Meeting Title: AI Service Morning Huddle Date: 2026-05-07 Meeting participants: Mustafa Raja, Samuel Roberts, Pranav, Casie Aviles


WEBVTT

1 00:00:54.800 00:00:55.710 Samuel Roberts: Hello?

2 00:00:58.570 00:00:59.430 Mustafa Raja: Hey, how are you?

3 00:01:00.950 00:01:02.919 Samuel Roberts: Doing alright. How about yourself?

4 00:01:03.560 00:01:04.989 Mustafa Raja: Good.

5 00:01:06.900 00:01:08.010 Samuel Roberts: Lipinov?

6 00:01:10.550 00:01:11.800 Pranav: Hey guys, good morning.

7 00:01:12.600 00:01:13.919 Mustafa Raja: Good morning.

8 00:01:13.920 00:01:15.410 Samuel Roberts: Morning. Oh. How you doing?

9 00:01:17.230 00:01:18.470 Pranav: Pretty good, pretty good.

10 00:01:19.360 00:01:20.350 Samuel Roberts: Where are you now?

11 00:01:21.830 00:01:22.890 Pranav: I’m still in Chicago.

12 00:01:23.280 00:01:24.260 Samuel Roberts: Okay, okay.

13 00:01:25.370 00:01:26.310 Samuel Roberts: Cool, cool.

14 00:01:26.860 00:01:30.920 Samuel Roberts: How have you… whereabouts in Chicago are you, like, Staying.

15 00:01:31.680 00:01:34.080 Pranav: Pretty close to downtown.

16 00:01:34.080 00:01:34.650 Samuel Roberts: Nice.

17 00:01:34.920 00:01:39.209 Pranav: Yeah, it’s, like the River North area, if you’re familiar with Chicago.

18 00:01:39.210 00:01:42.920 Samuel Roberts: A little bit, yeah, I’ve been there a couple times, but it’s been a minute, to be honest, so…

19 00:01:43.510 00:01:44.540 Pranav: Gotcha, okay.

20 00:01:48.110 00:01:49.200 Pranav: Hey, Casey.

21 00:01:50.540 00:01:50.900 Casie Aviles: There you go.

22 00:01:53.930 00:01:55.400 Pranav: Okay,

23 00:01:55.840 00:02:02.840 Pranav: Yeah, Mustafa, I saw that you sent over that report. I’m gonna take a look at it right after this, and then we can probably send something over.

24 00:02:02.970 00:02:10.570 Pranav: Yeah, Sam, I know you didn’t make the meeting in the afternoon yesterday, so if you can just kind of stick with your updates.

25 00:02:12.000 00:02:15.680 Samuel Roberts: I mean, that was the thing I sent you. I was out the rest of the day, so…

26 00:02:16.270 00:02:23.919 Samuel Roberts: Okay, cool. The current status is, like, I kind of pivoted to make sure I could generate something nice looking, and I’m just cleaning up a little bit to, like.

27 00:02:24.220 00:02:29.079 Samuel Roberts: Make sure that those are, you know, maybe not that specific thing, but make sure that the tool can still

28 00:02:29.230 00:02:48.740 Samuel Roberts: you know, fixing the edge cases with when it fails and, fallbacks and things like that, but it’s essentially there, and that was… I mean, if that’s the pilot, you know, if we have other queues and stuff, I can now run it whenever. So, and then, assuming that all seems good, or we need to iterate, we can, you know, use some of these examples as evals or something moving forward.

29 00:02:48.740 00:02:55.020 Samuel Roberts: But yeah, there’s… ticket-wise, it kind of has those… it’s kind of there for both of those tickets now.

30 00:02:55.180 00:02:57.770 Samuel Roberts: So, I can update the tickets and move on.

31 00:02:58.660 00:02:59.630 Pranav: Cool, cool.

32 00:02:59.790 00:03:03.710 Samuel Roberts: You have a meeting today, right? It wasn’t yesterday with that, what I sent you?

33 00:03:04.440 00:03:05.070 Samuel Roberts: For ABC.

34 00:03:05.070 00:03:05.719 Pranav: Wait, what?

35 00:03:06.290 00:03:07.859 Samuel Roberts: the meeting today with ABC?

