The products will be very useful here as well, just to let them know that, hey, you, like, restaurants have changed what’s going on here.

111 00:13:51.730 00:13:57.809 Demilade Agboola: But it’s, like, it’s just a bit, like, sometimes very frustrating, like, when there are other things to do.

112 00:13:57.930 00:14:00.430 Demilade Agboola: And then it’s things like this that kind of….

113 00:14:00.700 00:14:02.000 Uttam Kumaran: Yeah. ….

114 00:14:02.000 00:14:03.189 Demilade Agboola: Waste some time.

115 00:14:04.140 00:14:10.109 Uttam Kumaran: Well, I think, like, once we solve that, my second sort of cultural point is that

116 00:14:10.310 00:14:16.500 Uttam Kumaran: Someone has to be made aware at the top that we’re spending time with this because

117 00:14:16.760 00:14:24.089 Uttam Kumaran: We are not, like… if you guys have heard in project management of, like, racy diagram, but we’re not being informed.

118 00:14:24.540 00:14:39.449 Uttam Kumaran: Right? The I and RACI, R-A-A-C-I, we’re not being informed that these things are happening. And so, the expense of that is, we cannot get the tasks that we’re currently signed up for done, because we’re responding to these.

119 00:14:40.050 00:14:47.330 Uttam Kumaran: that needs to be escalated, right? And I would push that probably onto Robert, which is, like… or Amber, which is, like.

120 00:14:47.620 00:14:48.740 Uttam Kumaran: …

121 00:14:49.300 00:14:59.310 Uttam Kumaran: We’re currently dealing with this ad hoc issue. We’re not being informed of product name changes, or necessary changes to categories.

122 00:14:59.860 00:15:01.460 Uttam Kumaran: like, proactively.

123 00:15:01.820 00:15:07.699 Uttam Kumaran: There’s certainly a meeting, or at least a slack, where a decision’s happening on this, and we’re not being looped in.

124 00:15:07.890 00:15:09.200 Uttam Kumaran: And so…

125 00:15:10.290 00:15:15.079 Uttam Kumaran: what should we do? Should we continue, or should we not? As long as we get a decision.

126 00:15:15.330 00:15:22.450 Uttam Kumaran: Then we know, because for us, okay, if we have to keep doing this, then we have to slow down, the time has to come from somewhere.

127 00:15:22.790 00:15:24.790 Uttam Kumaran: But I don’t… I just think…

128 00:15:24.970 00:15:30.289 Uttam Kumaran: we have to make the, kind of, the problem very clear for Robert and Amber to be able to communicate.

129 00:15:33.990 00:15:50.670 Uttam Kumaran: Yeah, because this is… this is… this is really frustrating. So I think, like, short-term… I have some long-term ideas, we can talk about that. Short-term, I think those are the… probably the best two, is to, one, tighten the timeline by which we catch these, right? So, that’s a…

130 00:15:50.770 00:15:52.589 Uttam Kumaran: That’s a ticket we can work on.

131 00:15:52.870 00:15:57.109 Uttam Kumaran: The other ticket is to communicate. Hey, this is…

132 00:15:57.500 00:16:04.349 Uttam Kumaran: Like, can you guys put… put a estimation or… or something in terms of how many hours

133 00:16:04.550 00:16:11.239 Uttam Kumaran: or how long this is taking for a week to sort of manage these fixes. I mean, I see these PRs coming in all the time, so….

134 00:16:14.310 00:16:17.059 Demilade Agboola: Yeah, sure, we could try and just do an estimation on that.

135 00:16:17.270 00:16:25.470 Demilade Agboola: … But also, it just… it also… Plays into other things, like… …

136 00:16:28.880 00:16:43.510 Demilade Agboola: Yeah, I think, yeah, that’s… this is… this is the proactive thing. I wish, did you have any other thing you wanted to talk about, like, things that you wanted us to be more proactive about? Because I have something I wanted to mention, and I don’t want… I don’t want to change the topic without necessarily.

137 00:16:43.510 00:16:46.150 Awaish Kumar: Yeah, I’m gonna go ahead.

