Meeting Title: AI-App Standup Date: 2025-12-18 Meeting participants: Casie Aviles, Mustafa Raja, Samuel Roberts, Gabriel Lam, Uttam Kumaran
WEBVTT
1 00:00:17.120 ⇒ 00:00:18.350 Mustafa Raja: Hey, Casey.
2 00:00:20.510 ⇒ 00:00:21.969 Casie Aviles: There you go. Guys.
3 00:00:22.120 ⇒ 00:00:22.939 Casie Aviles: Name a cell phone.
4 00:00:28.060 ⇒ 00:00:29.050 Mustafa Raja: How are you?
5 00:00:29.780 ⇒ 00:00:34.179 Casie Aviles: Yeah, doing good. Yeah, thank you for taking on ABC.
6 00:00:34.270 ⇒ 00:00:35.090 Mustafa Raja: No worries.
7 00:00:36.860 ⇒ 00:00:39.729 Casie Aviles: I know there’s… there’s been a lot going on.
8 00:00:41.540 ⇒ 00:00:52.320 Casie Aviles: Yeah, I was just curious, like, what… Where can I… Help with… with the.
9 00:00:52.320 ⇒ 00:01:03.780 Mustafa Raja: I mean, I think, I think there’s a lot of missing data in the database, you know? So, a lot of the questions that they are asking and they’re not getting answers about is just…
10 00:01:04.050 ⇒ 00:01:07.629 Mustafa Raja: Because the assignments are not there, you know?
11 00:01:08.450 ⇒ 00:01:09.210 Casie Aviles: Hmm.
12 00:01:09.210 ⇒ 00:01:09.650 Mustafa Raja: in the day.
13 00:01:10.070 ⇒ 00:01:12.929 Mustafa Raja: And so we need to fill them. And then…
14 00:01:13.360 ⇒ 00:01:17.250 Mustafa Raja: We have some also… we also have some outdated data in the…
15 00:01:18.030 ⇒ 00:01:25.049 Mustafa Raja: Some of the people, they have moved departments or something, or services or something, and they’re still in there, you know?
16 00:01:26.010 ⇒ 00:01:27.769 Casie Aviles: Yes, okay, and…
17 00:01:28.430 ⇒ 00:01:38.379 Mustafa Raja: And yeah, it’s just things like that, and we have a sheet that we analyzed. Let me link it to you, so you could also take a look at that.
18 00:01:39.560 ⇒ 00:01:40.220 Casie Aviles: Okay.
19 00:01:41.270 ⇒ 00:01:42.350 Samuel Roberts: Hey, Casey.
20 00:01:43.440 ⇒ 00:01:44.769 Casie Aviles: Hey guys, hey Sam.
21 00:01:44.880 ⇒ 00:01:47.380 Casie Aviles: Yeah, thank you for taking on ABC.
22 00:01:47.930 ⇒ 00:01:51.339 Samuel Roberts: Oh, yeah, no, I mean, yeah, it,
23 00:01:51.670 ⇒ 00:01:54.679 Samuel Roberts: Of course, of course. I feel, you know…
24 00:01:55.100 ⇒ 00:02:00.740 Samuel Roberts: bad that, like, it’s been a lot on you in any way, so it’s, you know, I’m glad we’re all more familiar with it now.
25 00:02:02.120 ⇒ 00:02:09.619 Casie Aviles: I was just wonder… I was just asking Mustafa, like, where can I… what else can I pick up, and help with?
26 00:02:10.930 ⇒ 00:02:11.770 Samuel Roberts: Yeah.
27 00:02:12.410 ⇒ 00:02:18.770 Samuel Roberts: So we have… I mean, I don’t know what he said already, but, the sheet, I heard.
28 00:02:18.770 ⇒ 00:02:19.380 Casie Aviles: Sure.
29 00:02:19.950 ⇒ 00:02:22.590 Samuel Roberts: We went through a bunch of the…
30 00:02:24.000 ⇒ 00:02:29.169 Samuel Roberts: Bunch of the thumbs downs, figuring out what was wrong and classifying them.
31 00:02:29.350 ⇒ 00:02:30.700 Samuel Roberts: Best as we could.
32 00:02:31.950 ⇒ 00:02:35.420 Samuel Roberts: So, like, trying to use some error codes.
33 00:02:35.970 ⇒ 00:02:37.680 Samuel Roberts: And identify, like, is it…
34 00:02:38.360 ⇒ 00:02:42.109 Samuel Roberts: Is it a problem with the agent? Is it a problem with the data? What is it?
35 00:02:42.110 ⇒ 00:02:42.810 Casie Aviles: Okay.
36 00:02:43.010 ⇒ 00:02:45.140 Samuel Roberts: Can’t remember what some other ones are. And so…
37 00:02:45.710 ⇒ 00:02:50.210 Samuel Roberts: There’s a lot, I think we were finding that, like, the databases, they’re expecting different names.
38 00:02:50.340 ⇒ 00:02:56.240 Samuel Roberts: And I’m not sure people have been updating it, and I couldn’t remember where we ended up with the… them using the… the forms.
39 00:02:56.670 ⇒ 00:02:58.160 Samuel Roberts: But.
40 00:02:59.990 ⇒ 00:03:04.810 Samuel Roberts: Amber had mentioned there had been a problem, and I couldn’t remember exactly where that was.
41 00:03:05.320 ⇒ 00:03:06.800 Samuel Roberts: When that was canceled.
42 00:03:07.240 ⇒ 00:03:15.199 Casie Aviles: We did update, like, the forms, but I think… Jenny’s becomes the bottleneck there.
43 00:03:15.360 ⇒ 00:03:20.370 Casie Aviles: Since she has to update Everything… Okay.
44 00:03:20.370 ⇒ 00:03:21.519 Samuel Roberts: Yeah.
45 00:03:22.020 ⇒ 00:03:25.280 Casie Aviles: But also, there’s also the piece where…
46 00:03:25.900 ⇒ 00:03:30.020 Casie Aviles: The forms are, you know, sometimes there are errors as well.
47 00:03:30.510 ⇒ 00:03:32.100 Casie Aviles: With the forms.
48 00:03:32.600 ⇒ 00:03:39.930 Casie Aviles: That’s what I’ve… I think that’s what I last worked on before I was out, yeah.
49 00:03:40.450 ⇒ 00:03:42.650 Samuel Roberts: Okay, yeah, because I had looked at, like.
50 00:03:42.770 ⇒ 00:03:46.920 Samuel Roberts: the logs for certain people, or right, there were certain people that they said, like.
51 00:03:47.220 ⇒ 00:03:55.440 Samuel Roberts: hadn’t… or aren’t in that department anymore, or aren’t with the company anymore, and I went looking, and I didn’t even see any logs of them trying to change that, so I wasn’t sure if that was…
52 00:03:57.020 ⇒ 00:04:05.759 Samuel Roberts: that they hadn’t changed it, or that it hadn’t gone through, and so I wasn’t sure. But I think that’s something we need to hash out with them a bit better.
53 00:04:06.000 ⇒ 00:04:07.320 Casie Aviles: Yes. Okay.
54 00:04:07.970 ⇒ 00:04:16.340 Samuel Roberts: But I’m also… well, I don’t know. There’s a few other things we need to sort of sort out, and I think we’re also obviously working on migrating, master stuff
55 00:04:16.540 ⇒ 00:04:21.779 Samuel Roberts: But I don’t want to just do that right before, like, end of the year holiday stuff.
56 00:04:22.390 ⇒ 00:04:23.820 Samuel Roberts: And I want.
57 00:04:23.820 ⇒ 00:04:24.200 Casie Aviles: to be.
58 00:04:25.330 ⇒ 00:04:29.089 Samuel Roberts: I want us to know that it’s as good, if not better, as is.
59 00:04:29.260 ⇒ 00:04:34.370 Samuel Roberts: And so, the other thing is probably putting together, like, a… evaluation set.
60 00:04:35.290 ⇒ 00:04:43.000 Samuel Roberts: That we can test against. But… We also want to…
61 00:04:43.640 ⇒ 00:04:47.790 Samuel Roberts: I think Utam said, like, more brainpower going to fixing the N8in.
62 00:04:48.010 ⇒ 00:04:50.409 Samuel Roberts: The current flow, basically, while…
63 00:04:51.550 ⇒ 00:04:53.080 Samuel Roberts: Just in the next, you know.
64 00:04:53.290 ⇒ 00:04:55.690 Samuel Roberts: But just under a week that we have.
65 00:04:57.270 ⇒ 00:05:02.410 Samuel Roberts: Because we want to show progress there, rather than just say, like, yeah, we’re getting there with the migration, kind of thing.
66 00:05:05.660 ⇒ 00:05:06.250 Casie Aviles: Okay.
67 00:05:06.250 ⇒ 00:05:12.570 Samuel Roberts: So I think, yeah, those errors, and we want to start resolving those ones. That’s really the main…
68 00:05:13.790 ⇒ 00:05:15.850 Samuel Roberts: focus, I guess.
69 00:05:17.340 ⇒ 00:05:18.000 Samuel Roberts: Yeah.
70 00:05:19.310 ⇒ 00:05:20.689 Casie Aviles: Okay, great.
71 00:05:20.870 ⇒ 00:05:29.580 Casie Aviles: Yeah, so I guess there I can help with that, with, you know, fix, yeah, with the N8 inside.
72 00:05:30.300 ⇒ 00:05:39.140 Casie Aviles: Yeah, I saw that there are a bunch of updates as well in the repository, so…
73 00:05:39.650 ⇒ 00:05:47.029 Samuel Roberts: Yeah, yeah, we, trying to think. We moved the, actually, the DBQuery tool,
74 00:05:47.280 ⇒ 00:05:53.669 Samuel Roberts: off of Windmill, or just dropped in a new replacement, and that’s running on Heroku with Mastra.
75 00:05:55.160 ⇒ 00:05:56.120 Casie Aviles: Okay, nice.
76 00:05:57.270 ⇒ 00:05:58.720 Samuel Roberts: We were hitting…
77 00:05:59.560 ⇒ 00:06:03.079 Samuel Roberts: We had a few errors, but we’re also, like, the time that it takes for it to run.
78 00:06:03.530 ⇒ 00:06:08.609 Samuel Roberts: So we made a few tweaks to that prompt, I think? Is that right, Mustafa?
79 00:06:09.590 ⇒ 00:06:10.850 Samuel Roberts: There was something…
80 00:06:11.560 ⇒ 00:06:22.750 Mustafa Raja: Yes, we did make a tweak. So, what it would do is it would, so if it’s looking in… if it was looking into multiple services, what happened was it was trying to.
81 00:06:22.750 ⇒ 00:06:23.150 Samuel Roberts: God.
82 00:06:23.150 ⇒ 00:06:29.780 Mustafa Raja: match the name, and there we asked it to rather use, I like with an array.
83 00:06:32.550 ⇒ 00:06:34.840 Samuel Roberts: They were asking for something like, you know.
84 00:06:34.960 ⇒ 00:06:36.450 Samuel Roberts: Or even the one that I think.
85 00:06:36.450 ⇒ 00:06:37.570 Mustafa Raja: services, like.
86 00:06:37.570 ⇒ 00:06:38.590 Samuel Roberts: Yeah.
87 00:06:38.590 ⇒ 00:06:41.980 Mustafa Raja: And then there’s a lot, so it would name them all, you know?
88 00:06:44.790 ⇒ 00:06:48.570 Samuel Roberts: Yeah, so we’d switch that over to make it a little more,
89 00:06:49.590 ⇒ 00:06:55.170 Samuel Roberts: Yeah, the I like instead of the fixer, because that way the names don’t have to match exactly.
90 00:06:55.490 ⇒ 00:07:03.400 Samuel Roberts: But yeah, then moving it to Heroku also, for the most part, was an improvement. We did notice an error yesterday where pulling from LangFuse gave a timeout.
91 00:07:04.160 ⇒ 00:07:04.860 Casie Aviles: Hmm.