36 00:03:08.460 00:03:15.919 Pranav: I have a meeting today, but then I’ll also have our weekly meeting tomorrow with them, so I’ll probably present it tomorrow.

37 00:03:16.270 00:03:17.910 Samuel Roberts: Oh, okay, okay, cool.

38 00:03:18.040 00:03:18.680 Pranav: Yeah.

39 00:03:19.220 00:03:20.590 Pranav: Yeah. Okay.

40 00:03:20.590 00:03:21.020 Samuel Roberts: Okay, cool.

41 00:03:21.020 00:03:37.319 Pranav: Cool, and then, Casey, you were working on embedding those two documents, right? And I think those… I already saw your message to give that. It sounds like that went well. You were able to do, like, some testing as well to test the retrieval, especially from the table, which was our primary concern.

42 00:03:37.900 00:03:43.549 Casie Aviles: Yeah, basically I just… Converted it into a… Let’s tighten for now.

43 00:03:43.790 00:03:45.599 Casie Aviles: Or, yeah, a bulleted list.

44 00:03:46.190 00:03:51.510 Casie Aviles: Nice. I mean, I just made sure that it’s parsable by our current pipeline, so…

45 00:03:52.660 00:03:53.490 Pranav: Perfect.

46 00:03:53.750 00:03:56.219 Pranav: Cool. And then, was there anything else from yesterday?

47 00:03:56.550 00:04:02.589 Casie Aviles: I do have a ticket that we put back to in progress that I just wanted to…

48 00:04:02.750 00:04:05.170 Casie Aviles: Ask some questions on.

49 00:04:05.620 00:04:07.400 Pranav: Yeah, sure, let’s do that now.

50 00:04:07.980 00:04:16.000 Casie Aviles: So I have… so it’s this add weekly report with zero feedback, so if… so what we talked about before.

51 00:04:16.480 00:04:21.839 Casie Aviles: Was to create, like, Google Sheets, right, that would have, like, all the…

52 00:04:22.400 00:04:27.049 Casie Aviles: input that did not get a feedback. Is that… Yeah. Yeah.

53 00:04:27.180 00:04:32.150 Casie Aviles: So I guess my only… the only concern I have there is…

54 00:04:32.760 00:04:35.740 Casie Aviles: now that I thought about it, is… we’re gonna…

55 00:04:36.170 00:04:43.749 Casie Aviles: Create more, documents, and that might create some document sprawl, or… And, add additional processes.

56 00:04:43.920 00:04:47.040 Casie Aviles: And I’m not sure if… how…

57 00:04:47.160 00:04:50.460 Casie Aviles: If it… if it… if the overhead is better.

58 00:04:50.620 00:04:54.879 Casie Aviles: fine. What I was thinking of, if we want to just prevent, like.

59 00:04:55.640 00:04:58.230 Casie Aviles: Not getting any feedback at all is…

60 00:04:58.580 00:05:04.380 Casie Aviles: Maybe we can have, like, a… like, tags that we can… Like, clickable tags that…

61 00:05:05.130 00:05:07.000 Casie Aviles: The users would click.

62 00:05:07.100 00:05:11.049 Casie Aviles: If they don’t want to type something, because that might be a…

63 00:05:11.220 00:05:15.219 Casie Aviles: cause of friction, right? And when they have to type in, like.

64 00:05:15.630 00:05:20.029 Casie Aviles: Feedback, so maybe we could have, like, tags there.

65 00:05:20.210 00:05:26.439 Casie Aviles: Like, for example, this is a hallucination, or this is irrelevant, and that would, you know.

66 00:05:26.440 00:05:28.960 Pranav: With, document crawl, I don’t…

67 00:05:29.280 00:05:38.670 Pranav: I don’t see why that may be an issue here. You know, I see that maybe more so if that was something to do with the central docs, you know, we just had a bunch of different documents in different locations.

68 00:05:38.720 00:05:51.809 Pranav: This is more so just a feedback loop for us. So, we’re just gonna be able to give them a sheet saying, hey, we didn’t get any feedback on these, are they good to go or not? And it’ll just help us build a better golden dataset.

69 00:05:53.060 00:05:54.100 Pranav: Okay.

70 00:05:54.860 00:05:57.430 Pranav: Does that still seem like document scroll to you?

