Meeting Title: AI Service Office Hours Block Date: 2026-04-13 Meeting participants: Samuel Roberts, Mustafa Raja, Casie Aviles, Pranav
WEBVTT
1 00:00:12.460 ⇒ 00:00:13.390 Mustafa Raja: Hey.
2 00:00:17.050 ⇒ 00:00:17.840 Samuel Roberts: Mmm.
3 00:00:20.950 ⇒ 00:00:22.670 Samuel Roberts: -Oh, can you hear me?
4 00:00:23.120 ⇒ 00:00:23.720 Mustafa Raja: Yeah.
5 00:00:25.300 ⇒ 00:00:28.269 Samuel Roberts: I can’t hear you, one sec, I think it’s my headphones.
6 00:00:33.880 ⇒ 00:00:35.650 Samuel Roberts: Alright, I think I’m good now.
7 00:00:40.280 ⇒ 00:00:42.960 Samuel Roberts: Oh, no, now they’re just rebooting themselves? What?
8 00:00:44.080 ⇒ 00:00:45.100 Mustafa Raja: I can hear you.
9 00:00:46.420 ⇒ 00:00:49.690 Samuel Roberts: Yeah, I don’t know, maybe it’s just my audio out is bad, hold on.
10 00:00:50.270 ⇒ 00:00:50.960 Mustafa Raja: Right.
11 00:00:51.190 ⇒ 00:00:52.590 Samuel Roberts: Yeah, that’s the issue.
12 00:00:55.600 ⇒ 00:00:57.380 Samuel Roberts: Okay, now I think I’m good.
13 00:01:12.290 ⇒ 00:01:14.480 Samuel Roberts: Maybe I’m not good, why is it doing this?
14 00:01:18.740 ⇒ 00:01:20.120 Samuel Roberts: You guys can still hear me?
15 00:01:21.170 ⇒ 00:01:22.410 Mustafa Raja: Yeah, I do hear you.
16 00:01:22.410 ⇒ 00:01:28.669 Samuel Roberts: Okay, good, there we go. Yeah, for some reason, it selected my default speaker as my microphone, and I don’t know why.
17 00:01:30.850 ⇒ 00:01:33.550 Samuel Roberts: Anyway, how are you, Mustafa? Welcome back.
18 00:01:33.550 ⇒ 00:01:35.550 Mustafa Raja: Yeah, yeah, I’m doing good.
19 00:01:36.110 ⇒ 00:01:37.550 Samuel Roberts: Good trip.
20 00:01:37.550 ⇒ 00:01:41.919 Mustafa Raja: Yeah, super good trip. I want, I want this every year now.
21 00:01:41.920 ⇒ 00:01:43.740 Samuel Roberts: Yeah, I bet, I bet.
22 00:01:44.040 ⇒ 00:01:45.500 Samuel Roberts: That’s exciting.
23 00:01:46.020 ⇒ 00:01:46.410 Mustafa Raja: Yeah.
24 00:01:46.410 ⇒ 00:01:47.130 Samuel Roberts: That’s good.
25 00:01:48.820 ⇒ 00:01:50.139 Samuel Roberts: What was the highlight?
26 00:01:50.990 ⇒ 00:01:54.719 Mustafa Raja: Yeah, I just got to spend a lot of time with my family.
27 00:01:54.720 ⇒ 00:01:55.110 Samuel Roberts: No.
28 00:01:55.110 ⇒ 00:01:58.989 Mustafa Raja: places, and… Yeah, I needed that.
29 00:01:58.990 ⇒ 00:02:00.309 Samuel Roberts: Yeah, that’s great.
30 00:02:02.460 ⇒ 00:02:05.269 Samuel Roberts: Love… I love traveling with family and places. Yeah.
31 00:02:06.820 ⇒ 00:02:18.350 Samuel Roberts: I was just thinking about that, because I was… I was in California this weekend visiting a friend, and I was coming back, and I was flying by myself, and I was like, this is very different than when I have my wife and everyone with me, you know?
32 00:02:18.350 ⇒ 00:02:19.040 Mustafa Raja: Yeah.
33 00:02:20.990 ⇒ 00:02:22.620 Mustafa Raja: Cool. The flight wasn’t…
34 00:02:23.170 ⇒ 00:02:24.770 Samuel Roberts: Oh, yeah, first time flying, right?
35 00:02:24.770 ⇒ 00:02:25.790 Mustafa Raja: Yeah.
36 00:02:26.220 ⇒ 00:02:27.060 Samuel Roberts: Nice.
37 00:02:29.670 ⇒ 00:02:30.240 Samuel Roberts: Was it…
38 00:02:30.240 ⇒ 00:02:30.990 Mustafa Raja: Was it a direct flight?
39 00:02:30.990 ⇒ 00:02:31.590 Samuel Roberts: Wait.
40 00:02:32.070 ⇒ 00:02:33.209 Samuel Roberts: Through some more.
41 00:02:33.370 ⇒ 00:02:43.290 Mustafa Raja: Yes, so, the departure one was direct, the arrival one was, had a layover in… in Karachin, Pakistan.
42 00:02:43.290 ⇒ 00:02:44.170 Samuel Roberts: Oh, okay.
43 00:02:44.610 ⇒ 00:02:47.279 Mustafa Raja: Yeah, but it was only a 2-hour layover, so…
44 00:02:47.280 ⇒ 00:02:48.549 Samuel Roberts: Oh, that’s not bad, okay.
45 00:02:48.550 ⇒ 00:02:49.240 Mustafa Raja: Beautiful.
46 00:02:49.530 ⇒ 00:02:52.880 Mustafa Raja: I only bored in that time, and, you know…
47 00:02:52.880 ⇒ 00:02:53.510 Samuel Roberts: Yeah.
48 00:02:54.750 ⇒ 00:02:55.779 Samuel Roberts: Cool, cool.
49 00:02:57.170 ⇒ 00:02:58.700 Samuel Roberts: Casey, how you doing today?
50 00:03:00.380 ⇒ 00:03:01.539 Casie Aviles: Yeah, doing good.
51 00:03:01.650 ⇒ 00:03:05.660 Casie Aviles: Just had a chill weekend, as usual.
52 00:03:05.660 ⇒ 00:03:07.530 Samuel Roberts: Yeah, excellent, excellent.
53 00:03:08.460 ⇒ 00:03:14.969 Samuel Roberts: I need one of those soon, because I was flying back and forth to California, which is a long way from Ohio.
54 00:03:14.970 ⇒ 00:03:15.610 Casie Aviles: Oh, yeah.
55 00:03:15.610 ⇒ 00:03:17.270 Samuel Roberts: And I got in at, like.
56 00:03:18.660 ⇒ 00:03:21.430 Samuel Roberts: 2 AM this morning, by the time.
57 00:03:21.430 ⇒ 00:03:22.060 Casie Aviles: My flight’s got delayed.
58 00:03:22.060 ⇒ 00:03:23.470 Samuel Roberts: laid and everything.
59 00:03:23.840 ⇒ 00:03:31.970 Samuel Roberts: Of course, so… If I sound a little rough, it’s because I haven’t slept much, so I apologize.
60 00:03:33.230 ⇒ 00:03:35.590 Mustafa Raja: I just got… got back from…
61 00:03:36.090 ⇒ 00:03:39.709 Mustafa Raja: From an appointment with doctor, I had some fatigue.
