Meeting Title: AI Service Standup Date: 2026-01-21 Meeting participants: Clarence Stone, Gabriel Lam, Mustafa Raja, Samuel Roberts, Casie Aviles, Amber Lin
WEBVTT
1 00:01:19.380 ⇒ 00:01:21.130 Samuel Roberts: Hey, everyone.
2 00:01:22.830 ⇒ 00:01:23.640 Mustafa Raja: A…
3 00:01:24.550 ⇒ 00:01:25.869 Samuel Roberts: How we doing today?
4 00:01:29.210 ⇒ 00:01:30.260 Samuel Roberts: Cool.
5 00:01:36.920 ⇒ 00:01:45.340 Samuel Roberts: Yeah, I guess, let’s wait a minute or two… Amber, Utum, and Penal.
6 00:01:46.300 ⇒ 00:01:48.079 Samuel Roberts: It’s running a minute late.
7 00:01:48.930 ⇒ 00:01:51.330 Samuel Roberts: Otherwise, we’ll just jump in.
8 00:02:01.890 ⇒ 00:02:12.749 Samuel Roberts: Sorry, I’m a little distracted. 10 o’clock snuck up on me all of a sudden, and I’m in the middle of a PR review on some of the stuff Lutam has for the platform, so I was bouncing around a little bit,
9 00:02:13.240 ⇒ 00:02:16.089 Samuel Roberts: And then all of a sudden, I got the notification, it’s like, oh, wow, it’s already 10.
10 00:02:20.580 ⇒ 00:02:23.900 Samuel Roberts: How’s everyone’s, yesterday, I guess?
11 00:02:25.140 ⇒ 00:02:29.059 Mustafa Raja: Yeah, so yesterday for EBC, I, I worked on the evals.
12 00:02:29.150 ⇒ 00:02:30.030 Samuel Roberts: Yes.
13 00:02:30.330 ⇒ 00:02:34.850 Mustafa Raja: And I’ve sent the loom for review. Yeah, I saw that after…
14 00:02:35.040 ⇒ 00:02:35.890 Samuel Roberts: Yes, sir.
15 00:02:35.890 ⇒ 00:02:41.799 Mustafa Raja: Yeah. Apart from that, I did… I did take a look at, Pranav’s work on…
16 00:02:42.130 ⇒ 00:02:50.859 Mustafa Raja: the evals, I think I need… I think I need to sync with him. Yeah. To sort of understand,
17 00:02:51.110 ⇒ 00:02:53.559 Mustafa Raja: some of these, metrics, you’re suggesting?
18 00:02:54.150 ⇒ 00:03:06.440 Mustafa Raja: Yeah. Once that’s done. And I believe, we also need to… we also wanted to do, a live, eval stuff, right?
19 00:03:07.040 ⇒ 00:03:10.150 Samuel Roberts: I think we should, like, kind of ongoing…
20 00:03:10.400 ⇒ 00:03:11.040 Mustafa Raja: Hmm.
21 00:03:11.040 ⇒ 00:03:19.490 Samuel Roberts: as the requests are coming in, we’ll get a few scores, and then, like, we can kind of correlate those scores maybe to thumbs down over time or something.
22 00:03:19.490 ⇒ 00:03:20.260 Mustafa Raja: I think…
23 00:03:20.520 ⇒ 00:03:31.749 Mustafa Raja: I think, I’ll sync with Pranav on the metrics that he suggested, and also, a framework for this live, live evaluation stuff.
24 00:03:31.990 ⇒ 00:03:38.360 Samuel Roberts: Okay, cool, yeah. Yeah, I was hoping you guys would sync before, he got a chance to dig in, because I wanted him to get a little context more.
25 00:03:39.840 ⇒ 00:03:44.939 Samuel Roberts: at least now is good. So that’s something… something on paper is always, you know, a good starting place. That’s fine.
26 00:03:44.940 ⇒ 00:03:45.340 Mustafa Raja: Yeah.
27 00:03:45.340 ⇒ 00:03:51.910 Samuel Roberts: But… yeah, no, you have definitely more context into some of the stuff than he does, so don’t… don’t,
28 00:03:52.420 ⇒ 00:03:55.169 Samuel Roberts: Don’t worry about that. Like, use your, you know…
29 00:03:55.810 ⇒ 00:04:10.020 Samuel Roberts: your… your knowledge of the… of ABC and everything we’ve been doing. This is sort of just to, like, get him into ABC a little bit so that we have more support overall. So yeah, by all means, if things don’t make sense or whatever, be sure to clarify whatever you need.
30 00:04:10.210 ⇒ 00:04:11.750 Samuel Roberts: Cool. Okay.
31 00:04:12.180 ⇒ 00:04:14.900 Samuel Roberts: I guess, yeah, let’s keep going ABC.
32 00:04:15.580 ⇒ 00:04:17.310 Samuel Roberts: No one’s not top-not yet.
33 00:04:19.490 ⇒ 00:04:21.930 Samuel Roberts: Yeah, anything yesterday, Casey?
34 00:04:21.959 ⇒ 00:04:30.100 Casie Aviles: for ABC, I managed to… Get, you know, the tables live.
35 00:04:30.400 ⇒ 00:04:33.319 Casie Aviles: So that should be the one that they’re using right now.
36 00:04:35.840 ⇒ 00:04:42.000 Casie Aviles: What else? So, yeah, I think I’ll be able to, like, adjust the tickets much, much easier now.
37 00:04:42.260 ⇒ 00:04:43.890 Casie Aviles: with the admin UI.
38 00:04:44.580 ⇒ 00:04:50.599 Casie Aviles: Yeah, I think… I think that’s… that’s pretty much it there, so I’ll be working on…
39 00:04:51.080 ⇒ 00:04:54.380 Casie Aviles: using the UI now to fix whatever…
40 00:04:55.250 ⇒ 00:04:57.330 Casie Aviles: Issues that have been reported.
41 00:04:57.520 ⇒ 00:05:00.390 Samuel Roberts: Okay. You know, for, like, freshness issues.
42 00:05:00.960 ⇒ 00:05:01.780 Casie Aviles: And then…
43 00:05:02.980 ⇒ 00:05:15.859 Casie Aviles: Yeah, I guess I’ll also let the clients know that, you know, we’ve done… we’re doing that, and that we’ll continue to, like… if any errors will surface, then we can… we will just, you know…
44 00:05:16.110 ⇒ 00:05:18.399 Casie Aviles: Continue to work on that.