71 00:05:57.570 00:05:59.879 Pranav: And you can be honest, like, because we can…

72 00:05:59.880 00:06:00.550 Casie Aviles: Yeah, I’m…

73 00:06:00.730 00:06:03.660 Pranav: Well, seeing, kind of, where that would be an issue.

74 00:06:04.420 00:06:06.049 Casie Aviles: Yeah, I was just thinking, like.

75 00:06:06.760 00:06:15.140 Casie Aviles: we may be able to leverage Rail for that, if you want to just show them, like, which .

76 00:06:15.430 00:06:26.369 Pranav: Well, the idea is that they’ll be able to input as well. So I want them to, if they give a thumbs up, or if there’s no thumbs up or thumbs down, they will be able to provide the thumbs up and thumbs down, and…

77 00:06:26.560 00:06:31.179 Pranav: Additional, you know, if it’s a thumbs down, then giving a reason why is thumbs down.

78 00:06:32.690 00:06:34.070 Casie Aviles: Okay, okay.

79 00:06:34.070 00:06:38.849 Pranav: Is that something we can do in Rel? Most of what we’ve been showing on Rel is just, you know.

80 00:06:38.850 00:06:40.830 Casie Aviles: No, no, yeah, we can’t do that.

81 00:06:42.810 00:06:43.580 Pranav: Okay.

82 00:06:45.850 00:06:55.450 Pranav: Yeah, I mean, we write to Google Sheets, right? Like, I think you’ve built scripts that write to Google Sheets before. I don’t think it’s that big of a… a lift.

83 00:06:58.760 00:07:01.439 Casie Aviles: Okay, yeah, I was just, like, thinking if…

84 00:07:01.540 00:07:04.520 Casie Aviles: We have more sheets to maintain, but if…

85 00:07:04.970 00:07:08.970 Casie Aviles: we think that’s okay, then yeah, I’m fine with that as well, because…

86 00:07:09.240 00:07:11.389 Casie Aviles: Yeah, that was just what I was thinking of.

87 00:07:12.030 00:07:19.099 Pranav: Okay, yeah, so also the maintenance of this will just be, you know, after we get their feedback, we will just, like.

88 00:07:19.700 00:07:24.050 Pranav: We will probably run that through our system, you know, we’ll be…

89 00:07:24.170 00:07:30.659 Pranav: Primarily, what this is, is also just a measure of how are things working for Andy, and so…

90 00:07:31.060 00:07:35.389 Pranav: What this will… what this will do is give us a better estimate of

91 00:07:35.690 00:07:38.430 Pranav: When we’re giving the right answer and when we’re giving the wrong answer.

92 00:07:40.970 00:07:51.319 Pranav: But then after, you know, the week passes, we don’t need to support that document, you know? And it’ll be really easy to maintain, I think, because it’ll all be just within one Google Sheet.

93 00:07:51.770 00:07:59.429 Pranav: And so, when I say one Google Sheet as well, you know, on Google Sheets you can have multiple, like, sheet numbers within there, right?

94 00:07:59.620 00:08:00.470 Casie Aviles: Yeah, yeah.

95 00:08:00.720 00:08:04.979 Pranav: Yeah, so then each sheet number would essentially just correspond to each individual week.

96 00:08:05.240 00:08:13.469 Pranav: So, in terms of maintenance, I’m not too worried about that. We would have also probably, like, 5…

97 00:08:14.240 00:08:16.589 Pranav: Files, one per department.

98 00:08:17.060 00:08:19.179 Casie Aviles: Yeah, yeah, BDP, okay.

99 00:08:19.520 00:08:20.310 Pranav: Yeah.

100 00:08:21.810 00:08:39.599 Pranav: We can see how that goes. After a certain period of time, too, like, this might not be… we’ll just see how this works, right? Like, I’m trying to think of solutions here so that we can get more good data. Right now, we’re seeing probably, like, 10-15% in terms of feedback on…

101 00:08:40.200 00:08:46.870 Pranav: either, like, when I say feedback, like, thumbs up or thumbs down on Andy responses, which is…

102 00:08:47.320 00:08:53.400 Pranav: much lower than where I want to be, especially since, you know, most CSRs are only using ANDI

103 00:08:54.030 00:09:03.599 Pranav: a dozen times. That’s probably, like, the average amount of usage per CSR, and there’s a lot of CSRs that are still not using it at all. So if you’re only using it 12 times, you should be able to give it thumbs up, thumbs down.