62 00:03:39.710 ⇒ 00:03:40.930 Samuel Roberts: Oh, no.
63 00:03:41.150 ⇒ 00:03:44.009 Mustafa Raja: Yeah. But I’m good now. I’m good now.
64 00:03:44.010 ⇒ 00:03:44.790 Samuel Roberts: Okay.
65 00:03:45.100 ⇒ 00:03:46.350 Mustafa Raja: Yeah, that’s… that’s good.
66 00:03:46.830 ⇒ 00:03:47.740 Samuel Roberts: Okay.
67 00:03:49.180 ⇒ 00:03:50.499 Mustafa Raja: So, what’s been happening?
68 00:03:50.850 ⇒ 00:03:56.840 Samuel Roberts: Yeah, so, we made some pretty good progress on…
69 00:03:57.170 ⇒ 00:04:01.420 Samuel Roberts: Well, both ABC and Eden, I’m not really sure where we’re gonna plug you in yet,
70 00:04:01.680 ⇒ 00:04:04.919 Samuel Roberts: But I figure we can talk about, ABC.
71 00:04:05.190 ⇒ 00:04:13.010 Samuel Roberts: Right now… I mean kid. Sorry, I’m still reopening apps that were closed on my laptop.
72 00:04:14.580 ⇒ 00:04:17.160 Samuel Roberts: I don’t have linear open yet, let me get that open.
73 00:04:19.730 ⇒ 00:04:21.410 Samuel Roberts: Alright,
74 00:04:22.660 ⇒ 00:04:28.810 Samuel Roberts: So I was actually out Friday, too, so Casey, you might be able to give a better update on exactly where we are at this moment.
75 00:04:29.440 ⇒ 00:04:30.310 Casie Aviles: Oh, sure.
76 00:04:30.640 ⇒ 00:04:36.199 Casie Aviles: Let’s see, so I can also just quickly share my screen to show…
77 00:04:36.200 ⇒ 00:04:37.360 Samuel Roberts: Yeah, that’d be perfect, yeah.
78 00:04:40.500 ⇒ 00:04:42.059 Casie Aviles: Oh, okay.
79 00:04:49.200 ⇒ 00:04:51.250 Casie Aviles: You’re seeing my screen now, Earl.
80 00:04:51.250 ⇒ 00:04:52.420 Samuel Roberts: Yep, yep, you see.
81 00:04:52.800 ⇒ 00:04:54.530 Samuel Roberts: Yeah, I didn’t even bother.
82 00:04:55.730 ⇒ 00:05:03.379 Casie Aviles: So, basically, we’ve been doing, the central dock co-pilot work, we focused on the…
83 00:05:03.820 ⇒ 00:05:08.909 Casie Aviles: M3 milestone, and it’s… essentially what we’re doing here is we’re…
84 00:05:09.340 ⇒ 00:05:11.830 Casie Aviles: Listening to comments that are made.
85 00:05:12.340 ⇒ 00:05:17.909 Casie Aviles: So they’re… so the CSRs, or… sorry, maybe not the CSRs, but the users will be…
86 00:05:18.490 ⇒ 00:05:20.319 Casie Aviles: It’s, like, submitting…
87 00:05:21.210 ⇒ 00:05:28.290 Casie Aviles: Like, this structured comment here, if they want to make changes, or they want to propose changes to the central tool, but…
88 00:05:28.520 ⇒ 00:05:37.780 Casie Aviles: And then we have, like, an automation that’ll essentially listen to… to those. So, we were able to, like, get that, working, like, the… at least the MVP of that.
89 00:05:38.300 ⇒ 00:05:44.269 Casie Aviles: And yeah, thank you, Sam, as well, for, like, the work on the placement, or the central dock placement, that was…
90 00:05:44.270 ⇒ 00:05:45.030 Samuel Roberts: Yeah.
91 00:05:45.860 ⇒ 00:05:51.079 Casie Aviles: Yeah, the POC was helpful, and I was able… we were able to get, like.
92 00:05:51.540 ⇒ 00:05:57.060 Casie Aviles: some decent, like, results from that. I think the next steps here would be, like, having
93 00:05:57.470 ⇒ 00:06:02.699 Casie Aviles: As you mentioned last week, the preview of the change that will be.
94 00:06:02.700 ⇒ 00:06:03.410 Samuel Roberts: Yeah.
95 00:06:03.810 ⇒ 00:06:06.199 Casie Aviles: Like, yeah, that will happen, so…
96 00:06:06.310 ⇒ 00:06:12.359 Casie Aviles: That’s something I’ve yet to add. Last week, it was kind of weird, there’s, like, this weird…
97 00:06:13.120 ⇒ 00:06:18.960 Casie Aviles: timing issue that I’ve been experiencing, and I wasn’t entirely sure why that happened, but…
98 00:06:19.120 ⇒ 00:06:22.550 Casie Aviles: Good thing in the demo that…
99 00:06:22.680 ⇒ 00:06:28.539 Casie Aviles: it didn’t come up, but I would notice, like, sometimes the automation would just take, like.
100 00:06:28.930 ⇒ 00:06:32.579 Casie Aviles: A long time to respond, like, a couple of minutes.
101 00:06:33.290 ⇒ 00:06:36.299 Casie Aviles: So sometimes it takes around…
102 00:06:36.760 ⇒ 00:06:41.230 Casie Aviles: Five, or even more than 5… And I wasn’t entirely sure.
103 00:06:41.860 ⇒ 00:06:46.629 Casie Aviles: Yeah, like, normally it would just respond in, like, A few seconds.
104 00:06:47.500 ⇒ 00:06:47.970 Samuel Roberts: Hmm.
105 00:06:47.970 ⇒ 00:06:51.669 Casie Aviles: almost instantly. But, yeah, that was kind of…
106 00:06:51.990 ⇒ 00:06:55.909 Casie Aviles: I wasn’t sure, it was stressing me out, but… Yeah, yeah.
107 00:06:57.120 ⇒ 00:06:57.720 Casie Aviles: Yeah.
108 00:06:58.170 ⇒ 00:07:10.339 Samuel Roberts: I’m not super worried about the timing here, because I feel like it might not be something that needs to be acted on immediately. You know, we have, like, it’ll only be batched per day and stuff, so I’m hoping that that means it doesn’t need to be as quick as Andy, but…
109 00:07:10.340 ⇒ 00:07:10.860 Casie Aviles: Okay, huh.
110 00:07:10.860 ⇒ 00:07:15.479 Samuel Roberts: Still, I don’t like more than 5 minutes, seems, for this sort of work, seems crazy, but…
111 00:07:15.970 ⇒ 00:07:17.940 Samuel Roberts: I guess we can dig into that if we need.
112 00:07:21.340 ⇒ 00:07:26.209 Casie Aviles: But yeah, I think, I think that’s about it,
113 00:07:26.660 ⇒ 00:07:30.179 Casie Aviles: Yeah, that’s… that’s for the CDC workflow.
114 00:07:31.620 ⇒ 00:07:33.280 Samuel Roberts: Yeah, so,
115 00:07:34.190 ⇒ 00:07:44.380 Samuel Roberts: Basically, there were two main workflows. We kind of front-loaded it with this central doc co-pilot, and then there’s some work with the transcripts that is probably coming up now.