45 00:05:19.130 ⇒ 00:05:21.269 Samuel Roberts: Now, are they,
46 00:05:23.150 ⇒ 00:05:30.080 Samuel Roberts: are… so, like, are they confirming? Because, like, people will say, you know, oh, so-and-so is the person there.
47 00:05:30.240 ⇒ 00:05:36.719 Samuel Roberts: Are… are, like, Denise or Yvette confirming that before we make those changes for the triage tickets?
48 00:05:38.390 ⇒ 00:05:41.770 Casie Aviles: Oh, okay, yeah, I understand. So…
49 00:05:42.110 ⇒ 00:05:46.549 Casie Aviles: Some of those that have been assigned to me, I think I can work on, because… Okay.
50 00:05:47.080 ⇒ 00:05:50.759 Casie Aviles: those that were… because Janiece would assign those to me.
51 00:05:50.760 ⇒ 00:05:54.270 Samuel Roberts: Got it, okay, I just wanted to make sure that there was some step here where she had input.
52 00:05:54.850 ⇒ 00:05:55.429 Casie Aviles: Yes, yes.
53 00:05:55.430 ⇒ 00:06:03.089 Samuel Roberts: So if that’s fine, yeah, okay. And then I figure eventually, you know, knock on wood, ongoing, they’ll be able to do this proactively, right?
54 00:06:03.090 ⇒ 00:06:05.740 Casie Aviles: Yeah, that’s, that’s, that’s a goal. Okay.
55 00:06:05.960 ⇒ 00:06:14.330 Samuel Roberts: Okay, yeah, I just wanted to make sure that there was a step there where we were getting their confirmation of those changes. All right, cool, yeah, you’ve got it under control, so… keep at that.
56 00:06:14.330 ⇒ 00:06:15.440 Casie Aviles: Okay,
57 00:06:16.300 ⇒ 00:06:27.030 Casie Aviles: What else? So, I guess we were just running into some limits earlier, I was just on a call with Mustafa… with Heroku. We managed to resolve it, it was just, like, a memory issue.
58 00:06:27.380 ⇒ 00:06:31.289 Samuel Roberts: Oh, yeah, okay. I’ve hit that in the past with other stuff.
59 00:06:31.900 ⇒ 00:06:40.609 Casie Aviles: However, yeah, we were just talking, like, whether, you know, we should start using railway as well. I think railway has been really good so far.
60 00:06:40.610 ⇒ 00:06:49.390 Samuel Roberts: Yeah, I’ve been very satisfied with it, and I think especially as we’re gonna move to, like, the monorepo, multiple apps inside it, it’ll make sense to…
61 00:06:49.820 ⇒ 00:06:55.960 Samuel Roberts: do it that way, even if they don’t talk to each other. It’s just a nice way to have multiple things that we can
62 00:06:56.190 ⇒ 00:07:03.619 Samuel Roberts: spin up from one repo easily. I don’t know what kind of memory limits or anything might be there, so, you know, there’s still…
63 00:07:04.150 ⇒ 00:07:08.209 Samuel Roberts: But I think… I think it’s a good idea. Heroku has been…
64 00:07:08.520 ⇒ 00:07:14.690 Samuel Roberts: even just not having to log in every other time is a huge benefit, I find, because it just slows me down every single time.
65 00:07:15.200 ⇒ 00:07:18.560 Mustafa Raja: Yeah, and they ask for OTP, I don’t like that.
66 00:07:18.560 ⇒ 00:07:21.529 Samuel Roberts: Every time… yeah, fortunately, 1Password’s been pretty good for me, but.
67 00:07:21.530 ⇒ 00:07:22.089 Mustafa Raja: It’s annoying.
68 00:07:22.090 ⇒ 00:07:25.289 Samuel Roberts: It is, like, an extra step, which is… yeah.
69 00:07:25.530 ⇒ 00:07:26.020 Samuel Roberts: real way.
70 00:07:26.430 ⇒ 00:07:27.120 Samuel Roberts: right in.
71 00:07:28.300 ⇒ 00:07:41.590 Casie Aviles: Yeah, and I think the ben… I think what’s good also is what I’m seeing with Lilo now is we’re able to have, like, the database, you know, as well in there, and not… we don’t have to have, like, a separate SUPA base for that, so…
72 00:07:42.030 ⇒ 00:07:50.209 Samuel Roberts: Yeah, I mean, there’s still advantages, I think, for Supabase for some things. Like, the UI on Supabase is…
73 00:07:50.830 ⇒ 00:08:04.609 Samuel Roberts: is a nice feature. The auth and everything that we got for, you know, the platform and everything as well. You know, we don’t necessarily need the buckets if we’re using Railway, because that does help us there, but yeah, I see what you’re saying, because I was a little,
74 00:08:04.810 ⇒ 00:08:09.730 Samuel Roberts: when we were starting Lilo, I was internally debating, like.
75 00:08:10.010 ⇒ 00:08:13.590 Samuel Roberts: the database with Prisma versus Supabase.
76 00:08:15.040 ⇒ 00:08:24.760 Samuel Roberts: But since we were setting up BetterAuth anyway, I feel like it didn’t… basically didn’t give us the benefits we needed there, but, I think for some other things moving forward, yeah, I’d be able to spin up a database
77 00:08:24.870 ⇒ 00:08:30.530 Samuel Roberts: pretty easily, even if we need to use something else besides the Postgres database at some point, you know.
78 00:08:30.650 ⇒ 00:08:34.349 Samuel Roberts: But I think it’s nice to have just some local data there.
79 00:08:37.169 ⇒ 00:08:38.270 Casie Aviles: Okay, yeah.
80 00:08:38.270 ⇒ 00:08:44.720 Samuel Roberts: for, yeah, some things where we have to go in and do, you know, unless we want to build a UI or lean on something like Prisma Studio.
81 00:08:44.840 ⇒ 00:08:49.089 Samuel Roberts: You know, I don’t think we need to jump ship from Superbase yet, but…
82 00:08:49.750 ⇒ 00:08:52.829 Samuel Roberts: It could be something in the future worth thinking about, so…
83 00:08:53.890 ⇒ 00:08:54.440 Casie Aviles: Yep.