138 00:16:46.150 00:16:51.310 Uttam Kumaran: Yeah, maybe let me just say one long-term solution here that we can also explore.

139 00:16:51.700 00:16:55.310 Uttam Kumaran: One of the things that I saw in Snowflake, they released.

140 00:16:55.500 00:17:01.589 Uttam Kumaran: is the ability to use AI directly in query, like, directly on query time.

141 00:17:01.730 00:17:07.940 Uttam Kumaran: For example, you can pass in an input, pass it to a prompt, and get another column as an output.

142 00:17:08.730 00:17:14.310 Uttam Kumaran: I don’t know whether GCP has a version of that, But I feel like…

143 00:17:14.560 00:17:19.559 Uttam Kumaran: There could be some… there is, like, a… now a new opportunity to do something like this.

144 00:17:20.440 00:17:29.370 Uttam Kumaran: The other… I mean, one thing I’m… one thing we could do, and I think a ways of good output of this is if you could take this transcript and just toss it into…

145 00:17:30.080 00:17:32.349 Uttam Kumaran: Use AI, toss it into that dock.

146 00:17:32.990 00:17:38.830 Uttam Kumaran: There’s also probably some more dynamic ways that we can build these regex.

147 00:17:39.960 00:17:47.610 Uttam Kumaran: And we can continue to play with that, but I… I don’t know. I don’t think the alpha… there’s enough there. The only recent development

148 00:17:48.070 00:17:53.649 Uttam Kumaran: that I’ve seen that can solve some of this is having AI

149 00:17:54.010 00:17:58.439 Uttam Kumaran: like, basically, you would pass the input to an LLM, and the LLM would categorize.

150 00:17:58.640 00:18:01.609 Uttam Kumaran: So, which is just, like, a way better regex.

151 00:18:01.990 00:18:09.250 Uttam Kumaran: that… I don’t know whether GCP has that, Snowflake has the feature now. I thought… I actually thought about this last year, because…

152 00:18:09.490 00:18:12.930 Uttam Kumaran: For case WENs, or just sort of advanced regex.

153 00:18:13.100 00:18:16.480 Uttam Kumaran: All it is is, like, advanced, …

154 00:18:16.770 00:18:22.080 Uttam Kumaran: like, if-then trees, right? And so, that’s… we’d rather have something that’s more…

155 00:18:22.440 00:18:25.310 Uttam Kumaran: probabilistic, like, LOM, at least do it.

156 00:18:25.730 00:18:31.169 Uttam Kumaran: … But… that could be a solution. Another solution is….

157 00:18:31.170 00:18:32.199 Awaish Kumar: Oh, God.

158 00:18:32.200 00:18:33.209 Uttam Kumaran: Yeah, go ahead.

159 00:18:33.610 00:18:37.680 Awaish Kumar: Yeah, but with that, like, the liability goes on us, right? If we…

160 00:18:38.080 00:18:43.900 Awaish Kumar: If AI, like, categorized something wrong, then that’s on us.

161 00:18:44.940 00:18:47.399 Uttam Kumaran: Well, yeah, it’s all… but it’s all… it’s all on us, anyways.

162 00:18:48.710 00:19:05.399 Awaish Kumar: like, for example, now it’s, like, we can say, we have sent an alert, it’s not categorized, and they know the mapping. Like, it’s not that, like, Aiden team does not know the mapping, they know it, and they can share it, and with that, we can make it better.

163 00:19:05.540 00:19:09.070 Awaish Kumar: But with just AI, we are gen… it can…

164 00:19:10.070 00:19:13.119 Awaish Kumar: Can get confused between, like, some….

165 00:19:13.590 00:19:14.270 Uttam Kumaran: Okay.

166 00:19:15.270 00:19:24.700 Demilade Agboola: I think we could always… we could always use, like, maybe it’s for days, and then we modify it. So if, for instance, we understand that, like, there’s an order of priority.