116 00:07:46.700 ⇒ 00:07:47.460 Samuel Roberts: So, why not.
117 00:07:47.460 ⇒ 00:07:47.810 Casie Aviles: Figure.
118 00:07:47.810 ⇒ 00:07:50.870 Samuel Roberts: exactly where to plug Mustafa in there.
119 00:07:51.350 ⇒ 00:07:52.349 Samuel Roberts: I hate Pranav.
120 00:07:52.940 ⇒ 00:07:54.260 Pranav: Hey guys, can you hear me?
121 00:07:54.750 ⇒ 00:07:55.729 Samuel Roberts: Yes, we can.
122 00:07:56.010 ⇒ 00:07:56.600 Pranav: Perfect.
123 00:07:57.430 ⇒ 00:07:58.609 Pranav: Are you doing?
124 00:07:58.950 ⇒ 00:08:01.770 Pranav: Pretty good, pretty good. Welcome back, Mustafa.
125 00:08:02.620 ⇒ 00:08:03.689 Mustafa Raja: Thank you.
126 00:08:04.580 ⇒ 00:08:05.200 Pranav: Yeah.
127 00:08:05.490 ⇒ 00:08:13.100 Pranav: I think last week? Mustafa, you were gone from mid of the week prior, right?
128 00:08:13.340 ⇒ 00:08:15.899 Mustafa Raja: Yes, from yesterday, yes.
129 00:08:16.260 ⇒ 00:08:22.669 Pranav: Right, right, okay, cool. Yeah, Sam, it sounds like you’re kind of just going through linear. How’s…
130 00:08:22.980 ⇒ 00:08:27.920 Samuel Roberts: Yeah, yeah, well, I was on Friday, too, so Casey was just giving a quick update on where we were, there.
131 00:08:27.920 ⇒ 00:08:36.429 Pranav: Okay, cool. Yeah, so Casey gave me a good update on Friday on some of the automation work that’s been completed, so…
132 00:08:36.530 ⇒ 00:08:43.490 Pranav: It looks like the placement automation is working. I love the way that it looks. Janice was super excited about that. Oh, he goes.
133 00:08:43.490 ⇒ 00:08:44.899 Samuel Roberts: I’m curious to hear how it went, yeah.
134 00:08:44.900 ⇒ 00:08:50.510 Pranav: Yeah, and they’re already noticing improvements on the workflow, and so…
135 00:08:50.820 ⇒ 00:08:57.319 Pranav: Casey, it also looks like… like, that tagging stuff, too. Janiece’s… Janiece has seen,
136 00:08:58.080 ⇒ 00:09:07.860 Pranav: I want to make sure that it’s not just like a… maybe… because what she noticed, basically, from her end was just that things look a lot cleaner and linear.
137 00:09:07.980 ⇒ 00:09:19.459 Pranav: Now, I don’t know how much of that is just based on, you know, maybe manual work that we did, versus just, like, because, like, the automation-wise, I don’t know what is actually already in place,
138 00:09:20.420 ⇒ 00:09:26.599 Pranav: because you showed me, like, the placement automation, but I don’t know if we went over how that tagging…
139 00:09:26.710 ⇒ 00:09:27.480 Pranav: So…
140 00:09:27.480 ⇒ 00:09:28.190 Casie Aviles: Yeah.
141 00:09:28.710 ⇒ 00:09:34.169 Pranav: is in place for, like, incoming triage tickets. Is that already fully functional?
142 00:09:35.050 ⇒ 00:09:47.369 Casie Aviles: No, like, we’re still… we have… I haven’t been tagging, like, those, triage tickets that come in, just the… whenever someone makes a comment, that’s when something happens, basically.
143 00:09:48.110 ⇒ 00:09:50.019 Pranav: Gotcha. Okay, so I think…
144 00:09:50.020 ⇒ 00:09:50.580 Mustafa Raja: Boy.
145 00:09:51.150 ⇒ 00:09:55.740 Pranav: Yeah, so, Sam, this is, like, one thing that I’m noticing, is that…
146 00:09:56.160 ⇒ 00:09:58.670 Pranav: I think how we need to be thinking about…
147 00:09:59.250 ⇒ 00:10:18.449 Pranav: I think there’s one level of, like, when we’re first scoping out a project, we think about the technical approach, but then there’s also another… and maybe we bake this into just, like, one type of document. That’s, like, a step-by-step document, which is, like, yeah, there’s an overarching technical approach, but then there’s also, like, the individual…
148 00:10:18.610 ⇒ 00:10:21.219 Pranav: Steps that need to be taken place.
149 00:10:21.430 ⇒ 00:10:24.759 Pranav: Like, for… for the workflow to be…
150 00:10:25.450 ⇒ 00:10:33.089 Pranav: for the technical approach to be fully implemented. And so, like, for example, I don’t know if there’s a ticket specifically for.
151 00:10:33.340 ⇒ 00:10:33.880 Mustafa Raja: Boy.
152 00:10:34.750 ⇒ 00:10:37.770 Pranav: Updating the tagging mechanism for…
153 00:10:39.780 ⇒ 00:10:46.779 Pranav: incoming triage tickets, and I don’t know if that’s… that’s, like, being… that’s encompassed in a specific…
154 00:10:46.900 ⇒ 00:10:51.189 Pranav: Maybe that’s, like, a sub-task within a task? Maybe not…
155 00:10:51.590 ⇒ 00:10:59.259 Pranav: not technically a subtask in linear, but just, like, you know what I mean, like, maybe a bullet point in the acceptance criteria in a linear ticket.
156 00:11:01.110 ⇒ 00:11:04.360 Pranav: So, I think these are things that we… you know.
157 00:11:04.480 ⇒ 00:11:07.170 Pranav: This is, like, it’s a totally new process, right, for…
158 00:11:07.320 ⇒ 00:11:10.159 Pranav: Building all the winner tickets, so…
159 00:11:10.550 ⇒ 00:11:22.270 Pranav: maybe what this means is just, like, every Monday, we kind of look at the linear tickets, we have a discussion about, like, hey, what are the gaps based on what we understand, and then we update the linear tickets or create new ones.
160 00:11:22.370 ⇒ 00:11:28.750 Pranav: Because, yeah, Casey, do you feel like you have a linear ticket for updating The incoming triage tickets?
161 00:11:29.060 ⇒ 00:11:33.790 Pranav: To, like, follow suit with the tags that… We came up with.
162 00:11:35.130 ⇒ 00:11:42.369 Casie Aviles: I, I know that we have the 3.1 tickets, I think it should, yeah, there are slick…
163 00:11:42.740 ⇒ 00:11:51.560 Casie Aviles: I guess what’s… what wasn’t clear for me initially was if… whether we should… how… how much of the tagging or, like, the state machine should I…
164 00:11:51.870 ⇒ 00:11:56.199 Casie Aviles: implement. So, what I did was just a partial one.
165 00:11:56.480 ⇒ 00:12:00.809 Casie Aviles: Yeah. Based on the document that, that we have,
166 00:12:00.990 ⇒ 00:12:03.560 Casie Aviles: That, like, outlines the whole triage intake.