84 00:08:54.690 ⇒ 00:09:02.069 Casie Aviles: Yeah, I didn’t want to, like, also delay the actual going to live, so I didn’t work on the…
85 00:09:02.390 ⇒ 00:09:04.770 Casie Aviles: I didn’t transfer to railway right away.
86 00:09:04.940 ⇒ 00:09:07.460 Samuel Roberts: Yeah, that’s fine. That’s fine.
87 00:09:07.460 ⇒ 00:09:12.420 Casie Aviles: And… Yeah, I think Amber just forgot, and so…
88 00:09:12.420 ⇒ 00:09:12.770 Samuel Roberts: Yes.
89 00:09:12.770 ⇒ 00:09:15.129 Casie Aviles: going to… Recap that…
90 00:09:15.940 ⇒ 00:09:22.419 Casie Aviles: The database that we’re using for the admin UI should be the same one that the CSRs are using now.
91 00:09:22.730 ⇒ 00:09:25.320 Casie Aviles: So… We should be able to…
92 00:09:25.740 ⇒ 00:09:29.720 Casie Aviles: make changes using the new admin UI, and that would be…
93 00:09:30.330 ⇒ 00:09:35.730 Casie Aviles: You know, that would… so when the CSRs ask questions, they would get that now, so…
94 00:09:35.730 ⇒ 00:09:36.140 Samuel Roberts: Right.
95 00:09:36.140 ⇒ 00:09:39.750 Casie Aviles: I’m going to be resolving the triage tickets that we have.
96 00:09:40.290 ⇒ 00:09:41.980 Casie Aviles: Using the admin UI.
97 00:09:42.890 ⇒ 00:09:44.540 Casie Aviles: What, what I’m not…
98 00:09:44.640 ⇒ 00:09:49.750 Casie Aviles: But I haven’t taken a look at yet are the tickets that were new for this…
99 00:09:51.350 ⇒ 00:09:52.949 Casie Aviles: Yeah, the new tickets.
100 00:09:53.300 ⇒ 00:09:59.600 Casie Aviles: That we scheduled for today, so… or, like, I mean, this cycle, so… Wow.
101 00:09:59.720 ⇒ 00:10:06.230 Casie Aviles: We’ll see, I’ll see, I’ll let you guys know which ones, because I think that there’s a lot there, there’s a lot going on there.
102 00:10:06.990 ⇒ 00:10:09.120 Samuel Roberts: Okay, yeah, definitely keep us,
103 00:10:09.720 ⇒ 00:10:11.450 Samuel Roberts: Keep us updated when you take a look at them.
104 00:10:12.310 ⇒ 00:10:12.930 Samuel Roberts: Oh my god.
105 00:10:12.930 ⇒ 00:10:18.229 Casie Aviles: Yeah, I’ll let you guys know which ones I can take, which ones are…
106 00:10:20.510 ⇒ 00:10:26.370 Casie Aviles: Can be reassigned, or, yeah, if we can… Push those, but yeah.
107 00:10:26.510 ⇒ 00:10:28.559 Casie Aviles: That’s… that’s all I have for ABC.
108 00:10:28.930 ⇒ 00:10:36.140 Samuel Roberts: Great, sounds good. Amber, I believe I owe you a review of the Gantt, side, and…
109 00:10:36.260 ⇒ 00:10:39.230 Samuel Roberts: Get a chance yesterday. It kinda got away from me.
110 00:10:39.400 ⇒ 00:10:43.320 Samuel Roberts: So I’ll try to get that done ASAP this morning.
111 00:10:43.690 ⇒ 00:10:45.950 Samuel Roberts: And reply to your message.
112 00:10:47.780 ⇒ 00:10:52.220 Samuel Roberts: Cool. Any, anything else ABC-wise, Amber?
113 00:10:52.610 ⇒ 00:11:01.160 Samuel Roberts: Mustafa talked about evals, earlier before you hopped on, but I think, we… we may owe him a review there to Balloon.
114 00:11:01.980 ⇒ 00:11:05.400 Mustafa Raja: Yeah, one… One more review would be,
115 00:11:05.700 ⇒ 00:11:09.159 Mustafa Raja: Would be necessary for the cancellation.
116 00:11:09.720 ⇒ 00:11:13.770 Samuel Roberts: That’s right, and yeah, take a look at the cancellation, loom as well.
117 00:11:13.770 ⇒ 00:11:19.790 Mustafa Raja: Yeah, yeah. Oh, I tagged you earlier this morning, so… Boom.
118 00:11:19.960 ⇒ 00:11:20.670 Mustafa Raja: Yeah.
119 00:11:22.500 ⇒ 00:11:25.029 Amber Lin: Sounds good. I’ll write that down.
120 00:11:25.190 ⇒ 00:11:33.329 Samuel Roberts: Perfect. Thank you, thank you. Anything else, ABC? On what day are we at today? We have a… we have tomorrow’s the meeting, right? We have a working session today.
121 00:11:33.560 ⇒ 00:11:35.520 Samuel Roberts: On the calendar.
122 00:11:36.780 ⇒ 00:11:41.729 Samuel Roberts: Any other… Things we need to address now? Any other blockers or anything?
123 00:11:44.690 ⇒ 00:11:52.760 Samuel Roberts: All right, I guess let’s just jump to internal, in case Robert Utam join, we can talk legal after, but,
124 00:11:53.520 ⇒ 00:12:00.179 Samuel Roberts: Gabe, how… how are we looking there? I have a couple… I was in PR reviews for some of the things on the platform, I think I mentioned.
125 00:12:01.130 ⇒ 00:12:02.350 Gabriel Lam: Yeah,
126 00:12:02.690 ⇒ 00:12:17.340 Gabriel Lam: on that part, I’d love to, like, hear what your thoughts are, just to see what you… what you’ve noticed, or what, you know, is this something we can deploy? I also wanted to restart. I think we put the sort of linear tickets on the back burner, and I noticed that
127 00:12:17.440 ⇒ 00:12:26.659 Gabriel Lam: as I was talking to people, people are sort of just writing their own things again, and I wonder if there’s a way for, like, Cursor to really help us either groom or,
128 00:12:27.610 ⇒ 00:12:34.429 Gabriel Lam: Even ride tickets, sort of, In conjunction with all we have on the web platform.