167 00:19:24.960 00:19:38.459 Demilade Agboola: that, like, you know, for instance, with the MedKit stuff, it needs to catch everything. MedKit first, and if there are any ads for Medicit 1, 2, 3, 4, 5, then aggregate it to those ones, and then distribute. Like, they… we could use it to…

168 00:19:38.650 00:19:41.649 Demilade Agboola: Have a foundation, maybe not rely on it completely.

169 00:19:41.650 00:19:44.100 Uttam Kumaran: Because we also have some logic as well.

170 00:19:44.410 00:19:48.880 Uttam Kumaran: I mean, the other very, very, like, Rudimentary way, is the…

171 00:19:49.040 00:19:54.150 Uttam Kumaran: Do a meeting per week about new product for new campaigns.

172 00:19:54.370 00:19:59.460 Uttam Kumaran: … That way, there’s no hiding, right?

173 00:19:59.770 00:20:01.020 Uttam Kumaran: ….

174 00:20:02.870 00:20:11.550 Awaish Kumar: But I think, like, just as a starting point, if just sending alerts in client channel, and being proactive on that, will….

175 00:20:11.550 00:20:18.420 Uttam Kumaran: I just want to give you guys some… I just want to give you guys every option that I can think about. So, one is, like, setting the alerts.

176 00:20:18.570 00:20:30.960 Uttam Kumaran: Second is, like, having a meeting. Third is, like, we ask… I… and these are all, like, you can do and or. Third is escalate. Say, hey, we are spending one to two hours a week.

177 00:20:31.390 00:20:32.990 Uttam Kumaran: on this problem.

178 00:20:33.350 00:20:35.270 Uttam Kumaran: we can either, A,

179 00:20:35.560 00:20:40.010 Uttam Kumaran: get looped in, or B, we will have to have a recurring ticket that will take

180 00:20:40.200 00:20:42.760 Uttam Kumaran: X percent of demolade’s time per week.

181 00:20:43.020 00:20:45.159 Uttam Kumaran: There’s only two paths right now.

182 00:20:45.680 00:20:50.579 Uttam Kumaran: … Because the solution is to just have a recurring ticket on every sprint.

183 00:20:51.360 00:20:56.210 Uttam Kumaran: For 3 points, that goes to… regex adjustments.

184 00:20:57.470 00:21:02.460 Uttam Kumaran: Okay, right? That way, at least, we’re not getting assigned more work. We can’t take on more scope.

185 00:21:03.220 00:21:06.269 Uttam Kumaran: … So, I think we have all the…

186 00:21:06.720 00:21:11.220 Uttam Kumaran: I think we have a variety of solutions there, so let’s… we can write this up, and then…

187 00:21:12.200 00:21:13.540 Uttam Kumaran: We can decide.

188 00:21:13.710 00:21:20.039 Uttam Kumaran: We can… some are possible today, some we’ll have to explore, so go ahead, Demolade, on your next thing.

189 00:21:20.880 00:21:24.369 Demilade Agboola: So the next thing I wanted to mention is, like.

190 00:21:24.590 00:21:30.740 Demilade Agboola: I mean, obviously we know this, but I don’t necessarily know if we know the extent to which, like.

191 00:21:31.210 00:21:33.420 Demilade Agboola: how badly….

192 00:21:33.630 00:21:34.700 Uttam Kumaran: ….

193 00:21:34.700 00:21:41.040 Demilade Agboola: the data quality can… can be, right? So, for instance, we are trying to… we’re trying to do

194 00:21:41.170 00:21:44.649 Demilade Agboola: a dashboard for Josh, for FarmOps.

195 00:21:44.800 00:21:47.999 Demilade Agboola: And one of the things that they really want to see is.

196 00:21:48.330 00:21:49.560 Uttam Kumaran: the time….

197 00:21:50.030 00:21:58.869 Demilade Agboola: that is spent before it gets shipped, right? From when it gets to the pharmacy to when it gets shipped, because that’s a huge, SLA for them.

198 00:21:59.200 00:22:05.180 Demilade Agboola: But for instance, orders shipped, only…

199 00:22:05.360 00:22:11.529 Demilade Agboola: We can only see… we can only see the center pharmacy information in the order-shipped webhook.