167 00:12:04.170 ⇒ 00:12:07.060 Casie Aviles: But yeah, that part was…
168 00:12:07.570 ⇒ 00:12:16.080 Casie Aviles: Yeah, I probably glossed over that part, like, adding the new tags, or, like, labels for, like, incoming tickets, because
169 00:12:16.300 ⇒ 00:12:18.630 Casie Aviles: what I implemented was just for the…
170 00:12:19.220 ⇒ 00:12:22.589 Casie Aviles: comment spark, you know, when the automation for that.
171 00:12:22.590 ⇒ 00:12:23.750 Samuel Roberts: Yeah, I think.
172 00:12:23.750 ⇒ 00:12:24.140 Casie Aviles: comment.
173 00:12:24.140 ⇒ 00:12:41.009 Samuel Roberts: I think… I think you’re right, Casey. It was kind of baked into those, like, 3.1, 3. I wasn’t explicit enough, but I think also, like, the… when… when Casey and I were talking, like, most of… the bulk of the work was doing the rest… like, the work that happened, and, like, assigning that… that…
174 00:12:41.140 ⇒ 00:12:45.259 Samuel Roberts: That ticket, probably, or the, the label, when it first comes in, is…
175 00:12:45.750 ⇒ 00:12:48.399 Samuel Roberts: I don’t want to say trivial, but compared to what the other stuff was.
176 00:12:48.400 ⇒ 00:12:50.479 Pranav: Takes… you could probably do it in a few minutes on it.
177 00:12:50.480 ⇒ 00:12:56.469 Samuel Roberts: Yeah, exactly. So I think it kind of, like, that whole state machine got defined, and then the complexity got worked on, if that makes.
178 00:12:56.470 ⇒ 00:12:59.340 Pranav: Yeah, what I’m kind of seeing is, like, you know.
179 00:12:59.850 ⇒ 00:13:08.350 Pranav: Cursor is gonna be able to build out 95% of it before it to actually be, like, operational, and, like, the client is gonna see, like.
180 00:13:08.520 ⇒ 00:13:17.049 Pranav: the full benefit of things is, like, us putting in that 5%. So, like, maybe that’s just where we fill in the gaps, and it’s not necessarily building linear tickets. It’s, like, maybe.
181 00:13:17.050 ⇒ 00:13:23.770 Samuel Roberts: I mean, I think something like this is where we might say, like, okay, we need to make sure we add a specific ticket, because we noticed that that didn’t happen.
182 00:13:23.770 ⇒ 00:13:24.360 Pranav: Yeah.
183 00:13:24.600 ⇒ 00:13:27.609 Samuel Roberts: I think that’s fine. I don’t see any issue with that. Okay.
184 00:13:28.050 ⇒ 00:13:30.820 Pranav: So then, yeah, let’s maybe… let’s… let’s just…
185 00:13:31.180 ⇒ 00:13:37.619 Pranav: if this is one example where it’s, like, already encompassed in the ticket, then that’s great. Because, like, I think we just need to have the conversation, and then…
186 00:13:37.620 ⇒ 00:13:42.540 Samuel Roberts: Yeah, yeah, no, this is not… Make sure that we know that. This is why we can’t just, like, make the plan and feed it to cursor, so…
187 00:13:42.540 ⇒ 00:13:51.659 Pranav: 100%, 100%. Totally agree. Okay, did you guys already discuss, Casey and Mustafa, like, what you guys are gonna be working on this week?
188 00:13:52.490 ⇒ 00:13:53.250 Samuel Roberts: No, I think we’re kind of…
189 00:13:53.250 ⇒ 00:13:53.970 Pranav: Claire design.
190 00:13:53.970 ⇒ 00:13:55.699 Samuel Roberts: to that, so yeah, if you have thoughts…
191 00:13:55.700 ⇒ 00:14:06.190 Pranav: Cool. Let’s maybe just start with, what is the milestone for this Friday. So, honestly, let me just go to Notion real quick, so I can just see the latest.
192 00:14:06.410 ⇒ 00:14:07.010 Samuel Roberts: Okay.
193 00:14:08.010 ⇒ 00:14:09.960 Pranav: Alright…
194 00:14:11.020 ⇒ 00:14:21.959 Pranav: Yeah, so what we had originally scoped was for last week to complete the duplicate conflict engine, duplicate slash conflict engine as well.
195 00:14:22.730 ⇒ 00:14:30.000 Pranav: And so, at the end of this week was our proposed Deadline for completing the co-pilot.
196 00:14:31.780 ⇒ 00:14:43.030 Pranav: And so, I still feel pretty good about that. You know, I feel like in the next couple days we can get this duplicate slash conflict engine, you know, going. Sam, I know we had some discussion about…
197 00:14:43.360 ⇒ 00:14:48.759 Pranav: whether, using the GWCLI itself.
198 00:14:49.210 ⇒ 00:14:53.879 Pranav: would… or, you mentioned something how we didn’t necessarily need to use a rag.
199 00:14:53.990 ⇒ 00:14:56.550 Pranav: Like, we didn’t have to use the embeddings.
200 00:14:56.550 ⇒ 00:15:01.660 Samuel Roberts: Yeah, I looked around at that a little bit. Yeah, I looked around at that, and I played with it. Rags seemed.
201 00:15:01.660 ⇒ 00:15:04.289 Pranav: Did you get that POC working, or…
202 00:15:04.290 ⇒ 00:15:04.909 Samuel Roberts: Yeah. So that’s what.
203 00:15:04.910 ⇒ 00:15:05.230 Pranav: Yeah.
204 00:15:05.230 ⇒ 00:15:12.310 Samuel Roberts: that’s what Casey kind of built from… for the rest of it. For… that was for identifying where…
205 00:15:12.960 ⇒ 00:15:21.350 Samuel Roberts: Where in the document… so, like, once the ticket gets a comment, and that comment gets ingested, and we kind of understand the change.
206 00:15:21.760 ⇒ 00:15:29.810 Samuel Roberts: I put together a POC on finding that where in the document, and I think the rag actually ended up working better than I was thinking.
207 00:15:29.810 ⇒ 00:15:30.579 Pranav: Okay, cool.
208 00:15:30.580 ⇒ 00:15:42.170 Samuel Roberts: When I chatted with Cursor a bunch about some of the other ways of doing it, some of the downsides were not just the speed, but also just, like, the back and forth, the token usage, the non-deterministic nature of it.
209 00:15:44.180 ⇒ 00:15:53.810 Samuel Roberts: Cool. And so the rag, I think, because the way we’re trunking it, worked relatively well. I mean, there’s probably refinements that we can make, or maybe we could add another step if it’s really hard to…
210 00:15:54.030 ⇒ 00:15:56.030 Samuel Roberts: Fine in the rag, but it actually worked.
211 00:15:56.280 ⇒ 00:15:59.409 Samuel Roberts: Better than I was thinking it would.
212 00:16:01.240 ⇒ 00:16:10.530 Pranav: Okay, cool. So I’m adding to our working session on Wednesday to do an internal QA of the duplicate and the placement automation.
213 00:16:10.760 ⇒ 00:16:14.430 Pranav: I just want us to, like, in that session, maybe…
214 00:16:14.730 ⇒ 00:16:25.570 Pranav: since Casey, you’re the one working on this, the… the three of us, me, Sam, Mustafa, we can come into it with some ideas of, like, how can we stress test this, so, like.