104 00:09:03.940 00:09:09.599 Pranav: So that’s why I want closer to, like, you know, hopefully, like, we can get to, like, 75%, like…

105 00:09:09.770 00:09:14.120 Pranav: In terms of feedback, so… Yeah.

106 00:09:14.490 00:09:15.169 Casie Aviles: Okay, okay.

107 00:09:15.170 00:09:19.830 Samuel Roberts: Oh, yeah, question… question about that, actually. A little sidetracked, but I don’t know, I want you guys to finish what you were talking about.

108 00:09:21.170 00:09:33.619 Pranav: Yeah, I mean, does that make sense, Casey? Do you… is there, in terms of implementation, is there something that you’re feeling that there… this is more of a heavy lift, or are you just kind of talking about future maintenance?

109 00:09:34.070 00:09:40.619 Casie Aviles: No, no, I was just thinking of, like, you know, we may, yeah, maintenance, like, we may have a lot of…

110 00:09:40.730 00:09:48.510 Casie Aviles: new documents, and back then we were trying to kind of reduce that, but yeah, it’s fine. I think I get what…

111 00:09:48.640 00:09:52.169 Casie Aviles: the… What this solves, so… yeah.

112 00:09:52.340 00:09:57.500 Pranav: Okay. Yeah, I mean, I’m happy to hear, like, any other solutions as well.

113 00:09:57.610 00:10:02.179 Pranav: But I think this is the best solution from… From what I’ve thought of.

114 00:10:03.770 00:10:08.530 Casie Aviles: Okay, yeah, yeah, we can try this out. I’m good with this one.

115 00:10:09.260 00:10:10.600 Pranav: Okay, sounds good.

116 00:10:12.090 00:10:20.340 Pranav: Alright, yeah, Mustafa, I saw that you worked on… think…

117 00:10:20.860 00:10:33.290 Pranav: the day… or I think maybe that was the day before, where you worked on just the analysis comparing our command center with Danny’s, like, simulated Cloud code. And then yesterday, I think you were working on…

118 00:10:33.400 00:10:43.000 Pranav: Send, or… Using context of previous reports to, for the future reports?

119 00:10:44.260 00:10:49.169 Pranav: And so, that’s what you sent me. What’s the progress on the email sender?

120 00:10:50.670 00:10:56.339 Mustafa Raja: Yeah, so, If you prove this, what I’ll do is,

121 00:10:56.450 00:10:59.059 Mustafa Raja: I’ll connect your email, and then,

122 00:10:59.380 00:11:08.150 Mustafa Raja: what’s then going to happen once this workflow is done, it’s going to save it, save the email as draft in your…

123 00:11:08.630 00:11:11.160 Mustafa Raja: And you… Gmail.

124 00:11:12.190 00:11:20.430 Pranav: Okay, yeah, I mean, you can get started on that, now. Shouldn’t be a blocker for that, especially since we’re just sending things as drafts.

125 00:11:21.170 00:11:25.579 Pranav: And so, yeah… You can, start working on that.

126 00:11:25.960 00:11:29.049 Pranav: And then I’ll get that feedback over to you.

127 00:11:30.710 00:11:31.380 Mustafa Raja: Okay.

128 00:11:32.390 00:11:37.039 Pranav: Cool. So yeah, I would say that’s number one priority for today for you.

129 00:11:38.590 00:11:42.159 Pranav: Let’s see… let me… let me take a look at linear as well.

130 00:11:50.840 00:11:53.760 Pranav: Okay, is that… is that,

131 00:11:55.260 00:12:00.450 Pranav: defined in this ticket, EDE3-17 for you, Mustafa?

132 00:12:01.600 00:12:03.910 Mustafa Raja: Let me just look correctly, also.

133 00:12:05.870 00:12:07.389 Pranav: Or is that something else?

134 00:12:19.770 00:12:24.020 Mustafa Raja: Internal validation… No, this is something else, I believe.

135 00:12:24.460 00:12:27.400 Pranav: Yeah, so… Okay.

136 00:12:30.420 00:12:34.759 Mustafa Raja: I think you’ve seen… I went through all these tickets.