200 00:22:11.760 00:22:14.710 Demilade Agboola: But the only way we can see

201 00:22:15.210 00:22:32.359 Demilade Agboola: the order shipped, like, the values in that webhook is when they are shipped. That means at no time, between when the orders are placed and when they eventually get shipped, we have no idea when they get sent to the pharmacy. So as things stand, there’s a huge vacuum in between.

202 00:22:32.510 00:22:35.990 Demilade Agboola: To let them know, hey, right now, you have…

203 00:22:36.150 00:22:49.919 Demilade Agboola: do… you have, like, 200 orders that are… that have been sent to the pharmacy 2 days ago, but have not yet been shipped. We don’t know that until they eventually get shipped, and then we can now see the sent to pharmacy date.

204 00:22:50.110 00:22:51.589 Demilade Agboola: That’s a huge vacuum.

205 00:22:51.730 00:22:55.369 Demilade Agboola: So I’m pushing, you know, Zach for that, but, like, Zach…

206 00:22:55.570 00:22:58.500 Demilade Agboola: Basically, you can be very slow with these things.

207 00:22:58.650 00:23:01.020 Demilade Agboola: Same thing with, like, the COGS theme.

208 00:23:01.140 00:23:03.649 Demilade Agboola: We have fields for…

209 00:23:03.860 00:23:10.029 Demilade Agboola: Things like vowel sizes for each of these different orders, those ages and vowel sizes.

210 00:23:10.130 00:23:11.220 Demilade Agboola: …

211 00:23:11.900 00:23:24.130 Demilade Agboola: And I’m like, let’s try and populate that. Zach has basically ignored that request. The only other way to start doing something like COGS, historically speaking, is we have to either create, like, sheets.

212 00:23:24.780 00:23:26.250 Demilade Agboola: From all of time.

213 00:23:26.630 00:23:27.900 Demilade Agboola: or B,

214 00:23:28.980 00:23:40.200 Demilade Agboola: I have to manually, like, create a model that starts doing the calculations, and even that is not always going to be accurate, because number one,

215 00:23:40.660 00:23:44.339 Demilade Agboola: We don’t have every single variant and how it is shipped.

216 00:23:44.590 00:23:47.880 Demilade Agboola: So that, like, restricts us.

217 00:23:48.060 00:24:01.910 Demilade Agboola: It’s also… even the ones we do have are limited to a certain pharmacy, certain pharmacies, certain products, so things like, NAD, not there. So it’s not this expansive, all-inclusive.

218 00:24:01.940 00:24:09.809 Demilade Agboola: solution. We’re basically trying to polish a turd, and trying to, like, answer questions when we don’t have the data

219 00:24:09.850 00:24:14.680 Demilade Agboola: And we need to, like, straight up say, sometimes, like, like…

220 00:24:15.710 00:24:27.899 Demilade Agboola: we need better data. We need the data to come in great. Like, we need the solutions, the systems to work. And I understand that Basque is problematic, because, you know, we’re about moving off,

221 00:24:28.060 00:24:43.429 Demilade Agboola: And Zach himself isn’t always the most responsive person, and given that, like, you know, they’re building the EMR and all of that, and that’s part of why we’re trying to figure out ways for the EMR. But, like, some certain things, we need to just sort of, like, be able to say, hey.

222 00:24:44.030 00:24:58.249 Demilade Agboola: the issue here isn’t necessarily an us issue, it’s… we’re trying to make the best. Like, the COGS issue, the Jonah issue, how we solved it, was we had to get, like, every single invoice for the past couple of months.

223 00:24:58.500 00:25:07.120 Demilade Agboola: Then Annie created a script to scrape out the information, and then I started basically taking the information from

224 00:25:07.430 00:25:14.559 Demilade Agboola: the things that Rebecca had shared, and applied the COGS directly to those values.

225 00:25:14.860 00:25:19.540 Demilade Agboola: from this script-out invoices. Like, that’s not ideal. Like, that’s not…

226 00:25:19.760 00:25:24.280 Demilade Agboola: We’re not… do you understand? That’s not an automated process. That’s not a process that scales.