129 00:12:34.850 ⇒ 00:12:43.480 Gabriel Lam: Okay. Because what we have there is sort of… it’s still a little, like, stopgap-y, band-aid-y, so when I was doing some testing, it’s like…
130 00:12:43.580 ⇒ 00:12:44.710 Gabriel Lam: either…
131 00:12:46.480 ⇒ 00:12:59.429 Gabriel Lam: Yeah, I don’t know, I think that’s something on my end. I’m trying to figure out how to get people to sort of adopt that workflow in a more easy manner. And then now that we have linear and, sort of, you can connect the Superbase via cursor, is that a way that we can sort of circumvent that?
132 00:12:59.570 ⇒ 00:13:06.219 Gabriel Lam: The other thing would just be, I see Uten was working on, like, a superbase, sort of.
133 00:13:06.440 ⇒ 00:13:13.920 Gabriel Lam: pull into GitHub, and so I’m trying to figure out where that sits. But I’m trying to… I’m thinking of restarting that.
134 00:13:13.920 ⇒ 00:13:14.430 Samuel Roberts: Okay.
135 00:13:14.670 ⇒ 00:13:25.169 Gabriel Lam: that sort of workflow. I’ve also got a couple PRDs, I think, more for UTAM, but just thinking of, like, other features that we can have. So we talked about, like, the mini-podcast yesterday of.
136 00:13:25.170 ⇒ 00:13:25.520 Samuel Roberts: Yeah.
137 00:13:25.520 ⇒ 00:13:28.380 Gabriel Lam: How do we, you know, ingest,
138 00:13:28.600 ⇒ 00:13:34.240 Gabriel Lam: Or, like, you know, communicate information. Another thing I was thinking of was, like, a sort of…
139 00:13:34.540 ⇒ 00:13:37.360 Gabriel Lam: Personalized, like, dashboard.
140 00:13:37.460 ⇒ 00:13:44.960 Gabriel Lam: just, I just noticed people are like, okay, I’m looking at linear, this is what I do for today, but, like, what are the priorities today? What are, you know…
141 00:13:44.960 ⇒ 00:13:45.650 Samuel Roberts: I don’t know what?
142 00:13:45.650 ⇒ 00:13:53.120 Gabriel Lam: back when Ricoh had Slack messages for, like, hey, these are the priorities today, I think there was something helpful for just people to have an idea of, like.
143 00:13:53.820 ⇒ 00:13:57.610 Gabriel Lam: how do I rank my tickets aside from priorities and, like, you know, that kind of stuff.
144 00:13:58.000 ⇒ 00:14:07.129 Gabriel Lam: So I’m also finishing up a PRD for that, just thinking through, maybe it is linear, and just figuring out how to make it work for us, or…
145 00:14:08.530 ⇒ 00:14:12.120 Gabriel Lam: You know, is there a way to really, like, take what we have in the sign-up assistant?
146 00:14:12.320 ⇒ 00:14:16.230 Gabriel Lam: And then… like, improve that. So…
147 00:14:16.230 ⇒ 00:14:16.650 Samuel Roberts: Yeah.
148 00:14:16.650 ⇒ 00:14:18.150 Gabriel Lam: We’re in a planning phase, and…
149 00:14:18.150 ⇒ 00:14:18.560 Samuel Roberts: Okay.
150 00:14:18.700 ⇒ 00:14:25.820 Gabriel Lam: That’s sort of what I’m working on. And then, once you have the PR reviews done, I can hear your thoughts on that as well.
151 00:14:26.170 ⇒ 00:14:29.619 Samuel Roberts: Yeah, I merged the Slack notifications, because I was
152 00:14:30.050 ⇒ 00:14:34.699 Samuel Roberts: I attested to that, and I just… I guess I never merged it. And I’m looking at the,
153 00:14:35.150 ⇒ 00:14:37.770 Samuel Roberts: The deck one, I think, right now is what I was looking at.
154 00:14:39.220 ⇒ 00:14:44.089 Samuel Roberts: getting it running. It wasn’t building, so I gotta figure it out locally, but I’ll sort that out.
155 00:14:44.340 ⇒ 00:14:52.039 Samuel Roberts: Yeah, if you have any, if you want to bounce any ideas or anything while you’re working with the PRDs, or after, or anything, definitely let me know.
156 00:14:52.040 ⇒ 00:14:52.700 Gabriel Lam: Yeah.
157 00:14:52.870 ⇒ 00:14:55.120 Samuel Roberts: I have some thoughts there as well, like, I…
158 00:14:55.540 ⇒ 00:15:00.940 Samuel Roberts: Yeah, the linear thing is gonna be huge, if we can make that more…
159 00:15:01.220 ⇒ 00:15:06.899 Samuel Roberts: yeah, groomed and up-to-date, and keep that all on track, I think is gonna be…
160 00:15:08.080 ⇒ 00:15:17.489 Samuel Roberts: probably a good use of effort, and if we can make use from the stand-up assistant to set those priors, I think that might be a good way to do it.
161 00:15:18.750 ⇒ 00:15:22.779 Clarence Stone: Hey guys, this is, this is Utom, I’m with Clarence, actually.
162 00:15:22.780 ⇒ 00:15:24.560 Samuel Roberts: Oh, oh, hey! Okay, you are here.
163 00:15:24.560 ⇒ 00:15:25.230 Gabriel Lam: Okay.
164 00:15:25.230 ⇒ 00:15:41.500 Clarence Stone: I’m just listening. Oh, I’m listening, but working on some other stuff. I just wanted, like, maybe put a couple of thoughts, for this group, is one, yeah, we’re gonna be… there’s gonna be, kind of, PRs coming into the vault, the platform, into playbooks.
165 00:15:41.500 ⇒ 00:15:46.209 Samuel Roberts: So I do want to think about, like, how do we do a PR approval process?
166 00:15:46.210 ⇒ 00:15:55.669 Clarence Stone: I think maybe if we start with the vaults and playbooks, like, I think Gabe, as… really, what this is, is gonna be more related to, like, SLAs on PR approvals.
167 00:15:55.670 ⇒ 00:15:56.720 Gabriel Lam: So, like… Okay.
168 00:15:56.720 ⇒ 00:16:05.320 Clarence Stone: Gabe, my main thing is, like, it’s kind of, like, what my message to you today was, was like, do you want to be on the hook for reviewing? I would like someone to review those.