92 00:07:05.090 ⇒ 00:07:10.360 Samuel Roberts: So I’m wondering if we want to stick with languages there, or just…
93 00:07:11.040 ⇒ 00:07:14.499 Samuel Roberts: put it in a Markdown file for now, to make sure that doesn’t happen again.
94 00:07:15.190 ⇒ 00:07:22.580 Samuel Roberts: But yeah, besides that, it’s… it’s much faster, because I think we…
95 00:07:22.740 ⇒ 00:07:25.039 Samuel Roberts: The windmill had cold starts, right?
96 00:07:26.290 ⇒ 00:07:26.800 Mustafa Raja: No, it’s no…
97 00:07:26.800 ⇒ 00:07:28.530 Samuel Roberts: the actual, like, queries.
98 00:07:28.530 ⇒ 00:07:31.120 Mustafa Raja: Who now returns in, like, 2 seconds.
99 00:07:33.550 ⇒ 00:07:35.680 Samuel Roberts: Yeah, it’s pretty… it’s pretty good.
100 00:07:35.680 ⇒ 00:07:37.110 Casie Aviles: Okay, that’s nice.
101 00:07:37.370 ⇒ 00:07:42.630 Samuel Roberts: Yeah, yeah, so it wasn’t anything with the code, necessarily, it was just windmills spinning up, I think. It took forever.
102 00:07:43.090 ⇒ 00:07:48.240 Samuel Roberts: And now that it’s just running on Heroku, it’s just… it’s just always up, which is… Sign.
103 00:07:49.000 ⇒ 00:07:49.670 Casie Aviles: Boy.
104 00:07:50.070 ⇒ 00:07:51.160 Samuel Roberts: Sorry, yeah.
105 00:07:51.160 ⇒ 00:07:52.369 Casie Aviles: Hey, hey, Gabe.
106 00:07:54.920 ⇒ 00:08:03.729 Gabriel Lam: I think this is the same call for the AI stand-up, but I wasn’t sure, because it’s not the same one as the Google Calendar one.
107 00:08:05.300 ⇒ 00:08:06.519 Gabriel Lam: But I believe this is the one.
108 00:08:06.520 ⇒ 00:08:06.920 Samuel Roberts: Really?
109 00:08:06.920 ⇒ 00:08:08.440 Gabriel Lam: Correct. Yes.
110 00:08:08.440 ⇒ 00:08:10.490 Samuel Roberts: Wait, what’s on the Google Calendar?
111 00:08:10.760 ⇒ 00:08:14.059 Gabriel Lam: Let me bring it up.
112 00:08:17.330 ⇒ 00:08:19.370 Gabriel Lam: Oh, no, it is the same. Okay, that’s weird.
113 00:08:19.370 ⇒ 00:08:25.559 Samuel Roberts: Okay, yeah, because I just clicked the, like, the pop-up that came up on my… so I actually wasn’t sure where that even came from.
114 00:08:25.560 ⇒ 00:08:31.260 Gabriel Lam: Yeah, yesterday I was, like, on the call, and no one was on the call, and I was like, wait, that’s… not right.
115 00:08:31.540 ⇒ 00:08:38.170 Samuel Roberts: Oh, okay. Yeah, there was also a problem the other day when Tom transferred it, there was no meeting that got lost or something, and so I had to add one, but…
116 00:08:38.490 ⇒ 00:08:39.970 Samuel Roberts: Yeah.
117 00:08:40.820 ⇒ 00:08:43.700 Samuel Roberts: We’re just catching Casey up on ABC stuff.
118 00:08:43.700 ⇒ 00:08:44.260 Gabriel Lam: Okay.
119 00:08:44.260 ⇒ 00:08:47.040 Samuel Roberts: Which has been eventful, so…
120 00:08:50.450 ⇒ 00:08:53.380 Samuel Roberts: I lost my train of thought there, but…
121 00:08:53.380 ⇒ 00:08:57.839 Casie Aviles: Yeah, I was just going to ask, like, if we…
122 00:08:58.100 ⇒ 00:09:03.249 Casie Aviles: If we swapped out, like, the model for the… The query agent, or…
123 00:09:03.430 ⇒ 00:09:07.609 Casie Aviles: Was it still 4.0? Or 4.0? I get… yeah.
124 00:09:08.030 ⇒ 00:09:22.019 Samuel Roberts: Yeah, yeah, we went back to 4.0. I think… I think you had initially tried 5.2 there, or… that’s what we… but we added 4.0 just to make sure it’s on, like, the same thing, but it is definitely worth investigating. I just… I want us to have, like, a testing harness set up before we start.
125 00:09:22.290 ⇒ 00:09:29.979 Samuel Roberts: doing that. Because, you know, models are different and wildly different, and so I don’t want to…
126 00:09:29.980 ⇒ 00:09:30.600 Casie Aviles: Yeah.
127 00:09:30.910 ⇒ 00:09:35.970 Samuel Roberts: Tweak too much, but… The other side of this is we could probably use, like.
128 00:09:36.280 ⇒ 00:09:39.780 Samuel Roberts: Well, actually, I don’t know if it matters. Different models for different pieces of it, but…
129 00:09:41.330 ⇒ 00:09:42.570 Samuel Roberts: I think it’s fine for now.
130 00:09:42.780 ⇒ 00:09:48.700 Samuel Roberts: But yeah, the main priority there is getting these bugs fixed, and some of those bugs are…
131 00:09:49.330 ⇒ 00:09:51.550 Samuel Roberts: You know, data quality, which is…
132 00:09:52.020 ⇒ 00:09:58.839 Samuel Roberts: somewhat on us for making sure that they keep up with it, but also they just need to keep up with it. Some things we notice is just,
133 00:09:59.420 ⇒ 00:10:00.830 Samuel Roberts: the Central Docs…
134 00:10:01.910 ⇒ 00:10:08.780 Samuel Roberts: outdated. And so people are like, oh, that’s not right anymore, this guy doesn’t work here. I was like, well, he’s the one in the dock, I don’t know.
135 00:10:08.780 ⇒ 00:10:09.710 Casie Aviles: I see.
136 00:10:09.710 ⇒ 00:10:18.080 Samuel Roberts: And then we noticed one today where there’s a referral section that it didn’t know was a referral, and I think we just need to update the way the RAG does some metadata there.
137 00:10:19.890 ⇒ 00:10:20.680 Samuel Roberts: But…
138 00:10:21.090 ⇒ 00:10:23.159 Mustafa Raja: Yeah. That was just literally this morning, so…
139 00:10:24.980 ⇒ 00:10:29.999 Casie Aviles: Cool. So, yeah, the priority would be the zip codes. Okay, so…
140 00:10:32.100 ⇒ 00:10:41.699 Casie Aviles: So, I’m not sure if you guys already answered this, but what else are, like, the open… or, like, what did you resolve… what was already resolved, and what’s…
141 00:10:42.370 ⇒ 00:10:44.510 Samuel Roberts: That spreadsheet should have,
142 00:10:45.060 ⇒ 00:10:52.320 Samuel Roberts: I didn’t… I had to sign up a little bit before they finished going through it, but basically, we want to make sure that
143 00:10:52.650 ⇒ 00:10:53.420 Samuel Roberts: Sorry?
144 00:10:53.570 ⇒ 00:10:56.429 Mustafa Raja: We have an, isFixed column in there.
145 00:10:56.620 ⇒ 00:10:57.740 Mustafa Raja: So…
146 00:10:57.980 ⇒ 00:11:01.500 Casie Aviles: Oh, okay. I can see it now. Yeah, I can see it now.
147 00:11:02.940 ⇒ 00:11:11.889 Samuel Roberts: We also… oh, one thing we did, we, so this was all pulled from the Snowflake, and we just went through and found the last… initially the last 2 months.
148 00:11:12.200 ⇒ 00:11:16.950 Samuel Roberts: of or since the beginning of November, thumbs down.
149 00:11:17.080 ⇒ 00:11:19.060 Samuel Roberts: responses,
150 00:11:20.290 ⇒ 00:11:24.999 Samuel Roberts: But then we try to match it up with the N8N executions, which only stores the last month.
151 00:11:25.260 ⇒ 00:11:31.860 Samuel Roberts: And the way we had to do that was by the timestamps, which we extracted from the…
152 00:11:32.260 ⇒ 00:11:36.840 Samuel Roberts: record ID, or whatever it was, that had the… The date in it?
153 00:11:36.960 ⇒ 00:11:43.720 Samuel Roberts: But we have now added the execution ID to the Snowflake, so hopefully that will make it…
154 00:11:44.120 ⇒ 00:11:46.629 Samuel Roberts: Easier to do this going forward.
155 00:11:46.630 ⇒ 00:11:47.240 Casie Aviles: Okay.
156 00:11:47.380 ⇒ 00:11:49.740 Samuel Roberts: The idea is also maybe to do… add some…
157 00:11:49.970 ⇒ 00:11:55.810 Samuel Roberts: you know, every time there’s a thumbs down, have some AI take a look at it, maybe build a little tool that can
158 00:11:56.640 ⇒ 00:12:00.430 Samuel Roberts: I don’t know, pull the edit and execution, maybe, and just see, like.
159 00:12:00.770 ⇒ 00:12:06.070 Samuel Roberts: Why it’s not right, and give us a first pass at it. Classifying it, maybe?
160 00:12:06.760 ⇒ 00:12:08.530 Samuel Roberts: You know, a lot of this is…
161 00:12:08.530 ⇒ 00:12:09.220 Casie Aviles: Yeah.
162 00:12:09.220 ⇒ 00:12:12.099 Samuel Roberts: Making sure that they see that we’re on top of it.
163 00:12:13.010 ⇒ 00:12:15.959 Samuel Roberts: So, but yeah, a lot of the…
164 00:12:16.730 ⇒ 00:12:21.840 Samuel Roberts: I mean, if you look at the spreadsheet, there’s a lot that’s, like, them and us, so it’s pretty split, but…
165 00:12:23.010 ⇒ 00:12:24.359 Casie Aviles: I see, yeah, yeah.
166 00:12:25.260 ⇒ 00:12:30.689 Casie Aviles: Okay, so I’ll work on the remaining ones that have not been resolved yet.
167 00:12:31.300 ⇒ 00:12:33.180 Samuel Roberts: Yeah, I think,
168 00:12:36.140 ⇒ 00:12:41.579 Samuel Roberts: At least the ones we… we have that are, yeah, that are our own… that we’re owners of.
169 00:12:42.690 ⇒ 00:12:43.340 Casie Aviles: Okay.
170 00:12:44.900 ⇒ 00:12:46.739 Samuel Roberts: We gotta… I gotta think about…
171 00:12:47.580 ⇒ 00:12:51.649 Samuel Roberts: updating the database for them more, or… or maybe… I don’t know.
172 00:12:53.360 ⇒ 00:12:54.920 Samuel Roberts: Do we know… so, okay.
173 00:12:55.120 ⇒ 00:13:03.699 Samuel Roberts: This might be a dumb question, but I wasn’t fully in the loop when we started doing the database stuff for the zips, but that all came from the spreadsheets, right?
174 00:13:04.050 ⇒ 00:13:05.070 Samuel Roberts: Initially.
175 00:13:05.420 ⇒ 00:13:13.590 Casie Aviles: Yes, so… what I think, yeah, what we had before was just the…
176 00:13:14.660 ⇒ 00:13:21.800 Casie Aviles: The pest… it was just a few… a couple of… it was just the inspectors, and then just the…
177 00:13:22.910 ⇒ 00:13:31.809 Casie Aviles: what else, technicians, so skill… skill… I forgot what it’s called. Skills in Zips, pest division, yeah.
178 00:13:32.310 ⇒ 00:13:33.580 Casie Aviles: It’s just those.
179 00:13:33.790 ⇒ 00:13:35.619 Casie Aviles: And then the service areas.