227 00:25:24.420 00:25:28.750 Demilade Agboola: if Jonah has another request in, like, two months’ time.

228 00:25:28.920 00:25:33.570 Demilade Agboola: And we’re still in the same situation. It would have to be scripting off from invoices.

229 00:25:33.910 00:25:36.489 Demilade Agboola: Then, still, then applying in these sheets.

230 00:25:36.620 00:25:42.689 Demilade Agboola: That for every… Or any order between this date and this date, this is the value.

231 00:25:43.530 00:25:50.559 Demilade Agboola: like, I don’t know, sometimes we’re not… it’s not easy to be proactive with some of these things, because the data is so messed up that, like.

232 00:25:51.340 00:25:56.490 Demilade Agboola: It’s con… it’s like a consistent… reactionary thing.

233 00:25:59.230 00:25:59.890 Uttam Kumaran: Yeah.

234 00:26:03.280 00:26:09.310 Awaish Kumar: like, what I can say on this is that, like, we had COGS discussion I created to…

235 00:26:09.490 00:26:17.770 Awaish Kumar: Two kind of tickets, basically, which are, like, two different work streams to, … to optimize COGS, project.

236 00:26:17.970 00:26:26.459 Awaish Kumar: One is called, like, is for future improvements, which, like, we can plan how we want the data to come in in the warehouse.

237 00:26:26.640 00:26:30.519 Awaish Kumar: And prepare some documentation, which we can share with the MR team.

238 00:26:30.640 00:26:42.180 Awaish Kumar: when we are getting data, like, so they can make the data model ready, so we don’t have these problems in future. For right now, like, we have…

239 00:26:42.340 00:26:45.170 Awaish Kumar: We… we don’t have enough data to…

240 00:26:45.380 00:26:50.860 Awaish Kumar: and make… make things automated. Basque is not responsive.

241 00:26:51.200 00:26:53.679 Awaish Kumar: the only thing Demolara is doing is

242 00:26:53.940 00:26:59.340 Awaish Kumar: very manual work, which takes a long time. What needs to be done is

243 00:26:59.870 00:27:04.319 Awaish Kumar: Communicating to the client that this is really very… Thank you.

244 00:27:05.630 00:27:08.260 Awaish Kumar: Manual exercise, and it takes…

245 00:27:08.760 00:27:16.000 Awaish Kumar: Like, there’s that much of time to do that. So, if they wanna… if they’re, like, okay with…

246 00:27:16.220 00:27:24.600 Awaish Kumar: spending, like, our time on doing those manual exercises, then it’s okay, right? Otherwise, like, we can…

247 00:27:24.910 00:27:30.119 Awaish Kumar: We can basically get their consent on what they want us to spend our time on.

248 00:27:41.370 00:27:46.120 Uttam Kumaran: Yeah, this one is a little bit of a tougher one, because we don’t have the…

249 00:27:47.450 00:27:49.429 Uttam Kumaran: The data, but again, like….

250 00:27:50.540 00:27:56.189 Awaish Kumar: I think it just needs to get escalated, so maybe one… I have a document on this.

251 00:27:59.170 00:28:07.320 Uttam Kumaran: Yeah, so I think… I think for situations like this, for the engineering team, it’s important for us to translate to our

252 00:28:07.480 00:28:11.289 Uttam Kumaran: sort of project stakeholders, which is Robert and…

253 00:28:11.410 00:28:15.559 Uttam Kumaran: Amber, like, that we’re stuck between a rock and a hard place.

254 00:28:15.840 00:28:18.769 Uttam Kumaran: And then for them to escalate up.

255 00:28:19.140 00:28:24.020 Uttam Kumaran: So… I think that’s… blank.

256 00:28:28.220 00:28:36.860 Awaish Kumar: Yeah, like, I have a… I shared a Notion doc, so it has basically some context on this issue, but it’s, like, that… that’s what I…

257 00:28:37.910 00:28:41.090 Awaish Kumar: Thought, like, how can we approach this,

258 00:28:41.900 00:28:47.709 Awaish Kumar: One is getting consent from client, and second is, like, …

259 00:28:48.400 00:28:51.769 Awaish Kumar: I mean, again, this is another situation where, like.