169 00:16:05.320 ⇒ 00:16:06.800 Gabriel Lam: Yeah.
170 00:16:06.800 ⇒ 00:16:07.230 Clarence Stone: from…
171 00:16:07.230 ⇒ 00:16:08.399 Gabriel Lam: to be on the hook for that.
172 00:16:08.400 ⇒ 00:16:11.190 Clarence Stone: on the Vault and the playbooks, I don’t know…
173 00:16:11.320 ⇒ 00:16:17.499 Clarence Stone: yet whether there is someone that can review. Like, the only two people that are kind of close to
174 00:16:17.590 ⇒ 00:16:37.590 Clarence Stone: coming up to speed on that stuff is, like, Clarence or Sheshu. But for now, I think, like, me and you are probably the only people, so, like, I just wanted to confirm if you’re okay to just continue to be assigned as, like, reviewers, because I do want you to see all of that. Okay, okay, cool. Then I’m just not… I’m just not gonna… I’m gonna try not to bug you, but… No problem.
175 00:16:37.620 ⇒ 00:16:41.079 Clarence Stone: I’m, like, doing a lot in there, so there will be PRs pretty often, and then…
176 00:16:41.080 ⇒ 00:16:41.810 Gabriel Lam: You’re good.
177 00:16:41.940 ⇒ 00:16:49.829 Clarence Stone: Okay, and then I think, yeah, between Sam, like, our normal, like, internal AI team, I’m kind of curious how we want to handle this.
178 00:16:49.910 ⇒ 00:17:02.390 Clarence Stone: Because you can see, kind of, the breadth of stuff that I’m pushing, and it’s just… and I’m only one person, so we’re gonna have multiple people push… start pushing to the platform, I’m sure.
179 00:17:02.500 ⇒ 00:17:04.109 Clarence Stone: So, I’m wondering, like.
180 00:17:04.480 ⇒ 00:17:10.860 Clarence Stone: I guess I’ll kind of throw a question out there, like, how do we want to do code owners and PR reviews?
181 00:17:11.730 ⇒ 00:17:13.740 Clarence Stone: First, for the, for platform stuff.
182 00:17:14.250 ⇒ 00:17:23.870 Samuel Roberts: Yeah, we actually had a similar discussion, around Lilo, for PR reviews, because, you know, we didn’t want necessarily everything being, like… we didn’t want, like, one point of failure, so, like, if I’m…
183 00:17:23.960 ⇒ 00:17:43.240 Samuel Roberts: offline or something, and something needs to get done, like… So the idea that I had thrown out yesterday was, depending on, you know, how big of a thing it is, and these PRs have been pretty big, so that might not be the same, but, like, for Leelo, the idea was, as long as, like, one of the other two people gets to take a look at it, like, we can PR probably into staging.
184 00:17:43.760 ⇒ 00:17:51.469 Samuel Roberts: for the platform, we may even have a little more leeway there, I suppose, so I feel like as long as we get other eyes and testing done, I feel like that’s where I care…
185 00:17:51.590 ⇒ 00:17:54.150 Samuel Roberts: The most.
186 00:17:54.490 ⇒ 00:17:56.440 Samuel Roberts: so I have…
187 00:17:56.820 ⇒ 00:17:59.510 Clarence Stone: I guess…
188 00:17:59.690 ⇒ 00:18:11.499 Clarence Stone: Yeah, I guess, like, I care a lot, but I’m not… I’m not on that client. So, in that sense, it’s a bit selfish for me, where I’m like, hey, I’m pushing shit to the platform, like, when can I expect PR?
189 00:18:11.500 ⇒ 00:18:15.650 Samuel Roberts: That’s what I’m saying, that’s what I’m saying, I feel like as long as we get… if we have a list of people that could do a review.
190 00:18:15.650 ⇒ 00:18:35.370 Clarence Stone: Yeah, but this is where, like, consider me, like, what… the thing I don’t want to do is, like, merge. And I wanted to go through PR review, but I… like, think about me, again, like I mentioned yesterday, is like, I’m gonna be the most tech-savvy business person using the… pushing PR as a platform that we’re ever gonna have.
191 00:18:35.410 ⇒ 00:18:51.010 Clarence Stone: So, if I have ambiguity, it’s gonna be, like, tenfold with everybody else. Fair enough, yeah. So one thing I’m kind of trying to role-play here is, like, okay, la-di-da, I’m, like, a business person at Brainforge, and I’m pushing stuff, like, what are the questions that I have?
192 00:18:51.010 ⇒ 00:19:00.809 Clarence Stone: And that’s why I’m kind of, like, a lot of these features I’m, like, trying to do in different ways and stress testing, like, how the system works. One of which is gonna be, like, hey, again, like.
193 00:19:00.820 ⇒ 00:19:06.269 Clarence Stone: I just made a pull request, like, what can I expect from the platform team, you know?
194 00:19:06.270 ⇒ 00:19:08.900 Samuel Roberts: Sure, sure, yeah. Okay, yeah, I mean, I think…
195 00:19:09.250 ⇒ 00:19:17.699 Samuel Roberts: maybe the best way to handle that is just, like, tagging me, and I can decide if it’s, like, I will review or assign it to someone else kind of thing.
196 00:19:18.110 ⇒ 00:19:21.660 Samuel Roberts: But, like, it has to go through, I guess, someone to make sure that that review happens, right?
197 00:19:21.930 ⇒ 00:19:32.769 Samuel Roberts: And if there’s a single point that someone can focus on, as long as it’s not, like, a dire, critical bug fix… Sure. I think then… I think then that system can probably be okay. I mean…
198 00:19:33.350 ⇒ 00:19:35.430 Samuel Roberts: The other side of this,
199 00:19:35.840 ⇒ 00:19:49.829 Samuel Roberts: something I was talking a little bit about yesterday was just, like, you know, is this holding up other work? Like, we need to kind of have that sense. How robust is the PR description and stuff? Because I was looking at a few of that… like, the deck, I… I was a little confused about how to even test it at first, and I finally found that in the description, but…
200 00:19:50.180 ⇒ 00:19:55.840 Samuel Roberts: But the deck one, I think I should have just made a draft PR. I said, like, don’t worry about that, like, because…
201 00:19:55.840 ⇒ 00:19:56.720 Clarence Stone: 30 gallon.