137 00:12:35.360 00:12:48.020 Pranav: Yeah, yeah, I think there is a ticket for that in the backlog, so I’ll start assigning them later today, and you guys can let me know if, everything is clear in those, but I think, in terms of what you’re working on today, Mustafa, that should be pretty clear, right?

138 00:12:48.650 00:12:49.590 Mustafa Raja: Yes.

139 00:12:49.930 00:12:51.080 Pranav: Okay, cool.

140 00:12:51.300 00:12:58.500 Pranav: And Casey, yeah, just working on that, that report will be super helpful.

141 00:12:58.730 00:13:04.300 Pranav: Let me take a look at the other ABC tickets…

142 00:13:17.010 00:13:24.340 Pranav: Yeah, Sam, if you can start setting up, like, a… A… an automated weekly?

143 00:13:26.050 00:13:32.319 Pranav: Assessment of… Where we’re at in terms of, like, cancellations.

144 00:13:32.490 00:13:33.879 Pranav: That would be great.

145 00:13:34.400 00:13:42.819 Pranav: So I know right now you kind of, like, that POC of just, like, the one-time script. Let’s start, let’s start building this out so that it runs on a weekly cadence.

146 00:13:43.280 00:13:47.319 Samuel Roberts: Okay, yeah, cool. I wasn’t sure if you wanted to do that before or after they see it or not, but yeah, that works.

147 00:13:47.510 00:13:47.980 Samuel Roberts: Happening.

148 00:13:47.980 00:13:55.519 Pranav: Yeah, I mean, I think what you did was the proof of concept. Anything that they might have comments on is just gonna be, like, refining the output, so…

149 00:13:55.520 00:13:59.410 Samuel Roberts: Yeah, I figured. That makes sense, yeah. Okay. That works, yeah.

150 00:13:59.410 00:13:59.950 Pranav: You can start.

151 00:13:59.950 00:14:00.620 Samuel Roberts: Sounds good.

152 00:14:01.480 00:14:02.170 Samuel Roberts: Cool.

153 00:14:03.030 00:14:08.280 Samuel Roberts: Yeah, the question I was asking earlier, that email you sent from Yvette, or forwarded?

154 00:14:08.400 00:14:14.299 Samuel Roberts: When… The email asked about the guidelines, is that…

155 00:14:15.020 00:14:19.310 Samuel Roberts: Is he talking about the central docs, or do they have a separate thing that is also being maintained?

156 00:14:21.360 00:14:22.629 Pranav: Let me see…

157 00:14:25.770 00:14:29.909 Pranav: Do you know what the CSR is referring to for guidelines in the Central Docs?

158 00:14:39.500 00:14:41.570 Casie Aviles: I’m not too sure either, but…

159 00:14:41.570 00:14:41.900 Samuel Roberts: Okay.

160 00:14:41.900 00:14:47.900 Casie Aviles: They do have, like, a spreadsheet that has guidelines in the name.

161 00:14:48.140 00:14:51.640 Casie Aviles: But I’m not sure if that’s what they were talking about.

162 00:14:52.720 00:14:58.480 Casie Aviles: Like, lawn service guidelines for all branches. That’s what they… one of the sheets that they had.

163 00:15:01.740 00:15:02.330 Samuel Roberts: Okay.

164 00:15:02.590 00:15:10.680 Samuel Roberts: Yeah, I wasn’t sure if that’s, like, another source of truth that we either have folded in, or is not folded in, and that’s what he’s talking about, or…

165 00:15:13.850 00:15:19.140 Samuel Roberts: It’s just, yeah, the word guidelines was, like, a little bit of, like, a, oh no, I don’t know what he’s referring to here, and if that’s more dangerous.

166 00:15:19.140 00:15:19.830 Pranav: Yeah.

167 00:15:20.290 00:15:25.149 Pranav: Yeah, so, Casey, were you just saying that this is probably just guidelines, central docs is just synonymous?

168 00:15:26.720 00:15:32.219 Casie Aviles: No, no, I’m… I’m… it’s… that’s… let me just share my screen, and I can show you.

169 00:15:32.220 00:15:33.030 Samuel Roberts: Yeah, cool.

170 00:15:34.300 00:15:35.640 Casie Aviles: Let me just look for that.