215 00:16:25.570 ⇒ 00:16:25.950 Samuel Roberts: Yeah.
216 00:16:25.950 ⇒ 00:16:29.540 Pranav: Edge cases are… Are, taken care of.
217 00:16:30.480 ⇒ 00:16:31.020 Casie Aviles: Sure.
218 00:16:34.810 ⇒ 00:16:42.870 Samuel Roberts: Yeah, let me pull up… this cycle. So what do we… so we have the… So where, I guess.
219 00:16:43.520 ⇒ 00:16:46.890 Samuel Roberts: We still have to implement the duplicate and the conflict stuff, and…
220 00:16:47.300 ⇒ 00:16:47.820 Pranav: Yep.
221 00:16:48.300 ⇒ 00:16:56.569 Samuel Roberts: And we also were talking earlier about how do we show, back in linear where the change will be made best?
222 00:16:56.780 ⇒ 00:16:59.120 Samuel Roberts: How do you show best where the change will be made?
223 00:16:59.240 ⇒ 00:17:01.910 Samuel Roberts: Yeah, so I think…
224 00:17:01.910 ⇒ 00:17:11.920 Pranav: Activity is what we decided on, right? So, that’s what Casey showed with the placement automation, that there’s that Brainforge AI automation bot that basically comments.
225 00:17:12.190 ⇒ 00:17:13.079 Samuel Roberts: Yes.
226 00:17:13.089 ⇒ 00:17:13.839 Pranav: After the form.
227 00:17:13.839 ⇒ 00:17:20.969 Samuel Roberts: The question was, is that enough to show, like… well, so, is that enough to then, like, approve the change, and then see the change, if that makes sense?
228 00:17:21.429 ⇒ 00:17:23.389 Samuel Roberts: Before it’s batched, right?
229 00:17:25.369 ⇒ 00:17:26.039 Casie Aviles: Yeah, like…
230 00:17:26.040 ⇒ 00:17:26.730 Samuel Roberts: information.
231 00:17:26.730 ⇒ 00:17:28.329 Casie Aviles: There’s no preview right now, yeah.
232 00:17:28.339 ⇒ 00:17:38.229 Samuel Roberts: That’s sort of what we’re wondering. Does it need to show, like, these words will be replaced in the central doc? Like, and it will look like this after what it was before? Or is it just…
233 00:17:39.049 ⇒ 00:17:44.079 Samuel Roberts: That’s where we were starting to talk about how do we get that kind of… that sort of preview.
234 00:17:46.880 ⇒ 00:17:48.450 Pranav: I see what you’re saying, okay.
235 00:17:51.440 ⇒ 00:18:04.420 Pranav: Yeah, so I guess, what is the technical approach for replacing, right? Because for replacing, there’s gonna be certain information in there where it’s like, hey, we don’t wanna…
236 00:18:04.800 ⇒ 00:18:09.210 Pranav: Check, duplicate, or conflicting against that, right?
237 00:18:09.440 ⇒ 00:18:12.639 Pranav: So, whenever we’re doing the RAG,
238 00:18:12.850 ⇒ 00:18:25.340 Pranav: when we’re using this RAG approach for, assessing if something’s duplicate or conflicting, we need to make sure that we don’t look at that specific text section that we’re trying to replace.
239 00:18:26.190 ⇒ 00:18:27.630 Samuel Roberts: Sure.
240 00:18:27.630 ⇒ 00:18:34.390 Pranav: Now, given that… If we’re able to… Not use that text section.
241 00:18:34.810 ⇒ 00:18:36.730 Pranav: For the duplicate, or…
242 00:18:37.190 ⇒ 00:18:49.150 Pranav: duplicate or conflicting information, then that means we’re able to isolate that, and we should be able to display that in the linear activity, right? So we can say, hey.
243 00:18:49.420 ⇒ 00:18:53.190 Pranav: This is what there’s currently, this is what… this is what we’re going to replace it with.
244 00:18:53.400 ⇒ 00:18:55.470 Pranav: And what we’re gonna replace it with is just what the…
245 00:18:55.690 ⇒ 00:19:00.200 Pranav: What the trainer’s gonna be providing in the… in the form.
246 00:19:02.220 ⇒ 00:19:02.970 Pranav: So…
247 00:19:03.180 ⇒ 00:19:03.920 Samuel Roberts: Right.
248 00:19:03.920 ⇒ 00:19:11.039 Pranav: how do you see… do you have a good understanding, Sam, of what that… what that, technical approach is gonna be?
249 00:19:11.180 ⇒ 00:19:12.039 Pranav: Or is that our…
250 00:19:12.040 ⇒ 00:19:12.359 Samuel Roberts: It’s weird.
251 00:19:12.360 ⇒ 00:19:13.520 Pranav: find. Yeah.
252 00:19:13.520 ⇒ 00:19:21.010 Samuel Roberts: That’s what we were kind of just talking about, like, what do we need to actually… like, right now, we’re pulling from the rag to identify where the change needs to be made.
253 00:19:21.140 ⇒ 00:19:25.630 Samuel Roberts: Or have the options for, you know, human approval of that.
254 00:19:25.820 ⇒ 00:19:39.919 Samuel Roberts: We weren’t sure then the next step would be actually, like, identifying… because that’s just the chunk, right? We then need to identify… find that chunk in the dock and make the actual, like, substitution, or addition, or deletion.
255 00:19:41.320 ⇒ 00:19:48.610 Pranav: Yeah, well, I guess what makes more sense to me is, like, in that form, they shouldn’t be the ones defining replace or add, right?
256 00:19:48.710 ⇒ 00:19:57.130 Pranav: That should be defined by us, or at least we should give the first pass of deciding this should be a replacement or an addition.
257 00:19:57.290 ⇒ 00:19:58.909 Pranav: Because if they’re the ones defining that…
258 00:19:58.910 ⇒ 00:20:04.740 Samuel Roberts: They’re saying just, like, they want the central doc to say this, and then we identify if it’s saying something else, or…
259 00:20:05.550 ⇒ 00:20:06.850 Samuel Roberts: How to update that?
260 00:20:07.280 ⇒ 00:20:17.909 Pranav: that makes more sense to me, right? Because, like, or else, then we’re going to just… if they’re saying replace, or they’re saying to add, like, add… let’s say they say to add, for example, then…
261 00:20:18.500 ⇒ 00:20:19.370 Pranav: are…
262 00:20:20.060 ⇒ 00:20:35.310 Pranav: we’re just gonna be like, no, you can’t add. I think the right way to… the better product design here would be don’t give them that option, and then let the final say go to Janiece or Yvette to say, this can be added, or this can be replaced.
263 00:20:37.350 ⇒ 00:20:45.370 Pranav: That way, like, also if they have to do… if they have to, say, replace or add, then that means they have to go into the central doc and assess what is currently in there.
264 00:20:45.580 ⇒ 00:21:02.470 Pranav: Which I think defeats the purpose of this a little bit, right? The best workflow would be just, like, they just fill in the form. They… something doesn’t… something, isn’t being answered properly, all they have to do is answer a few questions, and then everything else is taken care of. They don’t need to go to different places, look for certain information.
265 00:21:02.890 ⇒ 00:21:09.289 Samuel Roberts: Okay, so… So that changes the intake form a little bit then, right?