202 00:19:57.130 ⇒ 00:20:08.169 Samuel Roberts: Yeah, yeah, yeah, no, you’re right, I didn’t realize that until I was digging through the PRs, but yes. The other side of it is, like, do we want… if people are gonna make things and show, like, do we require a loom to show how to test it and everything and make sure that things.
203 00:20:08.170 ⇒ 00:20:10.230 Clarence Stone: Yeah, I feel like yes.
204 00:20:10.230 ⇒ 00:20:19.660 Samuel Roberts: As I was digging, I’m like, oh, this is great, but, like, I’m not really sure what I’m looking at until I dig into the code, and that can take a minute. So I think, like, a loom.
205 00:20:20.350 ⇒ 00:20:37.580 Clarence Stone: But this is where, like, I’m kind of like, you guys, as I mentioned, like, you know, even when we talked, like, two months ago or so, like, y’all I want involved in the process because you’re gonna be gatekeeping and then also maintaining whatever gets in. Like, it’s gonna be very hard to require that of the business person to, like.
206 00:20:37.580 ⇒ 00:20:38.000 Samuel Roberts: Totally.
207 00:20:38.000 ⇒ 00:20:52.780 Clarence Stone: manage and understand, like, the deployment and, like, bug side, but yes, I think it’s… you guys should, like, set whatever requirements you need on the PR side, and enforce it, you know? Yeah.
208 00:20:52.780 ⇒ 00:20:53.360 Gabriel Lam: Yep.
209 00:20:53.360 ⇒ 00:21:13.279 Clarence Stone: treat… again, like, you have an opportunity to treat our internal stuff as a… as a client, where… but again, like, for me, I’m gonna push us to be the most AI-enabled client that we support, which means, like, okay, how do we get to a point where we’re deploying and pushing stuff daily? How do we get to the point where we’re like, yes, we have looms.
210 00:21:13.280 ⇒ 00:21:16.139 Clarence Stone: People are vibe coding a lot more.
211 00:21:16.140 ⇒ 00:21:24.849 Clarence Stone: And then there’s also these checks and balances, so, like, I’m game for what, like, for whatever you guys decide. I think more importance to me is that y’all decide instead of me.
212 00:21:25.140 ⇒ 00:21:41.449 Samuel Roberts: Yeah, yeah, no, I was… I was definitely giving these… these things thoughts as I was doing the PR, because I was like, a loom would be helpful, so I think something like that. Pr descriptions were pretty good, but obviously, like, make sure that it has that, like, a few things like that. And then as we… we can add some more testing and stuff, too, then it has to pass a few things before, because there were some…
213 00:21:41.550 ⇒ 00:21:49.630 Samuel Roberts: like, bug bot comments and stuff, and I was like, should I just hunt this back to you to fix those, or should I… you know, there’s a few other things.
214 00:21:49.790 ⇒ 00:21:51.730 Samuel Roberts: How… how technical are these people gonna be?
215 00:21:51.730 ⇒ 00:21:52.160 Clarence Stone: Yeah.
216 00:21:52.630 ⇒ 00:21:53.220 Clarence Stone: Exactly.
217 00:21:53.220 ⇒ 00:21:55.559 Samuel Roberts: So yeah, I think,
218 00:21:56.030 ⇒ 00:22:01.939 Samuel Roberts: Yeah, I can throw some thoughts together on how we want that. Maybe, Gabe, you and I can discuss that.
219 00:22:01.940 ⇒ 00:22:04.380 Gabriel Lam: Yeah, let me grab some time with you today.
220 00:22:04.380 ⇒ 00:22:05.820 Samuel Roberts: That’d be perfect, yeah, we’ve got a few things.
221 00:22:05.820 ⇒ 00:22:06.140 Gabriel Lam: Yeah.
222 00:22:06.140 ⇒ 00:22:06.740 Samuel Roberts: Yes, man.
223 00:22:07.840 ⇒ 00:22:08.710 Samuel Roberts: Okay.
224 00:22:10.220 ⇒ 00:22:13.159 Samuel Roberts: Other thoughts there, internally?
225 00:22:14.410 ⇒ 00:22:15.570 Gabriel Lam: That’s all for now, yeah.
226 00:22:15.760 ⇒ 00:22:17.130 Samuel Roberts: Alright, cool, cool.
227 00:22:19.240 ⇒ 00:22:25.010 Samuel Roberts: Great. And then, Lilo, I don’t know, Renov is, but,
228 00:22:26.000 ⇒ 00:22:29.260 Samuel Roberts: We got, everything into staging,
229 00:22:29.570 ⇒ 00:22:37.080 Samuel Roberts: I think he updated the reports and everything, so I think we’re pretty much good to push to production and let them start actually, like, getting people on board.
230 00:22:37.230 ⇒ 00:22:41.059 Samuel Roberts: I think,
231 00:22:44.850 ⇒ 00:22:51.600 Samuel Roberts: what was the other thing? Oh, the Shopify data, need to grab some time today with a way…
232 00:22:52.070 ⇒ 00:22:57.569 Samuel Roberts: To discuss some data warehouse things and get back to them.
233 00:22:59.020 ⇒ 00:23:00.470 Samuel Roberts: I’m trying to think what else…
234 00:23:03.050 ⇒ 00:23:06.440 Samuel Roberts: Yeah, Casey, am I missing anything else that we discussed yesterday or covered?
235 00:23:07.670 ⇒ 00:23:12.590 Casie Aviles: Yeah, it was mostly, like, the next steps for the data warehouse.
236 00:23:12.590 ⇒ 00:23:13.210 Samuel Roberts: Yeah.
237 00:23:13.650 ⇒ 00:23:18.989 Samuel Roberts: Yeah, oh, that’s right, I pulled down the Shopify data, I’m gonna get that into a database on…
238 00:23:19.930 ⇒ 00:23:21.530 Samuel Roberts: Dev or something.
239 00:23:22.190 ⇒ 00:23:25.129 Samuel Roberts: So we can start actually doing the forecast model,
240 00:23:27.300 ⇒ 00:23:46.709 Samuel Roberts: it was a little weird to pull that down, but Cursor… like, the… for context, you guys, the Shopify, docs have their own AI that wasn’t just asking questions, it was, like, writing code for me to actually do the stuff I wanted. Which at first was very cool, and then I got into Cursor to run it, and it… it didn’t work, so Cursor.