180 00:13:35.920 ⇒ 00:13:39.709 Casie Aviles: And then, of course, eventually, they’ve been adding…
181 00:13:40.080 ⇒ 00:13:42.509 Samuel Roberts: Right. More and more…
182 00:13:42.740 ⇒ 00:13:47.100 Casie Aviles: Inspectors, more… Technicians and
183 00:13:47.400 ⇒ 00:13:51.530 Casie Aviles: I think, yeah, they’re also up… so they’re kind of using…
184 00:13:52.230 ⇒ 00:13:55.840 Casie Aviles: They’re still using those spreadsheets, but…
185 00:13:55.840 ⇒ 00:13:58.969 Samuel Roberts: That’s… okay, that’s what I was kind of wondering, what is their source of truth?
186 00:13:59.290 ⇒ 00:14:05.530 Casie Aviles: Yeah, so, and whenever it gets updated, it doesn’t sync directly to… those… Yeah.
187 00:14:05.530 ⇒ 00:14:06.240 Samuel Roberts: Okay.
188 00:14:06.240 ⇒ 00:14:08.350 Casie Aviles: the database in Superbase.
189 00:14:08.350 ⇒ 00:14:14.000 Samuel Roberts: Right, they have to go update both now, I guess, which is… Alright. Yeah.
190 00:14:14.870 ⇒ 00:14:15.410 Samuel Roberts: Huh.
191 00:14:16.150 ⇒ 00:14:17.280 Samuel Roberts: I’ll put some bot into that then.
192 00:14:17.280 ⇒ 00:14:21.319 Casie Aviles: That… that’s… that’s why we did, like, the forms, in order to update.
193 00:14:21.320 ⇒ 00:14:22.280 Samuel Roberts: Yeah.
194 00:14:22.960 ⇒ 00:14:24.230 Samuel Roberts: But, I think… Right.
195 00:14:24.410 ⇒ 00:14:34.410 Casie Aviles: The reason why that also… wasn’t able to, like, update it as efficiently, I guess, with…
196 00:14:34.530 ⇒ 00:14:37.890 Casie Aviles: You know, even the form itself had some issues, or…
197 00:14:38.700 ⇒ 00:14:45.509 Casie Aviles: Yeah, sometimes they did not use it. Sometimes they don’t use it at all, and it gets… Yeah, that’s…
198 00:14:45.900 ⇒ 00:14:47.940 Samuel Roberts: That’s what it seems to be. Okay.
199 00:14:49.470 ⇒ 00:14:56.479 Samuel Roberts: Alright, yeah, I would say, yeah, so yeah, priority being the ones we have
200 00:14:56.800 ⇒ 00:15:03.260 Samuel Roberts: ownership of, I will give some thought to that. I’m wondering if now that we have this thing hosted on Heroku anyway for now.
201 00:15:03.450 ⇒ 00:15:08.230 Samuel Roberts: Maybe we give them another UI for updating or something? Or we just…
202 00:15:08.410 ⇒ 00:15:12.580 Samuel Roberts: Do some diff with the spreadsheet if they are keeping that up to date.
203 00:15:12.790 ⇒ 00:15:16.600 Samuel Roberts: But we can… we can brainstorm that a little bit more.
204 00:15:17.980 ⇒ 00:15:22.209 Samuel Roberts: Okay, so for this week, yeah, I’ll just work on debugs.
205 00:15:22.930 ⇒ 00:15:23.370 Samuel Roberts: Yeah.
206 00:15:23.370 ⇒ 00:15:24.670 Mustafa Raja: That we offer, yeah.
207 00:15:25.140 ⇒ 00:15:29.640 Mustafa Raja: Okay, so this would be for them to be able to update the data in database.
208 00:15:31.450 ⇒ 00:15:32.860 Casie Aviles: Yeah, the forms.
209 00:15:32.860 ⇒ 00:15:33.260 Mustafa Raja: Okay.
210 00:15:33.260 ⇒ 00:15:39.970 Samuel Roberts: Yeah, so right now we have the forms, because it was all NA to N, and, you know, I was hoping that as part of more scope, there might be another
211 00:15:40.200 ⇒ 00:15:50.510 Samuel Roberts: chat tool that we were building and a different UI, then we could add that. But I’m now starting to think that, like, if they’re not using the forms, I don’t know if it’s because they’re cumbersome, or they forget, or…
212 00:15:50.510 ⇒ 00:15:52.400 Mustafa Raja: You know, what the reasoning is.
213 00:15:52.400 ⇒ 00:16:02.540 Samuel Roberts: I’m wondering if we just… Either try to diff with the… with the… Spreadsheet.
214 00:16:02.640 ⇒ 00:16:06.190 Samuel Roberts: Or… I don’t know. Or we give them another.
215 00:16:06.190 ⇒ 00:16:12.280 Mustafa Raja: Oh, okay, okay. So you’re thinking maybe they could update the spreadsheet, and we could listen from there?
216 00:16:12.890 ⇒ 00:16:14.940 Samuel Roberts: Maybe, I don’t love that.
217 00:16:14.940 ⇒ 00:16:15.460 Mustafa Raja: Me too.
218 00:16:15.460 ⇒ 00:16:18.739 Casie Aviles: Yeah. But if that’s… I just want, like…
219 00:16:18.850 ⇒ 00:16:23.100 Samuel Roberts: I don’t like them having to maintain two things, because inevitably they’ll fall out of sync.
220 00:16:23.800 ⇒ 00:16:24.450 Mustafa Raja: Yeah.
221 00:16:25.330 ⇒ 00:16:30.559 Samuel Roberts: And so either that, or… We give them, like, a better…
222 00:16:30.740 ⇒ 00:16:36.920 Samuel Roberts: UI for editing the database, where they can maybe, you know, more…
223 00:16:37.050 ⇒ 00:16:40.580 Samuel Roberts: Easily change things without it being the form that has to do it.
224 00:16:40.690 ⇒ 00:16:47.960 Samuel Roberts: like a cred UI or something, I don’t know, that’s something I think we need to hash out with Janiece and see, like, what fits her flow better.
225 00:16:47.960 ⇒ 00:16:50.419 Mustafa Raja: forms are very basic… Exactly.
226 00:16:50.840 ⇒ 00:16:53.120 Mustafa Raja: Or look into other forms.
227 00:16:54.680 ⇒ 00:16:56.110 Samuel Roberts: Yeah, yeah, we could.
228 00:16:56.500 ⇒ 00:17:02.419 Samuel Roberts: Yeah, I’m not sure if that’s why things are falling out of sync, or if…
229 00:17:02.840 ⇒ 00:17:05.079 Samuel Roberts: But yeah, that’s… that’s the big question there, so…
230 00:17:05.400 ⇒ 00:17:05.929 Casie Aviles: Yeah, because.
231 00:17:05.930 ⇒ 00:17:06.500 Samuel Roberts: Hey, Tom.
232 00:17:07.160 ⇒ 00:17:08.069 Uttam Kumaran: Hey guys.
233 00:17:08.859 ⇒ 00:17:10.410 Uttam Kumaran: Hey Casey, welcome back.
234 00:17:11.250 ⇒ 00:17:12.839 Casie Aviles: Hey guys, yeah.
235 00:17:13.589 ⇒ 00:17:19.129 Uttam Kumaran: Pranav is, I think Pranav is out the next, 2 days, guys, by the way, so…
236 00:17:19.130 ⇒ 00:17:19.839 Samuel Roberts: Yeah.
237 00:17:20.680 ⇒ 00:17:21.300 Uttam Kumaran: Cool.
238 00:17:22.599 ⇒ 00:17:24.809 Samuel Roberts: Yeah, I don’t know if you guys have… interact with.
239 00:17:24.810 ⇒ 00:17:28.669 Uttam Kumaran: I don’t know if they met Pranav, but Pranav was working on the Lilo client, by the way, so… Yeah.
240 00:17:32.060 ⇒ 00:17:33.639 Samuel Roberts: Yeah.
241 00:17:33.640 ⇒ 00:17:39.030 Uttam Kumaran: Cool, yeah, I would love to hear about, like, ABC plans for today.
242 00:17:39.210 ⇒ 00:17:43.270 Uttam Kumaran: And then we can talk about… Talk about Lilo.
243 00:17:43.600 ⇒ 00:17:48.699 Samuel Roberts: Yeah. Yeah, we were just kind of catching Casey up and showing him the spreadsheet,
244 00:17:48.910 ⇒ 00:17:57.140 Samuel Roberts: And basically making that the priority to fix the… at least the ones that we’re the owner for there, and then, that’s what we started talking about, like, the ones that…
245 00:17:57.380 ⇒ 00:18:04.100 Samuel Roberts: ABC, would be the… the owner for, like, what is the best way to…
246 00:18:04.880 ⇒ 00:18:11.960 Samuel Roberts: One, fix those, and then mitigate it ongoing, so that, you know, if they’re maintaining two.
247 00:18:12.170 ⇒ 00:18:16.899 Samuel Roberts: If they’re still maintaining that spreadsheet that was the original thing we were using to do the lookups.
248 00:18:17.750 ⇒ 00:18:21.140 Samuel Roberts: You know, now they’re maintaining two different sources of truth, which is…
249 00:18:21.540 ⇒ 00:18:23.999 Samuel Roberts: inevitably, like I said, gonna fall out of sync.
250 00:18:24.130 ⇒ 00:18:26.080 Samuel Roberts: But…
251 00:18:27.290 ⇒ 00:18:34.110 Samuel Roberts: So then, we were just brainstorming a few other ways, maybe the forms are too cumbersome, maybe it’s just, you know, is it worth trying to
252 00:18:34.240 ⇒ 00:18:39.139 Samuel Roberts: read from that sheet still, and see what’s been changed, what’s updated.
253 00:18:40.290 ⇒ 00:18:46.760 Samuel Roberts: I think, realistically, the conversation needs to involve Denise if she’s the bottleneck for doing these updates right now.
254 00:18:48.960 ⇒ 00:18:49.630 Casie Aviles: Yeah.
255 00:18:50.480 ⇒ 00:18:55.429 Samuel Roberts: But, yeah. Yeah, besides that, I was just kind of pointing Casey to the spreadsheet and the ones that…
256 00:18:55.560 ⇒ 00:18:59.680 Samuel Roberts: We’re the owner of, and… Starting to check those boxes.
257 00:19:00.340 ⇒ 00:19:01.380 Uttam Kumaran: Okay, great.
258 00:19:04.070 ⇒ 00:19:11.330 Samuel Roberts: Yeah… I’m trying to think… I worked… I don’t know, Mustafa, if you ended up work…
259 00:19:11.720 ⇒ 00:19:13.840 Uttam Kumaran: Chatting with folks about the Gantt chart.
260 00:19:14.140 ⇒ 00:19:16.000 Mustafa Raja: Yeah, I did read some.
261 00:19:16.830 ⇒ 00:19:31.110 Samuel Roberts: Yeah, we went over a little bit, before this when we were catching up on some stuff. Yeah, it looks… it looks pretty good. I think there’s a couple things we were just talking about, like the routing and stuff, but, there was something else, and of course, I can’t remember now.
262 00:19:31.110 ⇒ 00:19:40.350 Mustafa Raja: Yeah, so, for the deadlines and for… because we want to, also, store the logs, or get observability over it.
263 00:19:40.920 ⇒ 00:19:42.949 Mustafa Raja: Which we need to figure out, I believe.
264 00:19:46.510 ⇒ 00:19:53.429 Samuel Roberts: Yeah, yeah, I mean, so, like I said, I have been testing Monster Cloud a little bit to see, you know, what it gives us out of the box, and it’s not…
265 00:19:54.220 ⇒ 00:19:58.829 Samuel Roberts: rate. It’s that… I mean, there’s a reason they’re not charging for it yet, I think. It’s still, like…
266 00:19:59.740 ⇒ 00:20:04.109 Samuel Roberts: It’s a little, rough around the edges.
267 00:20:04.630 ⇒ 00:20:05.380 Uttam Kumaran: Okay.