260 00:28:52.920 00:28:58.290 Uttam Kumaran: We need to just be… involved in the COGS process.

261 00:28:58.640 00:28:59.460 Uttam Kumaran: Right.

262 00:29:00.230 00:29:05.470 Demilade Agboola: Definitely. Let me tell you another issue that potentially can become an issue down the line.

263 00:29:05.660 00:29:12.609 Demilade Agboola: For the past, like, 30 days, 80% of orders that have come in, thereabouts, don’t have a pharmacy attached to it.

264 00:29:13.590 00:29:16.490 Uttam Kumaran: Yeah. So eventually… Yeah, you’re… I mean….

265 00:29:16.490 00:29:17.120 Demilade Agboola: Eventually, for

266 00:29:17.610 00:29:24.760 Demilade Agboola: pharmacy performance, yeah, if we’re trying to do pharmacy performance, there’s a lot of null values. So how do you… like, these are the things that, like.

267 00:29:25.090 00:29:32.570 Demilade Agboola: And be very… and I have brought this up to Zach, by the way, so it’s not like as if I haven’t mentioned it. Prod up to Zach, something like…

268 00:29:33.040 00:29:40.909 Demilade Agboola: these numbers are, like, going really high. We don’t have a lot of pharmacy information here. What’s going on? He hasn’t responded to that.

269 00:29:41.360 00:29:46.779 Demilade Agboola: And then, eventually, it becomes, like, a crunch time thing, where someone needs it for, like, some farm-ups analysis.

270 00:29:46.780 00:29:51.380 Uttam Kumaran: It just needs… it just needs to… so… so there are kind of, I think, two pieces here. One.

271 00:29:51.690 00:29:54.660 Uttam Kumaran: If we’re not hearing back from people.

272 00:29:54.950 00:30:11.309 Uttam Kumaran: like, we’re gonna get stabbed by it, so we need to escalate. It just needs to find its way up the channel. The one… a great way to do this is to lean on Amber. Just to give her clear instructions and say, hey, we asked this, this person is not responding, like.

273 00:30:11.430 00:30:13.649 Uttam Kumaran: We’re gu- it’s gonna come back on us.

274 00:30:13.770 00:30:23.580 Uttam Kumaran: we need help here. The second piece is, if we’re not able… like, closing the books on certain types of reporting is a very common practice.

275 00:30:23.860 00:30:33.170 Uttam Kumaran: Like, maybe we can’t agree to have, like, last month’s thing closed until a certain date, or we have SLAs on reporting.

276 00:30:33.680 00:30:42.089 Uttam Kumaran: Right? This is… this is something we have to mention. Everybody will think everything’s up-to-date every day, but we need to understand, like, what are the…

277 00:30:42.290 00:30:49.090 Uttam Kumaran: like, what are the drivers of slowness? Like, what’s the driver, what’s the minimum, like, the longest path, basically?

278 00:30:49.390 00:30:55.599 Uttam Kumaran: So if you’re like, hey, we can’t get this data in, and so for that reason, we can’t guarantee more than 7-day freshness.

279 00:30:55.890 00:30:57.929 Uttam Kumaran: Then it can escalate, right?

280 00:30:58.600 00:31:01.299 Uttam Kumaran: I mean, this is where, like, our team…

281 00:31:01.890 00:31:06.819 Uttam Kumaran: As part of the solution, and we’re driving to the solution, so we need to… you need to kind of…

282 00:31:07.150 00:31:10.620 Uttam Kumaran: Like, walk the line of, we’re not throwing this back on them.

283 00:31:10.720 00:31:17.800 Uttam Kumaran: But we’re probably the only people with the courage to, like, escalate. Or, like, with the purview to escalate, maybe.

284 00:31:20.140 00:31:21.270 Uttam Kumaran: …

285 00:31:23.570 00:31:30.340 Uttam Kumaran: So it’s kind of another example. So I don’t know, Robert, if you have any thoughts on… on either of these, or, like.

286 00:31:31.070 00:31:37.629 Uttam Kumaran: How… what’s our… like, what the client would think about any… either of these type of work streams of the solutions.