241 00:23:46.710 ⇒ 00:23:47.440 Casie Aviles: Fixed it.
242 00:23:47.440 ⇒ 00:23:51.620 Samuel Roberts: But it was just a few, like, weird versioning things, mostly.
243 00:23:52.390 ⇒ 00:24:01.209 Samuel Roberts: But it was… it was… it was actually, like, a really good starting point. And I was able to get the data out. It didn’t take too long for that, the bulk export and everything.
244 00:24:01.820 ⇒ 00:24:10.610 Samuel Roberts: So I’m gonna get that in, and we can make progress there with the forecasting app. I’ll probably have to get with Pranav to make sure that we’re on sync for that today.
245 00:24:10.810 ⇒ 00:24:17.380 Samuel Roberts: Yeah, I don’t know. Casey, anything else on Lila?
246 00:24:19.000 ⇒ 00:24:26.349 Casie Aviles: No, I think… We’re… I think we’re good with, but the image gen…
247 00:24:26.540 ⇒ 00:24:37.199 Samuel Roberts: Yeah, everything there seemed great, so he seemed very happy with everything. Things seemed to work, and like, no, you know, obviously there’s, like, a few things there, like the functionality and the outputs, and he seemed happy with everything, so…
248 00:24:38.340 ⇒ 00:24:43.409 Casie Aviles: So, I think, you know, if I can help in any way with the other…
249 00:24:43.720 ⇒ 00:24:49.150 Casie Aviles: the remaining POC stuff for forecasting, I can… I can probably take some on there.
250 00:24:49.620 ⇒ 00:24:56.139 Samuel Roberts: Okay, yeah, I would say, hopefully, I think OAS mentions a time I didn’t get a chance to respond yet.
251 00:24:56.270 ⇒ 00:24:57.889 Samuel Roberts: But…
252 00:24:57.890 ⇒ 00:24:59.079 Casie Aviles: can be there.
253 00:24:59.080 ⇒ 00:25:03.120 Samuel Roberts: Yeah, that would be… I was hoping, yeah, I was hoping we could all be on there to…
254 00:25:03.280 ⇒ 00:25:08.820 Samuel Roberts: give him all the context he needs, and you have more context into some of the data warehouses than even I do using them, so…
255 00:25:10.100 ⇒ 00:25:16.850 Samuel Roberts: I’ll try to get that on the calendar, and then we can make a decision about, like, tickets and stuff from there for…
256 00:25:17.400 ⇒ 00:25:23.299 Samuel Roberts: the forecast UI, and then there’s some Python code. I don’t know if you’ve seen the Google…
257 00:25:23.550 ⇒ 00:25:25.830 Samuel Roberts: collab notebook.
258 00:25:27.130 ⇒ 00:25:28.003 Samuel Roberts: The…
259 00:25:28.540 ⇒ 00:25:29.800 Casie Aviles: Did the client sit?
260 00:25:30.160 ⇒ 00:25:34.949 Samuel Roberts: Yeah, there’s a… so they have a spreadsheet that has a notebook that does a bunch of cohort analysis stuff.
261 00:25:35.950 ⇒ 00:25:37.330 Casie Aviles: I see, I see.
262 00:25:37.330 ⇒ 00:25:47.459 Samuel Roberts: And I think it, is kind of what we’re gonna be basically modeling this on, so I can share that with you if you want to just familiarize yourself with, like, what’s there. Don’t go too deep, but just to see it.
263 00:25:47.460 ⇒ 00:25:48.120 Casie Aviles: Okay.
264 00:25:48.600 ⇒ 00:25:53.830 Samuel Roberts: And then, you know, whether or not we do that there, or we do it in SQL, or however we end up doing it,
265 00:25:54.430 ⇒ 00:25:59.539 Samuel Roberts: at least for the POC, might be different than the final version, too, so we’ll figure that out, but… okay.
266 00:25:59.540 ⇒ 00:26:00.190 Casie Aviles: Okay, cool.
267 00:26:00.190 ⇒ 00:26:03.810 Samuel Roberts: Yeah, we’ll have a meeting today, we can, coordinate there.
268 00:26:04.050 ⇒ 00:26:08.790 Samuel Roberts: And then… Gabe, you’ll grab some time today as well.
269 00:26:10.270 ⇒ 00:26:16.579 Samuel Roberts: with me, and then, Mustafa, I owe you a review on the evals, right? Or a review…
270 00:26:18.570 ⇒ 00:26:23.579 Samuel Roberts: Okay, cool. Anything else, anyone needs from anyone else while we’re here?
271 00:26:26.790 ⇒ 00:26:27.180 Casie Aviles: I think that.
272 00:26:27.180 ⇒ 00:26:27.580 Samuel Roberts: Alright.
273 00:26:27.580 ⇒ 00:26:28.340 Casie Aviles: I have.
274 00:26:28.640 ⇒ 00:26:36.670 Mustafa Raja: Great. Yeah, for the emails, I would want to know what the next steps would be, because all of that is, local for now, right?
275 00:26:37.210 ⇒ 00:26:38.669 Samuel Roberts: Yeah, I mean, I think…
276 00:26:39.390 ⇒ 00:26:45.619 Samuel Roberts: Yeah, I can give some more feedback there. I mean, I guess part of it depends on how the evals went, like, what were the outputs?
277 00:26:45.620 ⇒ 00:26:46.210 Mustafa Raja: Okay.
278 00:26:46.560 ⇒ 00:26:51.549 Samuel Roberts: But, you know, whether or not we need to get that set up somewhere else, or just running it locally, like.
279 00:26:51.780 ⇒ 00:26:54.720 Samuel Roberts: I mean, what’s the TLDR of it right now? Like…
280 00:26:54.830 ⇒ 00:26:58.880 Samuel Roberts: Was it good results? Was it… Okay results?
281 00:26:58.880 ⇒ 00:27:02.839 Mustafa Raja: So I didn’t run it fully, oh, okay, okay.
282 00:27:03.120 ⇒ 00:27:06.140 Mustafa Raja: On the whole thing, because, we have, like,
283 00:27:06.140 ⇒ 00:27:08.090 Samuel Roberts: That’s a lot isn’t another thing. Okay, yeah.