171 00:15:35.990 00:15:36.870 Casie Aviles: File.

172 00:15:41.490 00:15:43.460 Casie Aviles: Okay.

173 00:15:45.270 00:15:49.130 Casie Aviles: So, it… there’s… they just have the sheet that has… that it’s…

174 00:15:49.280 00:15:51.719 Casie Aviles: Just loan service guidelines for all branches.

175 00:15:51.720 00:15:52.350 Samuel Roberts: Okay.

176 00:15:52.750 00:15:58.020 Casie Aviles: And… Yeah, this… this is what I use to get, like, some of the zip.

177 00:15:58.640 00:15:59.880 Casie Aviles: Assignments.

178 00:16:00.710 00:16:01.200 Pranav: Gotcha.

179 00:16:01.200 00:16:05.810 Casie Aviles: But they also have other sheets here that are just, you know, more… Kind of.

180 00:16:06.590 00:16:08.680 Casie Aviles: Okay. More plain text, I believe.

181 00:16:08.920 00:16:09.640 Casie Aviles: Or…

182 00:16:09.930 00:16:10.590 Pranav: Okay.

183 00:16:10.730 00:16:11.660 Casie Aviles: Summit.

184 00:16:12.700 00:16:14.240 Casie Aviles: Gotcha. They also have these.

185 00:16:15.010 00:16:17.499 Pranav: Yeah, so there’s probably certain CSRs that are…

186 00:16:17.990 00:16:27.509 Pranav: I just like going into the central doc and finding the information. Probably certain ones, I just like going into these Google Sheets to find certain stuff. Yeah.

187 00:16:29.880 00:16:35.300 Samuel Roberts: I just worry about, like, drift, basically, at this point, if they’re still maintaining additional dots.

188 00:16:36.280 00:16:43.869 Pranav: everything that they’re looking at, I think, what we’re… is something that we’re aware of that is supposed to be in…

189 00:16:43.990 00:16:46.910 Pranav: Either the central docs or in the zip code DB.

190 00:16:47.880 00:16:52.619 Samuel Roberts: Okay. Just making sure there wasn’t something else, because I didn’t know what he was exactly referring to, but okay, cool.

191 00:16:54.370 00:16:56.129 Samuel Roberts: then I’m not as worried then.

192 00:16:58.840 00:17:01.850 Samuel Roberts: Yeah, I just wasn’t sure the terminology was possible. Cool.

193 00:17:03.420 00:17:05.000 Pranav: Yeah, no, that makes sense.

194 00:17:05.250 00:17:08.920 Pranav: Okay, guys. Anything else?

195 00:17:11.920 00:17:13.320 Casie Aviles: No, that’s all for me.

196 00:17:15.450 00:17:16.050 Pranav: Alright, cool.

197 00:17:16.050 00:17:20.999 Samuel Roberts: Yeah, no, I guess, let’s make a ticket for the, like, automated, you know.

198 00:17:22.190 00:17:25.750 Samuel Roberts: Automation part of the… The transcripts now.

199 00:17:26.030 00:17:29.840 Pranav: Okay, perfect, yeah, if you wanna, if you wanna make that, that would be super helpful. Yeah.

200 00:17:29.840 00:17:31.700 Samuel Roberts: And then I’ll… I’ll write on that today.

201 00:17:32.330 00:17:32.700 Pranav: Perfect.

202 00:17:32.700 00:17:33.610 Samuel Roberts: Yeah.

203 00:17:34.560 00:17:36.219 Pranav: Alright. Cool, guys.

204 00:17:36.490 00:17:37.920 Pranav: Yeah.

205 00:17:38.670 00:17:50.040 Pranav: If, if you guys run into any blockers, just let me know. I’ll be around, so feel free to hit me up, we can hop into a meeting. But yeah, if not, we’ll… we’ll talk again later in this afternoon.

206 00:17:51.390 00:17:51.960 Samuel Roberts: Sounds good.

207 00:17:51.960 00:17:52.540 Casie Aviles: Okay.

208 00:17:52.790 00:17:53.390 Casie Aviles: Thank you.

209 00:17:53.390 00:17:54.070 Pranav: See you guys.

210 00:17:54.730 00:17:55.900 Samuel Roberts: Alright, bye.