287 00:31:37.880 00:31:48.950 Robert Tseng: Yeah, I mean, I hear the short-term things that, like, yeah, I mean, the Dimlade’s voice is something that I’m familiar with. Yeah, I think it’s…

288 00:31:50.980 00:31:59.250 Robert Tseng: We… we just… we’re… we’re always reactive. I’ve said this multiple times, get us into meetings and stuff where they’re discussing.

289 00:31:59.370 00:32:01.380 Robert Tseng: Just, like, everything…

290 00:32:02.070 00:32:11.880 Robert Tseng: everything just kind of… when I’ve raised that request before, all they do is add me to, like, 2 or 3 meetings, which I… I don’t want to be a part of those meetings. I’m not gonna sit there for hours….

291 00:32:11.880 00:32:14.749 Uttam Kumaran: We should probably run a meeting.

292 00:32:16.460 00:32:20.320 Robert Tseng: Just with them on this type of stuff, that’s just, like, product changes or product updates.

293 00:32:20.320 00:32:23.180 Uttam Kumaran: That is, yeah, and it’s being run by our side.

294 00:32:24.510 00:32:25.150 Robert Tseng: Yeah.

295 00:32:25.390 00:32:28.769 Uttam Kumaran: Because, yeah, we don’t care about… I don’t… we only care about what affects us.

296 00:32:29.190 00:32:32.550 Uttam Kumaran: Yeah. So… Yeah.

297 00:32:34.060 00:32:35.919 Robert Tseng: Yeah, we could, we could do that.

298 00:32:37.270 00:32:38.030 Uttam Kumaran: …

299 00:32:39.820 00:32:50.459 Uttam Kumaran: like… and it… it doesn’t have to be every week, but I would probably keep it high frequency until we don’t need to. And yeah, like, we run it. We run a cross-functional…

300 00:32:50.670 00:32:55.959 Uttam Kumaran: Sort of, like, data quality thing, where we get what are the new products, what are the new campaigns.

301 00:32:56.470 00:33:03.489 Uttam Kumaran: what’s going on with COGS? That way, because we just need to get ahead of it. Like, so, if we’re on the hook.

302 00:33:04.140 00:33:08.650 Uttam Kumaran: Then we have to run that meeting, And I think that’s it.

303 00:33:08.820 00:33:14.640 Uttam Kumaran: I don’t know, I don’t see another… except for the… I don’t see another solution for the second one.

304 00:33:17.830 00:33:30.690 Uttam Kumaran: I mean, and also, you should make it clear, the goal is to not have this meeting, or to start to, like, increase the time, but you could say, hey, this meeting can be Slack, but you guys are not… like, we’re not there yet, so….

305 00:33:30.940 00:33:31.560 Robert Tseng: Yeah.

306 00:33:38.100 00:33:41.970 Robert Tseng: Okay, yeah, I mean, I can… I can set that up with them.

307 00:33:44.270 00:33:45.259 Awaish Kumar: It’s a….

308 00:33:45.260 00:33:50.870 Uttam Kumaran: It’s the worst solution, is yet another meeting, but… It… it’s….

309 00:33:50.870 00:34:01.900 Robert Tseng: I don’t want to put it in the middle… I don’t want to put it early in the week. I want us to kind of do it towards the end, so, like, it’s not an urgent thing. Like, I don’t know, would… like, what if we set it up on, like, a Friday or something? I don’t know.

310 00:34:02.120 00:34:09.050 Robert Tseng: Thursday or Friday, and then that gives us time to, like, Make a change by….

311 00:34:09.050 00:34:14.250 Uttam Kumaran: Again, it’s probably something to align with them on, like, when are these changes being made, and, like, when can we…

312 00:34:14.960 00:34:17.469 Uttam Kumaran: Reasonably expect, you know, so….

313 00:34:18.989 00:34:20.479 Demilade Agboola: That’s fair enough.

314 00:34:21.350 00:34:22.010 Robert Tseng: Yeah.