266 00:21:09.290 ⇒ 00:21:09.940 Pranav: Yeah.
267 00:21:10.420 ⇒ 00:21:10.960 Samuel Roberts: Okay.
268 00:21:10.960 ⇒ 00:21:19.350 Pranav: Wha… Yeah, so Casey, how are you… Currently, like…
269 00:21:20.540 ⇒ 00:21:24.350 Pranav: Where is that, that field, which is,
270 00:21:24.890 ⇒ 00:21:28.869 Pranav: what is that? It’s just basically, you’re… you’ll be able to…
271 00:21:29.230 ⇒ 00:21:34.170 Pranav: decide whether this is a replace or addition? Like, how are you using that for your placement automation?
272 00:21:35.530 ⇒ 00:21:45.050 Casie Aviles: Right now, based on the doc, like, there’s, like, a change type field that is, that the commenter would specify.
273 00:21:45.270 ⇒ 00:21:46.020 Casie Aviles: So, yeah.
274 00:21:46.020 ⇒ 00:21:46.450 Pranav: Right.
275 00:21:46.450 ⇒ 00:21:50.580 Casie Aviles: Right? Like, the current state of this is that they have to…
276 00:21:51.050 ⇒ 00:21:56.340 Casie Aviles: Specify, like, what kind of operation or what kind of change they want to happen.
277 00:21:57.250 ⇒ 00:22:00.069 Casie Aviles: So if we are going to…
278 00:22:00.730 ⇒ 00:22:05.220 Casie Aviles: Update this, then this should be handled by, like, the workflow.
279 00:22:05.570 ⇒ 00:22:08.740 Casie Aviles: Underneath, so we would not require this anymore.
280 00:22:08.740 ⇒ 00:22:16.780 Samuel Roberts: So we would just get rid of that, but we would still keep, like, the target, the rationale, and the proposed update, or change, or whatever content, yeah.
281 00:22:17.160 ⇒ 00:22:19.889 Casie Aviles: Yeah, these are the minimum fields that we have.
282 00:22:20.080 ⇒ 00:22:20.400 Samuel Roberts: Right.
283 00:22:20.400 ⇒ 00:22:20.960 Casie Aviles: Alright.
284 00:22:21.650 ⇒ 00:22:23.180 Samuel Roberts: And so, if we get rid of…
285 00:22:24.210 ⇒ 00:22:27.689 Samuel Roberts: Change type, we just have to identify that as part of the step instead of just…
286 00:22:28.510 ⇒ 00:22:29.250 Casie Aviles: using that.
287 00:22:29.250 ⇒ 00:22:32.169 Samuel Roberts: Yeah, okay, I think that’s doable once we get the…
288 00:22:32.840 ⇒ 00:22:40.850 Samuel Roberts: Once we have the info, we can do the RAG lookup, of the target… And possibly some of the…
289 00:22:41.280 ⇒ 00:22:45.490 Samuel Roberts: proposed content, depending on how it’s written, I guess. So maybe, yeah, maybe we just have to identify…
290 00:22:46.170 ⇒ 00:22:46.880 Samuel Roberts: where they’re.
291 00:22:46.880 ⇒ 00:22:50.289 Pranav: Change type should be, I think, like, an output field, right? Like…
292 00:22:50.290 ⇒ 00:22:51.339 Samuel Roberts: That’s what I’m saying, that’s what it would become.
293 00:22:51.340 ⇒ 00:22:51.690 Pranav: Yeah.
294 00:22:51.690 ⇒ 00:22:52.300 Samuel Roberts: Yeah.
295 00:22:52.300 ⇒ 00:22:59.630 Pranav: Yeah, which I think is good, like, our, duplicate and conflicting information automation should be…
296 00:22:59.760 ⇒ 00:23:07.459 Pranav: like, one of the… we should kind of have, like, a proposed solution, which should define a change type, whether it’s delete, whether it’s to add, whether it’s direct.
297 00:23:07.460 ⇒ 00:23:07.880 Samuel Roberts: If that makes.
298 00:23:07.880 ⇒ 00:23:08.230 Pranav: Place.
299 00:23:08.230 ⇒ 00:23:09.000 Samuel Roberts: Yeah.
300 00:23:09.330 ⇒ 00:23:15.570 Samuel Roberts: Yeah, right now we’re just using it to identify, but we probably don’t need that change type even for the identification step, do we, really?
301 00:23:17.130 ⇒ 00:23:17.880 Pranav: Yeah.
302 00:23:18.050 ⇒ 00:23:33.740 Pranav: Okay, cool. Just to let you guys know, I have a hard stop in, like, 7 minutes, so maybe it makes sense for you guys, like, fully scope out that linear ticket, make sure, like, the… that ticket… because I think we’re adding some stuff here that’s probably not defined there. Let’s maybe spend…
303 00:23:33.740 ⇒ 00:23:39.609 Pranav: after this call, once I drop off, to kind of update that ticket, just so, you know, we all just make sure we…
304 00:23:39.760 ⇒ 00:23:42.350 Pranav: We’re all aligned on what that automation should look like.
305 00:23:42.480 ⇒ 00:23:44.890 Pranav: But yeah, sorry, Sam, what you were about to say?
306 00:23:45.460 ⇒ 00:23:51.180 Samuel Roberts: I was just gonna say, I don’t… Casey, I’m not sure if we’re even using that change type as part of that lookup step right now anyway.
307 00:23:52.140 ⇒ 00:23:54.410 Casie Aviles: Yeah, I don’t think we’re using it. Okay.
308 00:23:54.410 ⇒ 00:23:59.850 Samuel Roberts: I think that should be good for… well, I just have to add another step to identify what kind of change type we think it needs to be. I think that’s fine.
309 00:23:59.850 ⇒ 00:24:04.460 Pranav: Good. Okay, cool, so that means the placement automation can just be easily fixed, then.
310 00:24:04.820 ⇒ 00:24:10.889 Samuel Roberts: Yeah, I think so. The change type was going to come into effect when we actually implemented the change, and so I think that’s…
311 00:24:10.890 ⇒ 00:24:11.230 Pranav: Yeah.
312 00:24:11.400 ⇒ 00:24:12.539 Samuel Roberts: Yeah, that’s fine.
313 00:24:13.110 ⇒ 00:24:14.629 Pranav: Nice. Okay, cool.
314 00:24:15.180 ⇒ 00:24:32.139 Pranav: And so, that’s the main stuff for ABC this week. So, we kind of… I think Wednesday we should think of, hey, everything is completed, and then during that working session, we can do our internal QA. We’ll probably find some things that we need to,
315 00:24:32.240 ⇒ 00:24:38.710 Pranav: fix in terms of this workflow, and then we’ll have the rest of the day Wednesday, Thursday to figure that out, and then…
316 00:24:39.150 ⇒ 00:24:39.920 Pranav: Yeah.
317 00:24:41.200 ⇒ 00:24:43.860 Pranav: I think we should be… we should be good then.
318 00:24:44.160 ⇒ 00:24:46.519 Samuel Roberts: So then for Wednesday, you’re saying we need to…
319 00:24:46.740 ⇒ 00:24:53.699 Samuel Roberts: Get the placement working, get the conflict and duplicate detection working as part of that.