284 00:27:08.500 ⇒ 00:27:09.990 Mustafa Raja: Each row would take about…
285 00:27:10.150 ⇒ 00:27:15.670 Mustafa Raja: 10-15 seconds, so… it’s going to be… the whole thing is going to take a lot of time, really.
286 00:27:16.070 ⇒ 00:27:20.240 Mustafa Raja: But I can, I can, you know, run it on a subset.
287 00:27:20.840 ⇒ 00:27:22.520 Mustafa Raja: And share a seat.
288 00:27:22.520 ⇒ 00:27:28.120 Samuel Roberts: Yeah, I’ll watch what you do the loom, and I’ll give you some thoughts there. Yeah, even if we want to just get it set up somewhere to run.
289 00:27:28.530 ⇒ 00:27:30.730 Samuel Roberts: I mean, so you said 10 to 15 seconds?
290 00:27:32.710 ⇒ 00:27:35.300 Samuel Roberts: Per row, and how many rows are in that data set?
291 00:27:35.300 ⇒ 00:27:36.710 Mustafa Raja: 1003.
292 00:27:37.030 ⇒ 00:27:37.950 Samuel Roberts: Oh, boy.
293 00:27:39.500 ⇒ 00:27:43.690 Mustafa Raja: Yeah, let me know if you want to, you know, reduce the rows or something.
294 00:27:44.490 ⇒ 00:27:46.100 Samuel Roberts: Yeah, we may want to just run it on all the time.
295 00:27:46.100 ⇒ 00:27:53.009 Mustafa Raja: But this was, like, I believe this would run on a weekly cadence, rather than, you know…
296 00:27:53.070 ⇒ 00:27:57.479 Samuel Roberts: Oh, of course, of course, if anything, yeah. Really, what I’m thinking right now is I just want to validate
297 00:27:57.620 ⇒ 00:27:59.919 Samuel Roberts: what we’ve done in Maestra is good.
298 00:28:00.730 ⇒ 00:28:07.890 Mustafa Raja: Okay, then what I would do is I would run it locally, generate a CSV for the whole thing.
299 00:28:08.940 ⇒ 00:28:11.969 Mustafa Raja: And then, we can take a look at that also.
300 00:28:11.970 ⇒ 00:28:19.700 Samuel Roberts: Yeah, I think you’re right, eventually we want, like, a weekly cadence of, like, reviewing the last week, things like that, but I think for now.
301 00:28:20.520 ⇒ 00:28:36.309 Samuel Roberts: The things I wanted out of this were to make sure that we could do evals, and we can in Monster, and then use those evals to validate our selection of the Monster architecture, or tweak it however we need. Then ongoing will be another thing, yes, of course.
302 00:28:36.770 ⇒ 00:28:37.520 Mustafa Raja: Yeah, okay.
303 00:28:37.520 ⇒ 00:28:46.250 Samuel Roberts: Yeah, if you want to run it locally, or if you want to set it up on something, we can… we can figure that out, but I think if you can do it locally and just let it run later or something.
304 00:28:46.950 ⇒ 00:28:49.310 Mustafa Raja: Yeah, I’m just going to run it now.
305 00:28:49.540 ⇒ 00:28:52.749 Mustafa Raja: leave it. Once it’s done, I’m going to share the CSV.
306 00:28:52.930 ⇒ 00:28:57.470 Samuel Roberts: Okay, hold on, let’s just do a quick sanity calculation. So you said 10 to… so 15 seconds times…
307 00:28:57.610 ⇒ 00:28:59.790 Samuel Roberts: 1,003 rows.
308 00:29:00.250 ⇒ 00:29:01.850 Samuel Roberts: Divided by 6.
309 00:29:03.600 ⇒ 00:29:06.160 Samuel Roberts: Oh, I’m off fire. We’re talking, like.
310 00:29:07.270 ⇒ 00:29:10.540 Samuel Roberts: Wait, hold on, so that’s seconds… Minutes…
311 00:29:10.880 ⇒ 00:29:13.479 Samuel Roberts: Hours. Okay, so, like, 4 hours, it looks like.
312 00:29:13.700 ⇒ 00:29:14.520 Mustafa Raja: Yeah.
313 00:29:16.060 ⇒ 00:29:20.159 Samuel Roberts: Okay
314 00:29:23.260 ⇒ 00:29:27.029 Samuel Roberts: Yeah, I guess if you’re good to let that go on your machine, otherwise, you know, if you want to run
315 00:29:27.390 ⇒ 00:29:28.000 Samuel Roberts: else.
316 00:29:28.540 ⇒ 00:29:32.069 Mustafa Raja: No, no, no, yeah, this can handle it.
317 00:29:32.420 ⇒ 00:29:35.180 Samuel Roberts: Okay. Yeah, I just don’t want it to, like, affect other work today if you need to, you know.
318 00:29:35.180 ⇒ 00:29:36.610 Mustafa Raja: Yeah, yeah.
319 00:29:37.090 ⇒ 00:29:45.049 Samuel Roberts: Even if you want to do it in, like, chunks or something, if you want to do, like, all the thumbs up, all the thumbs up, however you want to make it run, but yeah, let’s just… whatever the results, that’s cool. Okay.
320 00:29:45.320 ⇒ 00:29:46.080 Mustafa Raja: Okay.
321 00:29:46.620 ⇒ 00:29:50.210 Samuel Roberts: I just, yeah. Alright. I think we’re good, then.
322 00:29:51.370 ⇒ 00:29:51.870 Mustafa Raja: Yep.
323 00:29:51.870 ⇒ 00:29:52.480 Gabriel Lam: Awesome.
324 00:29:53.100 ⇒ 00:29:54.419 Samuel Roberts: Alright, thank you all.
325 00:29:54.920 ⇒ 00:29:55.780 Gabriel Lam: Thank you all.
326 00:29:55.970 ⇒ 00:29:57.550 Samuel Roberts: Alright, talk to y’all later.
327 00:29:57.820 ⇒ 00:29:58.710 Gabriel Lam: Parkway.
328 00:29:58.970 ⇒ 00:29:59.490 Samuel Roberts: Bye.
329 00:29:59.490 ⇒ 00:29:59.900 Mustafa Raja: Right.
330 00:29:59.900 ⇒ 00:30:00.580 Gabriel Lam: Bye.