315 00:34:25.120 00:34:34.239 Awaish Kumar: Yeah, but, like, Today, like… Make it, like, find… for example, create a doc where we share

316 00:34:35.580 00:34:37.680 Awaish Kumar: Where do we write down, like, the…

317 00:34:38.000 00:34:42.339 Awaish Kumar: For example, COGS, what’s happening with COGS, how much time is spent there?

318 00:34:42.500 00:34:46.089 Awaish Kumar: What’s happening with campaigns and all of it, and it goes as an…

319 00:34:46.250 00:34:49.650 Awaish Kumar: Weekly updates, as part of weekly updates to the client.

320 00:34:49.650 00:34:50.480 Uttam Kumaran: Yes.

321 00:34:50.980 00:35:07.010 Uttam Kumaran: like, this is the thing. This is, like, a famous problem with data teams, is that we are not making the decisions, however, we are completely affected by those. And we are probably the only people that care about

322 00:35:07.300 00:35:19.759 Uttam Kumaran: like, this data quality, but a great positive outcome is that maybe we own the reporting, right? Like, meaning these are the outcomes. We have these things launching, we have these things, and we share that out.

323 00:35:19.930 00:35:25.360 Uttam Kumaran: that helps us look better, you know, in terms of communication. It keeps a lot of other people in the loop.

324 00:35:26.190 00:35:32.059 Uttam Kumaran: And, yeah, I mean, look, all those people care… I sometimes don’t want to blame them, because they…

325 00:35:32.210 00:35:39.789 Uttam Kumaran: or care about their own thing, right? Like, if their small size of reporting works, or maybe their work doesn’t need reporting, it’s fine.

326 00:35:40.030 00:35:41.860 Uttam Kumaran: And… but…

327 00:35:41.990 00:35:50.050 Uttam Kumaran: we… we have to care, and so we need to sort of own that meeting, and then we report out, here are the changes. I think that’s a positive outcome.

328 00:35:52.160 00:35:54.290 Uttam Kumaran: And I think all those people will lean on us.

329 00:35:56.480 00:36:06.670 Uttam Kumaran: Because there’s nobody, like, as cross-functional as we are right now, so I think it would be positive for all those folks to, like, have a concise set of updates that they can share on this.

330 00:36:06.810 00:36:09.780 Uttam Kumaran: And we needed anyways, for our work, so…

331 00:36:10.050 00:36:13.730 Uttam Kumaran: If we have to drive that, then I think so. Then we go ahead.

332 00:36:24.340 00:36:29.880 Demilade Agboola: I agree. I think that makes sense. It’s a bit frustrating, but yeah, it is what we have to do at this point.

333 00:36:30.770 00:36:36.640 Uttam Kumaran: Yeah, look, any sort of recurring meeting, we should find a way to… life.

334 00:36:37.170 00:36:39.030 Uttam Kumaran: Remove it over time.

335 00:36:39.240 00:36:45.690 Uttam Kumaran: But it’s just gonna be something here that, like, we need to solve it, and then we can optimize over time.

336 00:36:46.180 00:36:52.850 Uttam Kumaran: That would be my… that’d be my best thing. So, if we’re okay with that, then… …

337 00:36:53.080 00:36:54.949 Uttam Kumaran: I would probably recommend that.

338 00:36:55.300 00:36:59.169 Uttam Kumaran: OH Demulade, and you could have probably worked with Amber to book that.

339 00:36:59.400 00:37:07.289 Uttam Kumaran: And then, yeah, on the first piece, we can… if we can get everything into that dock, and then we can start to ticket out some of the changes.

340 00:37:08.460 00:37:10.180 Awaish Kumar: Yeah, sure, I’ll do that.

341 00:37:11.210 00:37:11.870 Uttam Kumaran: Okay.

342 00:37:17.150 00:37:18.419 Awaish Kumar: Okay, thank you.

343 00:37:19.400 00:37:20.460 Robert Tseng: Alright, thanks guys.

344 00:37:20.730 00:37:21.540 Uttam Kumaran: Yeah, thank you.

345 00:37:21.680 00:37:23.280 Demilade Agboola: Alright, thanks guys. Bye.