320 00:24:53.950 ⇒ 00:24:55.119 Samuel Roberts: Yep. And then…
321 00:24:55.880 ⇒ 00:25:01.739 Samuel Roberts: Like, surface what that change is back to linear, and then approve, and update, and batch, and all that stuff.
322 00:25:02.560 ⇒ 00:25:03.750 Pranav: Yes, yes.
323 00:25:06.090 ⇒ 00:25:07.129 Pranav: How does that sound?
324 00:25:07.850 ⇒ 00:25:15.170 Samuel Roberts: I think… I think that sounds pretty good. I mean, we’re not… I don’t remember the dates on the transcript stuff, but we’re not there yet, right? We’re still working on this part.
325 00:25:15.170 ⇒ 00:25:19.979 Pranav: Yeah, so I’m talking to Yvette literally right after this, which is why I have a stop.
326 00:25:19.980 ⇒ 00:25:20.580 Samuel Roberts: Oh, okay.
327 00:25:20.580 ⇒ 00:25:26.609 Pranav: a little bit more about transcripts, so honestly, if you guys are still on this call, like, at the end of that 20-minute meeting, I’ll…
328 00:25:26.780 ⇒ 00:25:35.339 Pranav: hop back in, and then I can give you guys some more, insight on how we should be thinking about transcripts this week, if we should be thinking about transcripts this week.
329 00:25:35.740 ⇒ 00:25:36.390 Samuel Roberts: Okay.
330 00:25:36.700 ⇒ 00:25:37.290 Pranav: Yo.
331 00:25:38.700 ⇒ 00:25:46.809 Pranav: Cool. So, one thing that I still think is unclear is how is this breakdown looking between Casey and Mustafa?
332 00:25:46.810 ⇒ 00:25:50.050 Samuel Roberts: Yeah, that’s… they’ll have to figure that out right now, so that was sort of… Okay.
333 00:25:50.650 ⇒ 00:25:51.240 Samuel Roberts: lot here.
334 00:25:51.240 ⇒ 00:25:51.880 Pranav: Okay.
335 00:25:53.130 ⇒ 00:25:55.849 Pranav: Cool. Sam, you want to figure that out, then?
336 00:25:55.850 ⇒ 00:25:58.339 Samuel Roberts: Yeah, yeah, we can, we can talk it through.
337 00:25:58.340 ⇒ 00:25:59.760 Pranav: Perfect, perfect.
338 00:26:00.040 ⇒ 00:26:00.540 Samuel Roberts: Probably.
339 00:26:00.540 ⇒ 00:26:00.870 Pranav: Very cool.
340 00:26:00.870 ⇒ 00:26:02.649 Samuel Roberts: combination of… yeah, well, look…
341 00:26:02.790 ⇒ 00:26:07.060 Samuel Roberts: Where… because Casey has most of the context, so I want to kind of make use of that.
342 00:26:07.180 ⇒ 00:26:10.950 Samuel Roberts: As best as possible, and then… .
343 00:26:10.950 ⇒ 00:26:14.490 Pranav: Yeah, so one way that I’m thinking about this, that…
344 00:26:14.740 ⇒ 00:26:18.449 Pranav: I think overall is gonna be the most optimal, is that…
345 00:26:18.720 ⇒ 00:26:32.210 Pranav: even with Eden, so I set up a call for us later today to go over Eden as well, because I want to onboard. You know, even Sam, I feel like you haven’t kind of talked to the client at all, so I’ll add you guys to the channel just to, like, do some intros, but I think
346 00:26:32.640 ⇒ 00:26:48.610 Pranav: for just, like, a longer term, like, let’s say, you know, somebody’s out, like, it works out better if everybody’s on every client. So even if it’s just, like, you know, instead of doing, you know, Casey 20 hours on ABC, must offer you 20 hours on
347 00:26:48.610 ⇒ 00:26:54.040 Pranav: Eden, I think it’d be better to do, like, 10 and 10, for example. I don’t know what the exact numbers are gonna be.
348 00:26:54.040 ⇒ 00:26:56.040 Samuel Roberts: Yeah, I see what you’re saying. Okay, yeah.
349 00:26:56.920 ⇒ 00:26:57.899 Samuel Roberts: We can look at that.
350 00:26:58.140 ⇒ 00:27:00.689 Samuel Roberts: I think that that’s not a bad idea.
351 00:27:01.230 ⇒ 00:27:01.900 Pranav: Cool.
352 00:27:02.600 ⇒ 00:27:09.609 Pranav: Okay, sounds good. Yeah, so that’ll be coming up later today. I’ll, I’ll give you guys a heads up when I… I’m gonna add you to the channel.
353 00:27:10.690 ⇒ 00:27:11.530 Mustafa Raja: Yeah.
354 00:27:11.530 ⇒ 00:27:12.240 Casie Aviles: Okay.
355 00:27:13.670 ⇒ 00:27:19.559 Pranav: Yeah, cool. Okay, guys, I’m gonna hop off and then just, like, prep for this, Yvette call, but
356 00:27:19.740 ⇒ 00:27:21.410 Pranav: Yeah, you guys can keep going.
357 00:27:22.660 ⇒ 00:27:23.240 Samuel Roberts: That’s good.
358 00:27:23.650 ⇒ 00:27:25.299 Pranav: Alright, talk to you guys later. See ya.
359 00:27:28.280 ⇒ 00:27:35.660 Samuel Roberts: Okay, so, given that, I’m trying to kind of think through where it’s best to… kind of…
360 00:27:35.960 ⇒ 00:27:37.920 Samuel Roberts: Slot you guys in.
361 00:27:38.360 ⇒ 00:27:41.510 Samuel Roberts: I’m thinking…
362 00:27:42.180 ⇒ 00:27:48.149 Mustafa Raja: So, so this automation right now, it’s only, updating the ticket, right?
363 00:27:49.750 ⇒ 00:27:50.630 Samuel Roberts: Yes, it’s not changing.
364 00:27:50.630 ⇒ 00:27:50.950 Casie Aviles: No.
365 00:27:50.950 ⇒ 00:27:51.730 Samuel Roberts: submitted all yet.
366 00:27:51.730 ⇒ 00:27:58.190 Mustafa Raja: Yeah, yeah, so, so, the goal, ultimately, is to make a suggestion in the document.
367 00:28:00.570 ⇒ 00:28:05.210 Samuel Roberts: Potentially, yeah. We’re not 100% sure the best way to do the, like,
368 00:28:06.700 ⇒ 00:28:22.240 Samuel Roberts: Yeah, is it make a suggestion, and then, like, batch all those, and have those change, or is it just identify where it needs to make the change, and then batch that, and then make a change? Like, do we want to, excuse me, make use of the suggestion feature, or just have it, like, queued up, ready to make a change?
369 00:28:23.600 ⇒ 00:28:25.510 Samuel Roberts: But either way, that’s the idea.
370 00:28:26.090 ⇒ 00:28:26.810 Mustafa Raja: Okay.
371 00:28:27.430 ⇒ 00:28:32.060 Samuel Roberts: So, there’s a few pieces here, that we just sort of talked about.
372 00:28:33.380 ⇒ 00:28:39.170 Samuel Roberts: I think… AC updating that, change type.