268 00:20:06.630 ⇒ 00:20:18.020 Samuel Roberts: But it does seem to have some of the observability stuff, so I think we just… if we’re gonna get that running on, like, our GCP, we just need to make sure that we have those things thought through and documented in the Gantt as well to make sure that
269 00:20:18.340 ⇒ 00:20:24.159 Samuel Roberts: We’re not, you know, we’re able to source or find these kinds of bugs and stuff that we’re, you know.
270 00:20:24.310 ⇒ 00:20:30.289 Samuel Roberts: able to find the N8N executions for right now, but don’t want to lose that if we jump over to code, so…
271 00:20:33.250 ⇒ 00:20:35.590 Samuel Roberts: Excuse me.
272 00:20:37.080 ⇒ 00:20:43.340 Samuel Roberts: Yeah, I definitely want to go through that a little bit more later, I just kind of looked at it quick this morning, Mustafa, but…
273 00:20:46.730 ⇒ 00:20:47.829 Samuel Roberts: Seems pretty good.
274 00:20:49.420 ⇒ 00:20:50.150 Mustafa Raja: Hmm.
275 00:20:51.170 ⇒ 00:20:58.210 Samuel Roberts: Do we know what their schedule is, ABC’s Holiday Wives? Do we… Know that?
276 00:20:58.210 ⇒ 00:21:02.890 Uttam Kumaran: I think maybe a good question to email about.
277 00:21:03.180 ⇒ 00:21:09.910 Samuel Roberts: Okay, that’s… yeah, because I… this is… I mean, I’m thinking if Amber’s out as of today, too, is that right?
278 00:21:11.440 ⇒ 00:21:13.339 Uttam Kumaran: Yes, Amber is out.
279 00:21:13.750 ⇒ 00:21:15.300 Samuel Roberts: Okay,
280 00:21:16.920 ⇒ 00:21:26.019 Samuel Roberts: because I definitely wanted to get Janiece’s input on, like, their database stuff, and figure out… because I… like I said, the two sources of truth is a little…
281 00:21:26.580 ⇒ 00:21:31.079 Samuel Roberts: not great, but if that’s, you know… I don’t want to make it too cumbersome for them if that’s what’s causing these issues.
282 00:21:31.080 ⇒ 00:21:31.840 Uttam Kumaran: Okay.
283 00:21:32.470 ⇒ 00:21:36.270 Samuel Roberts: So maybe I’ll try to reach out to her. End.
284 00:21:36.710 ⇒ 00:21:40.630 Samuel Roberts: See what her schedule’s like, at least, free Christmas.
285 00:21:42.580 ⇒ 00:21:43.590 Samuel Roberts: Okay.
286 00:21:45.120 ⇒ 00:21:48.470 Samuel Roberts: Anything else ABC-wise that we didn’t hit here?
287 00:21:49.260 ⇒ 00:21:57.540 Samuel Roberts: Mustafa and I talked a little bit about some of the alternative, like, structure for Mastra, but.
288 00:21:57.540 ⇒ 00:22:05.639 Uttam Kumaran: Did we talk about, like, the feedback from, like, having another agent do the, so, like, ultimately…
289 00:22:05.640 ⇒ 00:22:06.100 Samuel Roberts: Yes, yeah.
290 00:22:06.100 ⇒ 00:22:18.479 Uttam Kumaran: When we pitched ABC, we pitched a two-sided thing. One is the Andy agent, another thing is the agent that… another thing is an agent that helps the trainers. That way, you have this, like, great circular loop.
291 00:22:18.930 ⇒ 00:22:19.330 Samuel Roberts: Okay.
292 00:22:19.330 ⇒ 00:22:38.129 Uttam Kumaran: what’s important is that we’ve nailed the first one, we have yet… we’ve built structure for the trainers and Janiece to fix things, but we need to turn that into something AI. So, our first foray that we proposed, right, yesterday was, like, have an agent triage the error codes first.
293 00:22:38.330 ⇒ 00:22:42.409 Uttam Kumaran: Figure out the error code, go through and, like, think about what it could be.
294 00:22:43.120 ⇒ 00:22:54.610 Uttam Kumaran: I think if you, like, maybe I’ll pose the question to you guys. Yeah. I don’t want… maybe instead of giving, like… I’ll let you think about it, like, what else do you think we could do on that side of the house?
295 00:22:56.000 ⇒ 00:22:56.899 Uttam Kumaran: And think big.
296 00:22:56.900 ⇒ 00:22:58.410 Samuel Roberts: I’m just, like, triaging?
297 00:22:58.880 ⇒ 00:23:02.969 Uttam Kumaran: Beyond just, like, oh, like, what could the… what could the issue be?
298 00:23:03.590 ⇒ 00:23:04.730 Samuel Roberts: Yeah, yeah, I mean…
299 00:23:05.150 ⇒ 00:23:11.939 Samuel Roberts: I mean, there’s a lot of data that’s not just bugs in the Snowflake, too, you know, that, like, we could see, like, patterns and refine
300 00:23:12.320 ⇒ 00:23:15.869 Samuel Roberts: From there? Is that the story we’re talking about, or are you talking about…
301 00:23:16.570 ⇒ 00:23:17.989 Samuel Roberts: I mean, when you say big, I don’t really…
302 00:23:17.990 ⇒ 00:23:27.090 Uttam Kumaran: What other AI features could we develop to improve the speed at which issues get remediated, basically.
303 00:23:30.760 ⇒ 00:23:31.310 Samuel Roberts: Yeah.
304 00:23:31.920 ⇒ 00:23:36.509 Samuel Roberts: Let’s think.
305 00:23:36.530 ⇒ 00:23:44.540 Casie Aviles: Like, spotting… Spotting errors before they get to the clients? Something like that.
306 00:23:46.240 ⇒ 00:23:46.760 Casie Aviles: Is that…
307 00:23:46.760 ⇒ 00:23:49.900 Samuel Roberts: Well, I think it’s really sort of talking about the, like, the triaging the errors right now.
308 00:23:49.900 ⇒ 00:23:52.799 Uttam Kumaran: Go one, go, go, go one step further.
309 00:23:53.200 ⇒ 00:23:54.219 Samuel Roberts: How do we fix it?
310 00:23:56.070 ⇒ 00:23:58.210 Uttam Kumaran: No, no, no, even further.
311 00:23:58.210 ⇒ 00:23:59.239 Samuel Roberts: Further than that.
312 00:24:00.330 ⇒ 00:24:03.409 Mustafa Raja: Do you want to have the agent update the database?
313 00:24:03.620 ⇒ 00:24:04.360 Uttam Kumaran: Yes.
314 00:24:06.050 ⇒ 00:24:06.660 Mustafa Raja: But this.
315 00:24:06.660 ⇒ 00:24:08.869 Samuel Roberts: Oh, you want to use Andy to update itself?
316 00:24:09.420 ⇒ 00:24:10.130 Uttam Kumaran: Yes.
317 00:24:11.530 ⇒ 00:24:13.059 Uttam Kumaran: But what is itself?
318 00:24:14.500 ⇒ 00:24:21.479 Mustafa Raja: I mean, based on the feedback, we’ve seen that the feedback clearly mentions that this person does not belong here anymore, and is there.
319 00:24:21.660 ⇒ 00:24:23.849 Mustafa Raja: It’s somewhere else, right? So…
320 00:24:24.930 ⇒ 00:24:25.420 Samuel Roberts: Yeah.
321 00:24:25.420 ⇒ 00:24:28.879 Mustafa Raja: is clear enough. Only then it should, you know…
322 00:24:29.520 ⇒ 00:24:31.609 Uttam Kumaran: Sorry, I don’t mean to be… I don’t mean to be, like, pedantic.
323 00:24:31.610 ⇒ 00:24:34.850 Samuel Roberts: No, I see what you’re saying, yeah, yeah, no, I hear it, I hear it, okay.
324 00:24:34.850 ⇒ 00:24:49.330 Uttam Kumaran: what I want you guys is to dream, right? And it’s rare that as engineers we get told to dream, but I’m asking you guys to really dream, not just about having AI tell a human what to do.
325 00:24:50.090 ⇒ 00:24:52.719 Uttam Kumaran: If the AI is good enough to tell the human.
326 00:24:52.820 ⇒ 00:25:10.910 Uttam Kumaran: the AI’s probably good enough to go get 98% of the fix, and be like, do you want me to go make this fix? Think about cursor, right? So, like, think about a cursor-style experience where the AI looks the feedback, looks at the central doc, oh, it’s not here, and the person suggested, do you want me to go
327 00:25:11.120 ⇒ 00:25:14.609 Uttam Kumaran: push this. You could have a human in the loop at that point.
328 00:25:14.690 ⇒ 00:25:19.569 Samuel Roberts: Sure. But imagine a future where, like, yeah, just, like, basically self-heals. And so this is, like… right, right.
329 00:25:19.570 ⇒ 00:25:24.200 Uttam Kumaran: This is the self-healing sort of, like, concept, but yes. Yeah. That’s sort of what I’m getting at.
330 00:25:24.490 ⇒ 00:25:25.350 Samuel Roberts: I see what you’re saying.
331 00:25:29.170 ⇒ 00:25:30.090 Samuel Roberts: Okay.
332 00:25:31.760 ⇒ 00:25:33.599 Uttam Kumaran: So that would be my challenge.
333 00:25:36.520 ⇒ 00:25:37.609 Uttam Kumaran: To you guys.
334 00:25:38.060 ⇒ 00:25:38.840 Samuel Roberts: Yeah.
335 00:25:40.300 ⇒ 00:25:47.080 Samuel Roberts: Sorry, gears are turning now. Yeah, I mean, I definitely see that, especially, like, once we’re off N8M.
336 00:25:48.310 ⇒ 00:25:50.350 Samuel Roberts: It’ll be a lot easier to tool call and…
337 00:25:50.990 ⇒ 00:25:53.010 Samuel Roberts: proposed changes and human in the Loop.
338 00:25:53.710 ⇒ 00:25:56.469 Samuel Roberts: With the master framework.
339 00:25:57.620 ⇒ 00:25:59.699 Samuel Roberts: Updating the central doc means…
340 00:25:59.810 ⇒ 00:26:05.600 Samuel Roberts: What? I’m just, I’m just spitballing now, too. That’s updating a Google Doc, and then re-ragging it? Okay. Okay.
341 00:26:06.570 ⇒ 00:26:07.710 Samuel Roberts: See, yeah.
342 00:26:12.470 ⇒ 00:26:13.100 Samuel Roberts: Okay.
343 00:26:13.100 ⇒ 00:26:17.910 Uttam Kumaran: So, like, I mean, you may have to have some, like, evaluation on, like.
344 00:26:18.520 ⇒ 00:26:23.309 Uttam Kumaran: what change to make, something like that, but I want you guys to think to that level.
345 00:26:23.580 ⇒ 00:26:24.090 Samuel Roberts: Sure.
346 00:26:24.090 ⇒ 00:26:41.930 Uttam Kumaran: Because, ultimately, I am not going to expect… I don’t… I want us to spend less time on this project, right? And so, we realize that this triage thing is brutal, but as AI people, we need to realize that our job is not just to give them more things to, like, do every day, it’s like.
347 00:26:41.930 ⇒ 00:26:52.459 Uttam Kumaran: the AI, maybe 80% of the triage, the AI could probably do with the humans is just say, go do, do, do, do, do, right? So, I would like us to, like, drive towards that type of future here.
348 00:26:54.680 ⇒ 00:26:55.290 Samuel Roberts: Okay.
349 00:26:59.480 ⇒ 00:27:11.750 Uttam Kumaran: So, I think it starts with this, like, error logging thing, but for me, I would like to see the three of you guys come up with a plan to drive towards that, like, North Star. Because ultimately, we’re… we’re never gonna be able to…
350 00:27:12.080 ⇒ 00:27:21.370 Uttam Kumaran: Amber is working her best to drive usage, but simultaneously, you guys need to work to drive two things. Continue to drive, like.
351 00:27:21.520 ⇒ 00:27:28.520 Uttam Kumaran: the amount of time we’re spending on this client down. But additionally, think about this two-sided
352 00:27:28.820 ⇒ 00:27:45.560 Uttam Kumaran: this two-sided thing as something we can package and sell to many different clients, right? Like, both the user-facing and the ability to remediate issues in a much more faster cycle, that is an amazing package that we can go sell. So, I think one thing
353 00:27:45.570 ⇒ 00:27:52.070 Uttam Kumaran: you know, I would love to see from y’all is just, like, what is a… what are our options? What is, like, a roadmap on the…
354 00:27:52.490 ⇒ 00:27:54.920 Uttam Kumaran: Trainer, remediation support side.
355 00:27:57.420 ⇒ 00:27:58.090 Mustafa Raja: Yep.
356 00:28:00.540 ⇒ 00:28:01.190 Samuel Roberts: Okay.
357 00:28:02.760 ⇒ 00:28:03.770 Samuel Roberts: Okay, yeah.
358 00:28:07.070 ⇒ 00:28:11.670 Samuel Roberts: Some bot into that. Yeah, I mean, definitely the start of that, I think, is…
359 00:28:11.790 ⇒ 00:28:13.330 Samuel Roberts: What we’ve talked about with the…
360 00:28:13.630 ⇒ 00:28:17.359 Samuel Roberts: identifying the errors, obviously, like, already. But yeah, like…
361 00:28:18.100 ⇒ 00:28:24.759 Samuel Roberts: why not update the database, or propose a RAG, or propose a central doc change, if that’s… yeah, okay.
362 00:28:28.340 ⇒ 00:28:29.310 Samuel Roberts: Alright.
363 00:28:31.390 ⇒ 00:28:32.130 Samuel Roberts: Cool.
364 00:28:34.460 ⇒ 00:28:37.669 Samuel Roberts: Probably talk that… or what are we at today? Thursday? Okay.
365 00:28:39.910 ⇒ 00:28:40.680 Samuel Roberts: Alright.
366 00:28:41.030 ⇒ 00:28:45.200 Samuel Roberts: Other… other thoughts on ABC before we… Jump.
367 00:28:48.100 ⇒ 00:28:48.589 Casie Aviles: No, no.
368 00:28:48.590 ⇒ 00:28:51.090 Uttam Kumaran: Anything on… on… sorry, go ahead, Casey.
369 00:28:51.450 ⇒ 00:28:56.360 Casie Aviles: Yeah, yeah, I was just… I just wanted to reiterate that I’d be up for…
370 00:28:56.570 ⇒ 00:29:00.910 Casie Aviles: this week, at least, I’ll work on the P0 ones that we own.
371 00:29:02.190 ⇒ 00:29:02.820 Uttam Kumaran: Okay.
372 00:29:03.620 ⇒ 00:29:04.469 Casie Aviles: Yeah, that’s it.
373 00:29:04.470 ⇒ 00:29:08.220 Uttam Kumaran: Yeah, and as you fix them, feel free to send in the channel, and then…
374 00:29:08.240 ⇒ 00:29:26.409 Uttam Kumaran: we’ve basically been trying to send a message in the morning and in the evening, so I’ll go… I’ll send a message today saying we’re continuing through the P0 stuff. We’re also coming up with a little bit of a plan on, like, how to make the issue resolution faster. If there’s anything else you’d like me to highlight, let me know.
375 00:29:26.470 ⇒ 00:29:27.829 Uttam Kumaran: I can do that.
376 00:29:30.800 ⇒ 00:29:31.400 Casie Aviles: Okay.
377 00:29:31.950 ⇒ 00:29:35.709 Mustafa Raja: Yeah, let me know what would be some good action items for me to take on.
378 00:29:40.240 ⇒ 00:29:42.349 Samuel Roberts: Yeah, I mean, what are the…
379 00:29:43.450 ⇒ 00:29:49.829 Samuel Roberts: Well, we can, we can split those up, I suppose, by category. Yeah, let’s talk after.
380 00:29:49.950 ⇒ 00:29:51.609 Samuel Roberts: Or Slack after, at least.
381 00:29:52.620 ⇒ 00:29:58.130 Samuel Roberts: Okay, cool. We were talking really low, even though I was not here.
382 00:29:58.130 ⇒ 00:29:58.870 Uttam Kumaran: Yes.
383 00:30:00.720 ⇒ 00:30:01.650 Samuel Roberts: Yeah.
384 00:30:01.850 ⇒ 00:30:06.019 Samuel Roberts: Caught up with the loom, pulled the code and stuff, but haven’t gotten a chance to, like, dig in.
385 00:30:06.640 ⇒ 00:30:13.339 Samuel Roberts: I think I left Cursor running a review of it, but I haven’t stepped back to it. Looking good. They gave us a new key, I saw.
386 00:30:14.310 ⇒ 00:30:16.720 Samuel Roberts: for, fabio?
387 00:30:17.550 ⇒ 00:30:18.250 Uttam Kumaran: Yes.
388 00:30:19.100 ⇒ 00:30:22.949 Samuel Roberts: If I’m thinking with printout out, I may have to take a stab at.
389 00:30:23.320 ⇒ 00:30:26.850 Samuel Roberts: given that MCP like, integrated.
390 00:30:26.980 ⇒ 00:30:30.989 Samuel Roberts: So that we get something out by… Christmas for that, at least.
391 00:30:31.100 ⇒ 00:30:34.000 Samuel Roberts: Showing that we’re back to the rough level that they were at.
392 00:30:36.660 ⇒ 00:30:41.910 Samuel Roberts: Besides that, yeah, I mean, I haven’t dug into the code, it looks good, the room look good.
393 00:30:42.840 ⇒ 00:30:46.570 Samuel Roberts: Yeah, SDK seemed like it was exactly what was needed there, so that’s good.
394 00:30:46.710 ⇒ 00:30:54.859 Samuel Roberts: Yeah, I don’t have much more on that yet until I’ve dug in, but, without Pranava, yeah.
395 00:30:57.440 ⇒ 00:30:58.240 Uttam Kumaran: Okay.
396 00:30:58.440 ⇒ 00:31:00.499 Uttam Kumaran: Cool. Yeah, if you can just…
397 00:31:00.650 ⇒ 00:31:11.739 Uttam Kumaran: get us set up today in order to… so I would maybe try to delegate as much of the ABC stuff to Casey and Mustafa, and then as much as you can get done on Lilo.
398 00:31:11.850 ⇒ 00:31:17.569 Uttam Kumaran: And then I wanna… I… I don’t see that I’ve booked a call, so I will book a call
399 00:31:18.030 ⇒ 00:31:23.650 Uttam Kumaran: Yeah, I’m gonna… I’m gonna make this weekly.
400 00:31:25.460 ⇒ 00:31:25.960 Samuel Roberts: with Lilo?
401 00:31:26.830 ⇒ 00:31:27.540 Uttam Kumaran: Yeah.
402 00:31:27.750 ⇒ 00:31:28.650 Samuel Roberts: Yeah, okay.
403 00:31:31.380 ⇒ 00:31:32.190 Samuel Roberts: Cool.
404 00:31:32.190 ⇒ 00:31:34.980 Uttam Kumaran: So I’m gonna make it at 11.
405 00:31:44.680 ⇒ 00:31:48.279 Samuel Roberts: Oh, right, I live in your town. Okay, yes. Cool, cool, perfect.
406 00:31:51.900 ⇒ 00:32:02.659 Uttam Kumaran: So yeah, as much as we can keep pushing on Lilo this week, this week would be great, Sam, and then we’ll have our meeting with them tomorrow. As you can tell, they’re, like, pure output kind of guys.
407 00:32:02.660 ⇒ 00:32:03.290 Samuel Roberts: Yeah, so…
408 00:32:03.290 ⇒ 00:32:11.260 Uttam Kumaran: I’m not gonna do… I’m not gonna do decks. I can always buy us time to talk through Gantt charts and PRDs.
409 00:32:11.370 ⇒ 00:32:16.529 Uttam Kumaran: Okay. But I would… but they’re just… it’s just not gonna win them over.
410 00:32:16.890 ⇒ 00:32:17.580 Samuel Roberts: Yeah, yeah.
411 00:32:17.580 ⇒ 00:32:27.170 Uttam Kumaran: gonna win them over is, like, a shit ton of output. What’s also gonna win for us, in terms of long-term, is getting them to think about more shit they want, so we can… we can keep building it for them.
412 00:32:27.170 ⇒ 00:32:27.750 Samuel Roberts: Right, right.
413 00:32:28.250 ⇒ 00:32:35.020 Uttam Kumaran: So, I would say, if I can give you one thing today, is to try to spend time on Lilo and see if you can
414 00:32:35.340 ⇒ 00:32:40.230 Uttam Kumaran: send comms in that channel, and sort of, like, yeah, I think they’re really enjoying seeing that, so…
415 00:32:40.230 ⇒ 00:32:43.810 Samuel Roberts: Okay. Yeah, that was… that was my plan with Pranabout. He…
416 00:32:43.810 ⇒ 00:32:44.370 Uttam Kumaran: Cool.
417 00:32:44.780 ⇒ 00:32:54.710 Samuel Roberts: we… we… he… I don’t know if he grabbed time with me or not. We were talking about he was gonna just kind of brain dump to me this morning, just so I could catch up, but I don’t know if that’s happening or not with him out.
418 00:32:55.350 ⇒ 00:33:01.329 Samuel Roberts: But yeah, I’m gonna dig through the code, and that was exactly my plan, was to make progress there without him, so…
419 00:33:01.330 ⇒ 00:33:02.230 Uttam Kumaran: Okay, okay.
420 00:33:02.650 ⇒ 00:33:03.450 Samuel Roberts: Cool.
421 00:33:04.420 ⇒ 00:33:07.860 Samuel Roberts: I guess then internal, Gabe?
422 00:33:08.770 ⇒ 00:33:15.610 Gabriel Lam: Yeah, there’s a couple things. I… have…
423 00:33:15.970 ⇒ 00:33:18.759 Gabriel Lam: Let me bring it up first.
424 00:33:20.420 ⇒ 00:33:30.779 Gabriel Lam: So I pushed some changes to the playbook, and I added a new repo. I just called it a vault for now, I think we can change the name later, to store…
425 00:33:30.980 ⇒ 00:33:32.630 Gabriel Lam: all the…
426 00:33:34.630 ⇒ 00:33:40.260 Gabriel Lam: transcripts, summaries, and that kind of stuff. I’m just gonna… I’m gonna try to share screen, which I hope will make it a little clearer.
427 00:33:40.610 ⇒ 00:33:42.580 Samuel Roberts: One second…
428 00:33:51.870 ⇒ 00:33:53.140 Gabriel Lam: So…
429 00:33:55.320 ⇒ 00:34:04.400 Gabriel Lam: I’ve been working on, basically, how to get people using Cursor and using GitHub, and one of the things that I’ve been trying to add is
430 00:34:04.900 ⇒ 00:34:12.729 Gabriel Lam: where did it go? This AgentsMD file, which is basically allowing… Our agents to find…
431 00:34:13.010 ⇒ 00:34:18.179 Gabriel Lam: where prompts are. And so having the playbook be sort of the place in which
432 00:34:18.350 ⇒ 00:34:26.060 Gabriel Lam: all our standard prompts for emails, summaries, PRDs, SOPs, all that to live here. And at the same time, it can reference
433 00:34:26.440 ⇒ 00:34:33.000 Gabriel Lam: raw documentation, whether it’s, like, transcripts, notes, that we already have saved, in our platform, and then Superbase.
434 00:34:33.400 ⇒ 00:34:37.710 Gabriel Lam: And then to then throw them into client repos. And so, I think…
435 00:34:37.719 ⇒ 00:34:38.089 Samuel Roberts: Okay.
436 00:34:38.090 ⇒ 00:34:40.120 Uttam Kumaran: Wait, so what’s the… what’s the vault?
437 00:34:40.889 ⇒ 00:34:42.839 Gabriel Lam: So, the vault, I…
438 00:34:43.069 ⇒ 00:34:48.079 Gabriel Lam: have been trying to think of where we can store all our internal documentation in GitHub.
439 00:34:48.279 ⇒ 00:34:51.559 Gabriel Lam: as in, like, all… we have everything in Supabase, and I don’t know…
440 00:34:51.560 ⇒ 00:34:52.130 Uttam Kumaran: Oh.
441 00:34:52.130 ⇒ 00:34:55.480 Gabriel Lam: cursor can… access our superbase.
442 00:34:55.489 ⇒ 00:34:56.129 Samuel Roberts: point.
443 00:34:56.130 ⇒ 00:34:57.110 Gabriel Lam: tables?
444 00:34:57.350 ⇒ 00:35:04.350 Gabriel Lam: So, the thought was like, hey, if we already have them saved, is there a way to then push them into GitHub, into some sort of repository that…
445 00:35:04.800 ⇒ 00:35:12.280 Gabriel Lam: via, like, cursor and GitHub cloning that we can then Use as context, right?
446 00:35:13.350 ⇒ 00:35:20.680 Uttam Kumaran: I mean, I guess, like, my… yeah, I guess… well, how does everyone else feel? Like, should we put this all in one repo, or have them separate?
447 00:35:23.900 ⇒ 00:35:24.760 Samuel Roberts: The transcriptions.
448 00:35:24.760 ⇒ 00:35:25.320 Gabriel Lam: Bye.
449 00:35:25.430 ⇒ 00:35:40.409 Gabriel Lam: Yeah, I think the main thing we talked about yesterday was whether to save… A, save client-specific transcripts into client repos, and B, if it’s, like, a stand-up meeting that crosses multiple clients, where that sits. And then the last thing is.
450 00:35:41.670 ⇒ 00:35:49.149 Gabriel Lam: like, clients don’t really need to see… I mean, yes, we have… all our repos are private, but clients don’t really care to see their transcripts, I don’t think.
451 00:35:49.840 ⇒ 00:35:50.230 Samuel Roberts: Right.
452 00:35:50.230 ⇒ 00:35:55.459 Uttam Kumaran: Yeah, like, they don’t… they don’t really care about… they care about the outcomes from that, so…
453 00:35:55.990 ⇒ 00:36:01.079 Uttam Kumaran: I guess my… it’s more of an ergonomic question, like, should we just have everything into, like.
454 00:36:01.410 ⇒ 00:36:04.380 Uttam Kumaran: Should we just have a Brainforge vault?
455 00:36:04.830 ⇒ 00:36:13.500 Uttam Kumaran: repo that has all of the everything. Like, both the stuff you’re showing here and all the transcripts just in the subfolder.
456 00:36:13.640 ⇒ 00:36:14.370 Uttam Kumaran: You know?
457 00:36:14.370 ⇒ 00:36:18.050 Samuel Roberts: That… that would be my, like, initial thought.
458 00:36:18.460 ⇒ 00:36:20.200 Uttam Kumaran: Like, why or why not?
459 00:36:20.200 ⇒ 00:36:31.510 Samuel Roberts: Yeah, I think… I think ergonomically, that makes the most sense. I think pointing to things… like, I’ve been trying to open up, like, my… on my computer, like, my Brainforge folder that has, like, clients in it, and, like, so I can have everything in cursor. I only worry about it
460 00:36:32.050 ⇒ 00:36:36.670 Samuel Roberts: leaking across in some way, and making things, like, reading…
461 00:36:37.170 ⇒ 00:36:41.099 Samuel Roberts: The wrong file, because it just saw a transcript, and
462 00:36:42.320 ⇒ 00:36:50.409 Samuel Roberts: Not that keeping them in separate repos would change that if we’re still opening them all in cursor, you know, together, so, like, it probably doesn’t matter at that point, we just have to be careful with it.
463 00:36:50.680 ⇒ 00:36:55.209 Samuel Roberts: But yeah, I think if it’s just an internal repo.
464 00:36:55.800 ⇒ 00:36:57.560 Samuel Roberts: That’s the best way to do it.
465 00:36:58.320 ⇒ 00:36:59.280 Samuel Roberts: Personally.
466 00:37:03.170 ⇒ 00:37:12.070 Samuel Roberts: I mean, the other option is, like, maybe this playbook is one repo, and, like, meeting transcripts, client info is another one, and this is more, like.
467 00:37:12.600 ⇒ 00:37:20.630 Samuel Roberts: How to operate, and the other one is more of the actual, like, Data that we would… use…
468 00:37:21.890 ⇒ 00:37:24.480 Uttam Kumaran: I guess, like, is there any reason to keep it separate?
469 00:37:26.420 ⇒ 00:37:27.640 Samuel Roberts: My only thought is…
470 00:37:27.640 ⇒ 00:37:28.260 Uttam Kumaran: just…
471 00:37:28.820 ⇒ 00:37:31.679 Samuel Roberts: It’s just how many times we’re gonna be, like, pushing stuff to it.
472 00:37:31.870 ⇒ 00:37:35.270 Samuel Roberts: And having to keep re-pulling, and… I don’t know.
473 00:37:35.940 ⇒ 00:37:41.369 Uttam Kumaran: But, like, that would be a problem anyways, so… Even if it’s separate.
474 00:37:42.370 ⇒ 00:37:45.920 Samuel Roberts: Yeah, I suppose. It’s just, yeah, keeping that other one up to date. Yeah, either way, you keep it.
475 00:37:45.920 ⇒ 00:37:51.420 Uttam Kumaran: I guess, like, my… if there is no reason to have multiple repos, then I would just put it all into one.
476 00:37:51.640 ⇒ 00:37:52.120 Gabriel Lam: Okay.
477 00:37:52.120 ⇒ 00:37:53.269 Samuel Roberts: Okay. Yeah.
478 00:37:53.770 ⇒ 00:37:54.620 Uttam Kumaran: Contact Leader.
479 00:37:54.620 ⇒ 00:37:55.110 Gabriel Lam: Can you also.
480 00:37:55.110 ⇒ 00:38:02.499 Uttam Kumaran: go… can you also go back to the Brainforge AI, like, list of repos? Because I want to show you guys a couple of others that I also want
481 00:38:02.620 ⇒ 00:38:10.649 Uttam Kumaran: some help making a decision on. So, we have, Brain Forge Playbook, BrainForge Vault, New Client.
482 00:38:11.020 ⇒ 00:38:19.750 Uttam Kumaran: Brainforge files, if you keep scrolling down, there’s… our platform.
483 00:38:19.960 ⇒ 00:38:22.470 Uttam Kumaran: And click… if you click on… if you click on this…
484 00:38:23.580 ⇒ 00:38:29.080 Uttam Kumaran: And, you’ll see there is Brainforge Data Platform, there is… keep going.
485 00:38:29.080 ⇒ 00:38:29.990 Gabriel Lam: file analytics.
486 00:38:29.990 ⇒ 00:38:35.620 Uttam Kumaran: File analytics, web platform, AI backend, all this shit is internal.
487 00:38:36.050 ⇒ 00:38:42.620 Uttam Kumaran: Yeah. This is, again, just, like, an annoying thing, but if we’re gonna talk about it, these are all the things that I would like for us to just decide on, like.
488 00:38:43.050 ⇒ 00:38:49.099 Uttam Kumaran: the… like… the AI platform.
489 00:38:49.330 ⇒ 00:38:56.130 Uttam Kumaran: I understand, like, there’s the AI platform, there’s also these client templates, and there is this, like, vault playbooks concept.
490 00:38:56.490 ⇒ 00:39:03.820 Uttam Kumaran: You know? Right? And there’s also… see, there’s something in here. I probably did that, like, a year or two ago. There’s probably some other stuff in the next page.
491 00:39:03.820 ⇒ 00:39:04.460 Samuel Roberts: Yeah, yeah.
492 00:39:04.460 ⇒ 00:39:09.119 Uttam Kumaran: like, Yeah, there’s internal Brainforge analytics.
493 00:39:09.120 ⇒ 00:39:10.880 Gabriel Lam: This website…
494 00:39:12.480 ⇒ 00:39:13.210 Gabriel Lam: And this backend.
495 00:39:13.210 ⇒ 00:39:14.560 Uttam Kumaran: this thing…
496 00:39:16.060 ⇒ 00:39:22.189 Uttam Kumaran: So, I think if we’re comfortable making the decision to combine Playbooks and vaults, I would like that.
497 00:39:23.970 ⇒ 00:39:30.999 Uttam Kumaran: Okay. I don’t see any reason not to, because it doesn’t matter if an automated thing is adding transcripts there, it doesn’t.
498 00:39:31.000 ⇒ 00:39:31.440 Samuel Roberts: Yeah.
499 00:39:31.440 ⇒ 00:39:32.150 Uttam Kumaran: anything.
500 00:39:32.570 ⇒ 00:39:33.310 Gabriel Lam: Hmm.
501 00:39:33.860 ⇒ 00:39:41.339 Gabriel Lam: Yeah, I think that’s fair. I think the main thing was I… and you guys can chime in, is if context bleeding or, like.
502 00:39:41.480 ⇒ 00:39:43.710 Gabriel Lam: That’s gonna be a problem of drawing.
503 00:39:44.460 ⇒ 00:39:47.870 Gabriel Lam: Context from other clients that might not be accurate.
504 00:39:49.200 ⇒ 00:39:49.590 Samuel Roberts: I mean.
505 00:39:49.590 ⇒ 00:39:55.810 Gabriel Lam: But I think we can also specify in our prompts, and we’re just like, hey, if I’m asking about this client, just stick to this folder structure.
506 00:39:55.810 ⇒ 00:40:04.239 Samuel Roberts: Yeah, I mean, we can mitigate that as best as possible, but like I said, if you’re opening multiple folders in cursor anyway, like, it already has that. Like, if you’re creating a workspace that is…
507 00:40:04.240 ⇒ 00:40:07.290 Uttam Kumaran: Yeah, you’re just gonna add both… you’re gonna have to have both of these, and so…
508 00:40:07.290 ⇒ 00:40:09.020 Samuel Roberts: So I don’t think… yeah, it doesn’t matter then.
509 00:40:09.020 ⇒ 00:40:09.970 Gabriel Lam: Okay, yeah.
510 00:40:11.850 ⇒ 00:40:12.390 Uttam Kumaran: Yeah.
511 00:40:12.540 ⇒ 00:40:15.889 Samuel Roberts: okay. Yeah, I think that makes sense then.
512 00:40:17.260 ⇒ 00:40:26.490 Gabriel Lam: And then for Utam, just a couple things I’m gonna show. I’m gonna hop on a call in about 5 minutes, with Ian, but if I…
513 00:40:28.950 ⇒ 00:40:34.949 Gabriel Lam: basically show… like, I’ve… Put in our data store, just…
514 00:40:35.560 ⇒ 00:40:41.150 Gabriel Lam: A client profile that you might imagine, a sample insurance document.
515 00:40:41.410 ⇒ 00:40:51.019 Gabriel Lam: I’m gonna try to get the right documentation from Ian, like, today. But so far, we’re just able to do gap analysis for insurance.
516 00:40:51.340 ⇒ 00:40:54.780 Gabriel Lam: And to be able to, like, draft an email. And so…
517 00:40:54.890 ⇒ 00:40:56.900 Gabriel Lam: I’m gonna keep refining this, basically, with
518 00:40:57.030 ⇒ 00:41:06.169 Gabriel Lam: The sort of outputs that he’s looking for, and if there’s any, like, sort of hierarchy with information and hierarchy of response.
519 00:41:06.470 ⇒ 00:41:13.970 Gabriel Lam: Also looking at, like, If we put in the transcript, that it can cite a transcript and
520 00:41:14.080 ⇒ 00:41:21.630 Gabriel Lam: match it so far to different parts of a sample supplemental application with follow-up questions. And so I think we’re…
521 00:41:22.000 ⇒ 00:41:27.370 Gabriel Lam: Taking a first step to reaching that out-market, solution.
522 00:41:28.010 ⇒ 00:41:31.249 Gabriel Lam: That webinar example that you’ve also shown.
523 00:41:31.970 ⇒ 00:41:33.690 Gabriel Lam: And…
524 00:41:33.950 ⇒ 00:41:47.019 Gabriel Lam: yeah, I think we’re taking steps. I think what they have is a lot more, sort of, siloed for different use cases, and I think that’s something that we can discuss about how siloed we want these demos to be, which I think, for now, it makes sense that they’re siloed.
525 00:41:47.280 ⇒ 00:41:48.380 Gabriel Lam: And then…
526 00:41:49.190 ⇒ 00:41:54.380 Gabriel Lam: For comparison, the first thing I thought of was, like, just using healthcare plans to test, because those are the most
527 00:41:54.800 ⇒ 00:41:59.440 Gabriel Lam: like, those are the clearest and most easy to find. So here’s an example of, like.
528 00:41:59.840 ⇒ 00:42:04.030 Gabriel Lam: differences between, like, last year and this year, and how things change. And so I’ve…
529 00:42:05.640 ⇒ 00:42:09.709 Gabriel Lam: Try to refine and tweak the… the system prompts.
530 00:42:10.090 ⇒ 00:42:12.040 Gabriel Lam: And the agent to get out.
531 00:42:13.230 ⇒ 00:42:21.719 Gabriel Lam: mostly, like, table and structured output versus unstructured output. That’s really been the sort of main pain point that I’ve been noticing, that it sometimes
532 00:42:22.120 ⇒ 00:42:28.240 Gabriel Lam: extracts Like, structured data as unstructured… With an unstructured response.
533 00:42:28.670 ⇒ 00:42:31.510 Gabriel Lam: So, I’m gonna work on that, and then…
534 00:42:31.800 ⇒ 00:42:37.889 Gabriel Lam: Yeah, try to get documentation and sort of precise questions and precise outputs out.
535 00:42:39.200 ⇒ 00:42:40.490 Uttam Kumaran: Okay. In the next call.
536 00:42:42.510 ⇒ 00:42:47.560 Uttam Kumaran: Yeah, I think start the call by just showing Ian what, like, the examples.
537 00:42:48.300 ⇒ 00:42:54.529 Uttam Kumaran: Like, don’t lead with questions first, because it’s just… he doesn’t… he doesn’t even know this world at all.
538 00:42:54.530 ⇒ 00:42:54.870 Gabriel Lam: Right.
539 00:42:54.870 ⇒ 00:42:55.800 Uttam Kumaran: So… Right.
540 00:42:55.950 ⇒ 00:43:03.630 Uttam Kumaran: Try to show it… try to show what… okay, based on your last feedback, here’s what we worked on, and let’s try to spend as much of the time getting, like.
541 00:43:03.800 ⇒ 00:43:13.130 Uttam Kumaran: Driving towards, like, 3 to 5, like, killer demos that each demo we should be able to execute end-to-end and, like.
542 00:43:13.890 ⇒ 00:43:17.980 Uttam Kumaran: like, I honestly would say, like, 3 minutes or less.
543 00:43:17.980 ⇒ 00:43:19.300 Gabriel Lam: Minutes, yeah. Yeah.
544 00:43:19.300 ⇒ 00:43:24.459 Uttam Kumaran: Like, we just don’t have much time with a prospect, and so, commonly.
545 00:43:24.540 ⇒ 00:43:40.960 Uttam Kumaran: as we all have sat with vendors before, they spend most of their time talking about their product and their demo, and then they leave, like, 90 seconds to be like, what do you think? Like, how does this work for you? We have to do the opposite. Our goal with these demos is to flash in the pan, be like.
546 00:43:41.100 ⇒ 00:43:43.049 Uttam Kumaran: Here’s, like, what we could do.
547 00:43:43.180 ⇒ 00:43:48.669 Uttam Kumaran: to get them to see themselves using this, and then to drive towards, like, a deal, you know? So…
548 00:43:48.670 ⇒ 00:43:49.270 Gabriel Lam: Yep.
549 00:43:49.720 ⇒ 00:43:50.370 Uttam Kumaran: Yeah.
550 00:43:50.550 ⇒ 00:43:51.280 Gabriel Lam: Awesome.
551 00:43:51.470 ⇒ 00:43:53.930 Gabriel Lam: Okay, perfect. I’ll be there. I’ll be listening in.
552 00:43:54.090 ⇒ 00:43:57.400 Gabriel Lam: Okay. Yeah, yeah. I’m gonna start the meeting now, so I’ll hop off.
553 00:43:58.160 ⇒ 00:44:03.889 Uttam Kumaran: Alright, okay. Thanks, guys. Yeah, let me know on Slack about the, kind of, couple things we talked about, so…
554 00:44:05.280 ⇒ 00:44:09.800 Gabriel Lam: Alright, sounds good. You guys want to stand and talk ABC real quick and split those up, or do you guys think you got that?
555 00:44:12.380 ⇒ 00:44:18.510 Casie Aviles: Yeah, we could, we could talk about, which… Yeah, which one’s to split?
556 00:44:21.220 ⇒ 00:44:26.090 Samuel Roberts: I mean, there might be a little bit of, like, overlap, depending on, like, what kind of changes are needed, so it might be even, like, a…
557 00:44:26.580 ⇒ 00:44:30.100 Samuel Roberts: Working together for some of them, but, like, the basic… let’s see, hold on.
558 00:44:30.340 ⇒ 00:44:38.980 Samuel Roberts: I’m going to sift out… the ABC team, you know, some of the Brainforge team… Mmm…
559 00:44:39.880 ⇒ 00:44:47.899 Samuel Roberts: I don’t know if it counts these easily, does it? SQL error, bad filter, insufficient content, so we got, like, 1, 2…
560 00:44:48.750 ⇒ 00:44:52.500 Samuel Roberts: 3, 4… license, okay.
561 00:44:54.660 ⇒ 00:44:57.629 Samuel Roberts: Oh, it’s not even all that, it’s just a pinball, our P0s up here, okay.
562 00:45:00.380 ⇒ 00:45:09.030 Samuel Roberts: Yeah, I’m… I’m thinking a little bit now about what he… what, Tom was saying about how we…
563 00:45:10.130 ⇒ 00:45:13.659 Samuel Roberts: Build towards something that is self-healing.
564 00:45:16.650 ⇒ 00:45:21.510 Samuel Roberts: So… I mean, obviously we want to resolve the ones that we can.
565 00:45:22.280 ⇒ 00:45:26.780 Samuel Roberts: But I think while doing that, keeping in mind that, like, what…
566 00:45:27.090 ⇒ 00:45:39.209 Samuel Roberts: what might have been a good way to surface this to something that could then repair it, or prompt a human to say, like, hey, do you want me to do this? So I would just say, keep that in mind now, but let’s prioritize fixing them.
567 00:45:41.080 ⇒ 00:45:41.530 Casie Aviles: Delaware.
568 00:45:41.530 ⇒ 00:45:44.730 Samuel Roberts: Okay, that’s… sorry, this is top of mind just now.
569 00:45:45.810 ⇒ 00:45:48.770 Samuel Roberts: Yeah, I don’t know, Mustafa, you might have more insight into these…
570 00:45:49.070 ⇒ 00:45:54.839 Samuel Roberts: actual errors at this point. So I don’t know if there’s, like, ones you think are, like, which way to split these.
571 00:45:57.360 ⇒ 00:46:06.709 Mustafa Raja: Yeah, so, the ones that we will be working on, I believe, are only, you know, updates in database.
572 00:46:07.910 ⇒ 00:46:08.840 Mustafa Raja: You know?
573 00:46:09.690 ⇒ 00:46:11.070 Samuel Roberts: Oh, okay.
574 00:46:11.410 ⇒ 00:46:13.249 Samuel Roberts: Oh, assignments? No…
575 00:46:13.370 ⇒ 00:46:14.810 Mustafa Raja: Yeah…
576 00:46:15.360 ⇒ 00:46:16.830 Samuel Roberts: Well, that’s…
577 00:46:16.830 ⇒ 00:46:17.910 Mustafa Raja: And then there was…
578 00:46:18.220 ⇒ 00:46:19.440 Samuel Roberts: Okay, hold on.
579 00:46:20.600 ⇒ 00:46:28.139 Mustafa Raja: So assignments… assignments none mean, that assignments table has nothing for these ones.
580 00:46:28.400 ⇒ 00:46:29.160 Mustafa Raja: You know?
581 00:46:30.710 ⇒ 00:46:31.030 Casie Aviles: Yeah.
582 00:46:31.030 ⇒ 00:46:31.380 Samuel Roberts: Okay.
583 00:46:31.380 ⇒ 00:46:38.350 Casie Aviles: So, a lot of these are… I’m sorry, but there are no… you know, it’s not found in the database, so…
584 00:46:38.510 ⇒ 00:46:39.979 Casie Aviles: Yeah, yeah.
585 00:46:40.870 ⇒ 00:46:47.710 Mustafa Raja: And some of these are going to be, you know, the data is, the data needs to be updated or something.
586 00:46:47.940 ⇒ 00:46:50.529 Mustafa Raja: And some of these are going to be…
587 00:46:50.700 ⇒ 00:47:07.220 Mustafa Raja: Or a lot less of these are going to be, errors in… or not errors, actually, but outdated information in Central Doc, or needs… the information needs a better structure in Central Doc.
588 00:47:07.460 ⇒ 00:47:11.040 Mustafa Raja: So, this, this is just the gist of it, you know?
589 00:47:11.780 ⇒ 00:47:13.020 Samuel Roberts: Yeah, okay.
590 00:47:13.200 ⇒ 00:47:17.950 Samuel Roberts: I would say…
591 00:47:17.950 ⇒ 00:47:25.709 Mustafa Raja: And then, there was one query error, and then there was one, other query where,
592 00:47:26.070 ⇒ 00:47:32.659 Mustafa Raja: You know, there was a query here, let me actually find that one, because that is important.
593 00:47:39.220 ⇒ 00:47:41.040 Mustafa Raja: Give me a moment, Andy.
594 00:47:41.630 ⇒ 00:47:42.170 Samuel Roberts: Sure.
595 00:47:45.640 ⇒ 00:47:48.190 Mustafa Raja: Yeah, it’s this one… okay.
596 00:47:48.710 ⇒ 00:47:50.649 Mustafa Raja: Let me share my screen also.
597 00:47:57.650 ⇒ 00:48:10.939 Mustafa Raja: Okay, so, so it says that, who are the chem-free best techs, right? And then the query, what query does is it goes to the branch chem-free, but this.
598 00:48:10.940 ⇒ 00:48:11.730 Samuel Roberts: Oh.
599 00:48:11.760 ⇒ 00:48:18.670 Mustafa Raja: more like the service should become free pest, and I need the technicians for that, right?
600 00:48:20.630 ⇒ 00:48:26.840 Mustafa Raja: So… so this is something that… but this is… this isn’t something very frequent.
601 00:48:27.050 ⇒ 00:48:32.659 Mustafa Raja: This is the only example we have in our set for this sort of stuff.
602 00:48:33.740 ⇒ 00:48:37.919 Samuel Roberts: Kemphrey is, like, the… sister company, right? Or…
603 00:48:37.920 ⇒ 00:48:38.390 Mustafa Raja: Yeah.
604 00:48:38.770 ⇒ 00:48:39.770 Casie Aviles: Yeah, I think so.
605 00:48:39.770 ⇒ 00:48:40.210 Mustafa Raja: leaves.
606 00:48:40.210 ⇒ 00:48:44.750 Samuel Roberts: How is that… how is that even… is that even represented in the ZIPS database at all, properly?
607 00:48:46.160 ⇒ 00:48:56.080 Casie Aviles: Yeah, so that’s… I think that’s also something we can optimize, because right now, it’s… Branch-based, so…
608 00:48:56.430 ⇒ 00:48:57.390 Samuel Roberts: Right.
609 00:48:57.810 ⇒ 00:49:08.710 Casie Aviles: So, yeah, so, for example, it’s either Austin, or College Station, or Camphrey, so it’s, it’s, it’s under, like, a…
610 00:49:09.640 ⇒ 00:49:14.900 Casie Aviles: Yeah, it’s under the branch, but it’s not like… there’s no clear distinction whether it’s…
611 00:49:15.150 ⇒ 00:49:19.480 Casie Aviles: ABC versus scam-free, like, maybe that could be a separate column, I guess.
612 00:49:19.480 ⇒ 00:49:21.540 Samuel Roberts: Okay, yeah, it might have to be then.
613 00:49:21.680 ⇒ 00:49:23.280 Samuel Roberts: Okay, I think…
614 00:49:23.750 ⇒ 00:49:31.389 Mustafa Raja: Yeah, and the query that needed… I believe that needed to be run was this one. You know, the service name should have
615 00:49:31.590 ⇒ 00:49:36.819 Mustafa Raja: Chem-free in there, and not… not just restricting a location, you know.
616 00:49:37.580 ⇒ 00:49:42.110 Samuel Roberts: Well, yeah, would it be service name, or do we need to add another… column.
617 00:49:42.110 ⇒ 00:49:44.520 Mustafa Raja: So, which name’s our deer?
618 00:49:44.520 ⇒ 00:49:44.970 Samuel Roberts: free?
619 00:49:44.970 ⇒ 00:49:46.649 Mustafa Raja: Chem-free audio, yeah.
620 00:49:46.650 ⇒ 00:49:48.139 Samuel Roberts: Oh, okay. Alright, well, maybe…
621 00:49:48.140 ⇒ 00:49:48.490 Casie Aviles: Yes, it’s.
622 00:49:48.770 ⇒ 00:49:49.680 Mustafa Raja: I appreciate that.
623 00:49:49.680 ⇒ 00:49:50.649 Samuel Roberts: that it knows.
624 00:49:50.650 ⇒ 00:49:55.900 Mustafa Raja: that these… this would… I didn’t… I believe that these don’t have…
625 00:49:56.170 ⇒ 00:50:05.319 Mustafa Raja: assignments, because when I run this query, I got nothing, but, just looking at the service name and services came free is there.
626 00:50:05.670 ⇒ 00:50:06.220 Casie Aviles: Okay.
627 00:50:06.510 ⇒ 00:50:10.870 Casie Aviles: Gemphrey’s also hard-coded into some of the services.
628 00:50:11.440 ⇒ 00:50:17.370 Samuel Roberts: Okay, alright, that’s… alright, so that’s probably just a prompting thing to make sure that it… when it sees the name Kemphrey, it knows to include that into the service name.
629 00:50:17.370 ⇒ 00:50:20.240 Mustafa Raja: Yeah, I think, I think,
630 00:50:20.480 ⇒ 00:50:25.580 Mustafa Raja: I think my question for Casey is, do we… do we have technicians?
631 00:50:26.010 ⇒ 00:50:30.170 Mustafa Raja: For chemistry pest service, because this query returned nothing.
632 00:50:31.870 ⇒ 00:50:36.510 Casie Aviles: Oh, you mean chem-free technicians? Yeah, we do have chem-free technicians.
633 00:50:36.910 ⇒ 00:50:39.050 Mustafa Raja: Okay.
634 00:50:39.370 ⇒ 00:50:45.210 Samuel Roberts: I would say, as part of this then, like, when there’s something that’s not clear, and we need to, like, clarify with them if it’s…
635 00:50:45.820 ⇒ 00:50:51.440 Samuel Roberts: who the right person is, or, you know, is the sheet up to date, or something like that. Like, keep that list going.
636 00:50:52.040 ⇒ 00:50:55.739 Samuel Roberts: Of those sorts of questions, because I want to get with Janiece at some point.
637 00:50:55.920 ⇒ 00:51:01.599 Samuel Roberts: And… and try to understand her workflow, and, like, maybe we can clarify a few things like this.
638 00:51:01.960 ⇒ 00:51:04.660 Samuel Roberts: So… Keep that in mind.
639 00:51:06.180 ⇒ 00:51:06.830 Casie Aviles: Sure.
640 00:51:07.730 ⇒ 00:51:09.860 Samuel Roberts: Alright, cool.
641 00:51:26.880 ⇒ 00:51:28.119 Mustafa Raja: Yeah, so he’s a good one.
642 00:51:28.260 ⇒ 00:51:29.910 Mustafa Raja: I’ve done nothing, yeah.
643 00:51:30.550 ⇒ 00:51:32.609 Samuel Roberts: Yeah, okay, so maybe…
644 00:51:36.770 ⇒ 00:51:41.499 Samuel Roberts: Yeah, I would say just keep working through these, and like, if there’s things that are like this, where we’re just like, oh, I don’t know…
645 00:51:41.500 ⇒ 00:51:42.220 Mustafa Raja: Nasu.
646 00:51:42.470 ⇒ 00:51:45.100 Mustafa Raja: So, chem-free service has only inspectors.
647 00:51:45.360 ⇒ 00:51:46.100 Mustafa Raja: You know?
648 00:51:46.830 ⇒ 00:51:50.010 Samuel Roberts: Okay, so maybe that’s… that’s an issue with the data, then.
649 00:51:50.480 ⇒ 00:51:54.620 Casie Aviles: I think it’s… it’s the branch that we have to look at for…
650 00:51:54.620 ⇒ 00:51:57.889 Samuel Roberts: Oh… oh, the branch will have chemistry and.
651 00:51:57.890 ⇒ 00:51:58.240 Casie Aviles: Yeah.
652 00:51:58.240 ⇒ 00:51:59.150 Samuel Roberts: That is not an issue.
653 00:51:59.580 ⇒ 00:52:03.170 Casie Aviles: For technicians, at least. That’s why it’s confusing.
654 00:52:03.760 ⇒ 00:52:04.200 Samuel Roberts: Okay.
655 00:52:04.200 ⇒ 00:52:06.469 Casie Aviles: But it needs to be optimized.
656 00:52:06.690 ⇒ 00:52:07.780 Samuel Roberts: Yeah, okay.
657 00:52:16.880 ⇒ 00:52:17.580 Mustafa Raja: Okay.
658 00:52:19.270 ⇒ 00:52:26.390 Mustafa Raja: Yeah, so I see that this service, ChemFreePest, which is… which was in the question, ChemFreePest, right?
659 00:52:27.330 ⇒ 00:52:31.459 Mustafa Raja: It’s only… it says that it’s assigned the role…
660 00:52:31.750 ⇒ 00:52:37.830 Casie Aviles: Yeah, the road ID is… The ID2 means inspector. That should be inspected.
661 00:52:38.780 ⇒ 00:52:39.620 Casie Aviles: So…
662 00:52:40.700 ⇒ 00:52:43.779 Mustafa Raja: So if the… so, is this correct question, then?
663 00:52:43.780 ⇒ 00:52:47.599 Samuel Roberts: Well, there would be a… Casey’s saying there’s a branch that has chem-free.
664 00:52:48.010 ⇒ 00:52:56.269 Casie Aviles: Yeah, this is the correct question. I guess we have to prompt… we just have to prompt it to look at the branch for technicians.
665 00:52:56.980 ⇒ 00:52:57.700 Mustafa Raja: Okay.
666 00:53:01.380 ⇒ 00:53:02.210 Mustafa Raja: Okay.
667 00:53:08.300 ⇒ 00:53:14.500 Casie Aviles: Yeah, but I think that’s… that’s honestly… that’s a little confusing, because that should be…
668 00:53:14.960 ⇒ 00:53:17.879 Casie Aviles: I mean, I mean, the way I set it up, so I think that’s…
669 00:53:18.030 ⇒ 00:53:22.419 Casie Aviles: down the line, we should probably re… I should probably think about, you know…
670 00:53:23.420 ⇒ 00:53:26.170 Casie Aviles: Like, a better way to differentiate.
671 00:53:26.470 ⇒ 00:53:30.309 Casie Aviles: CAM-free from non-CAM-free services and assignments.
672 00:53:30.310 ⇒ 00:53:30.980 Samuel Roberts: Okay.
673 00:53:31.530 ⇒ 00:53:37.729 Samuel Roberts: Okay, yeah, I would say… Add that to a list of things that we can bring up and understand.
674 00:53:37.980 ⇒ 00:53:38.580 Samuel Roberts: how…
675 00:53:38.580 ⇒ 00:53:39.160 Casie Aviles: Okay.
676 00:53:40.000 ⇒ 00:53:45.039 Samuel Roberts: Like, yeah, the list doesn’t necessarily just have to be things that we’ll bring to them, but, like, just things that, like, surface all the way up to…
677 00:53:45.180 ⇒ 00:53:49.020 Samuel Roberts: Like, another conversation about restructuring things and stuff, so…
678 00:53:49.500 ⇒ 00:53:50.740 Casie Aviles: Yes, definitely.
679 00:53:50.970 ⇒ 00:53:52.000 Samuel Roberts: Okay, cool.
680 00:53:54.580 ⇒ 00:53:56.799 Samuel Roberts: Alright, yeah, so I would say,
681 00:53:57.280 ⇒ 00:54:06.650 Samuel Roberts: If, yeah, if the two of you need to do, like, working sessions together, or talk through things, feel free. I would say just keep me in the loop. I had a bunch to do with Lilo today, so…
682 00:54:07.170 ⇒ 00:54:10.560 Samuel Roberts: I’ll be on that, but if I can answer or help or think through things, let me know.
683 00:54:11.310 ⇒ 00:54:18.900 Mustafa Raja: Okay, so, for today, do we only want to make progress in, this?
684 00:54:19.350 ⇒ 00:54:24.110 Mustafa Raja: In fixing these errors, or do we want to also look into some of the migration stuff?
685 00:54:25.440 ⇒ 00:54:30.440 Samuel Roberts: I would say the two of you working on this is probably the best thing right now, to show…
686 00:54:30.860 ⇒ 00:54:34.620 Samuel Roberts: That we can message to them, and then if there’s time later.
687 00:54:34.820 ⇒ 00:54:37.450 Samuel Roberts: We can talk migration stuff, maybe.
688 00:54:38.160 ⇒ 00:54:38.840 Mustafa Raja: Okay.
689 00:54:38.840 ⇒ 00:54:42.670 Samuel Roberts: But… but focus on this, just because I think we want to show that there’s progress made.
690 00:54:43.260 ⇒ 00:54:43.920 Mustafa Raja: Okay.
691 00:54:44.820 ⇒ 00:54:45.830 Samuel Roberts: Cool, thanks.
692 00:54:47.510 ⇒ 00:54:54.249 Samuel Roberts: Alright, I’m gonna hop… do you guys wanna stay on here, or go somewhere else, or hop off yourselves? Should I end the meeting, or just leave?
693 00:54:54.560 ⇒ 00:54:58.910 Mustafa Raja: Yeah, I guess we can set up a new, new meeting for welcome session.
694 00:54:59.230 ⇒ 00:55:06.750 Samuel Roberts: Okay, yeah, keep me, like I said, keep me in the loop, but, yeah, put together, like, a list of stuff if we can’t resolve things that are unclear.
695 00:55:06.880 ⇒ 00:55:07.350 Samuel Roberts: Alright.
696 00:55:09.570 ⇒ 00:55:10.260 Samuel Roberts: Thanks, guys.
697 00:55:11.000 ⇒ 00:55:11.770 Mustafa Raja: Yeah, thank you very much.
698 00:55:11.830 ⇒ 00:55:12.920 Samuel Roberts: Talk to you later.