373 00:28:39.380 ⇒ 00:28:42.020 Samuel Roberts: To be after the identification.
374 00:28:43.660 ⇒ 00:28:46.200 Samuel Roberts: I think that probably makes sense for you to tackle that.
375 00:28:47.320 ⇒ 00:28:48.020 Casie Aviles: Yes.
376 00:28:48.250 ⇒ 00:28:49.810 Samuel Roberts: And then…
377 00:28:50.800 ⇒ 00:28:58.280 Samuel Roberts: I would potentially also say that the update to the dock might be worth it for you.
378 00:28:58.570 ⇒ 00:29:03.370 Samuel Roberts: 2, and then the, like, Conflict and change agent, maybe.
379 00:29:03.560 ⇒ 00:29:06.000 Samuel Roberts: Or the conflict and duplicate age, it might be.
380 00:29:06.900 ⇒ 00:29:07.800 Samuel Roberts: Or the, the…
381 00:29:07.800 ⇒ 00:29:08.120 Casie Aviles: Okay.
382 00:29:08.120 ⇒ 00:29:14.740 Samuel Roberts: or the parts of that workflow might be good for Mustafa, just because, like, you’re already a little plugged into the other part, and this is a little separate.
383 00:29:15.760 ⇒ 00:29:18.300 Samuel Roberts: Does that… Sound good?
384 00:29:19.260 ⇒ 00:29:22.669 Mustafa Raja: Yeah, and these are built-in, Mashra, or anything?
385 00:29:24.470 ⇒ 00:29:25.439 Samuel Roberts: This is in Moscow.
386 00:29:25.840 ⇒ 00:29:26.360 Casie Aviles: Yeah.
387 00:29:26.360 ⇒ 00:29:27.580 Mustafa Raja: Yeah, that’s good.
388 00:29:27.790 ⇒ 00:29:30.110 Samuel Roberts: So, yeah, I can… we can,
389 00:29:30.350 ⇒ 00:29:36.450 Samuel Roberts: We can actually maybe even look at that if we want, but…
390 00:29:37.310 ⇒ 00:29:41.239 Samuel Roberts: Yeah, so basically what happened last week was, Casey was working on the…
391 00:29:41.810 ⇒ 00:29:48.340 Samuel Roberts: Linear automation, reading the ticket, getting, like, parsing that and pulling it out, responding, and all that stuff.
392 00:29:48.520 ⇒ 00:29:54.489 Samuel Roberts: And then I put together a little proof of concept for using the RAG system to identify where the…
393 00:29:55.780 ⇒ 00:30:04.779 Samuel Roberts: change should be made, or surfacing what places might be the right part of the change, and then Casey kind of ran with that and refined that so that it…
394 00:30:04.960 ⇒ 00:30:06.959 Samuel Roberts: Post back to linear and everything.
395 00:30:08.530 ⇒ 00:30:16.520 Samuel Roberts: So then I think… I think… let me think, sorry, I’m just… Processing.
396 00:30:17.840 ⇒ 00:30:26.890 Samuel Roberts: the… yeah, I think… So, moving that change type to after the human comment and after identification.
397 00:30:27.470 ⇒ 00:30:32.180 Samuel Roberts: And then, actually, when identifying
398 00:30:32.550 ⇒ 00:30:38.419 Samuel Roberts: Using the RAG information that we get to identify where in the dock the change should be made.
399 00:30:38.940 ⇒ 00:30:39.900 Samuel Roberts: Right.
400 00:30:40.640 ⇒ 00:30:44.539 Samuel Roberts: Like, we’re identifying the brag element that signals…
401 00:30:45.220 ⇒ 00:30:51.270 Samuel Roberts: where it is, but we actually need to then make that change with the API or the CLI or whatever.
402 00:30:52.440 ⇒ 00:30:53.140 Samuel Roberts: Right?
403 00:30:55.420 ⇒ 00:30:57.929 Casie Aviles: You mean Google’s… Google’s API, right?
404 00:30:58.020 ⇒ 00:31:07.530 Samuel Roberts: Right, right. The RAG is just the chunks in the database, so we need to use those headings and everything to then say, okay, here’s what would actually happen.
405 00:31:07.750 ⇒ 00:31:10.820 Samuel Roberts: When we want to batch that.
406 00:31:10.960 ⇒ 00:31:16.259 Samuel Roberts: change. So, it might… might actually be the make the suggestion thing, that might be the…
407 00:31:16.390 ⇒ 00:31:19.710 Samuel Roberts: best way. But that might have to be a little bit of exploration to see.
408 00:31:20.640 ⇒ 00:31:22.439 Casie Aviles: We might need to spike if…
409 00:31:22.760 ⇒ 00:31:27.859 Casie Aviles: Like, if we can do, like, suggestions here, right? Like, through the API.
410 00:31:28.030 ⇒ 00:31:32.350 Samuel Roberts: Yeah, I think that would be good to identify. That would be helpful to know in future things, too.
411 00:31:32.800 ⇒ 00:31:44.870 Samuel Roberts: Either that, or with the CLI as well, which is… Maybe less deterministic, but… could be…
412 00:31:45.590 ⇒ 00:31:48.989 Samuel Roberts: easier to implement, and maybe more powerful? I don’t really know.
413 00:31:49.140 ⇒ 00:31:51.859 Samuel Roberts: that kind of uses the API underneath, but…
414 00:31:52.020 ⇒ 00:31:59.329 Samuel Roberts: if we’re passing it to the agent, it might be smart enough to identify that change, or identify, based on our RAG stuff.
415 00:31:59.860 ⇒ 00:32:03.880 Samuel Roberts: And then, make the suggestion in the doc.
416 00:32:04.910 ⇒ 00:32:05.660 Mustafa Raja: Yeah.
417 00:32:06.120 ⇒ 00:32:08.980 Samuel Roberts: But I think, I think that’s the idea. And then…
418 00:32:09.510 ⇒ 00:32:15.840 Samuel Roberts: The other step there is the… excuse me, sorry. What does…
419 00:32:17.140 ⇒ 00:32:22.410 Samuel Roberts: The, as part of that placement process, is it making sure that
420 00:32:22.650 ⇒ 00:32:29.809 Samuel Roberts: it’s not a duplicate anywhere, and it’s not a conflict with anything else, because they’re not going to have that information. They’re just gonna say.
421 00:32:29.920 ⇒ 00:32:32.740 Samuel Roberts: Oh, I think in the… wherever.
422 00:32:32.920 ⇒ 00:32:35.449 Samuel Roberts: We gotta make this change, right?
423 00:32:36.070 ⇒ 00:32:37.170 Samuel Roberts: And so…
424 00:32:37.480 ⇒ 00:32:43.169 Samuel Roberts: As part of that, we can identify that change and do it, but we also need to identify, is this…
425 00:32:43.700 ⇒ 00:32:48.719 Samuel Roberts: Like, conflicting with something else? Or is this duplicate already, and it’s just…
426 00:32:48.990 ⇒ 00:32:51.569 Samuel Roberts: not being pulled right from Andy or something, so…
427 00:32:54.120 ⇒ 00:33:00.439 Samuel Roberts: And then, are the tickets outlined for those other steps? Is that… I’m not sure what’s aligned with what we currently have.
428 00:33:00.810 ⇒ 00:33:05.360 Samuel Roberts: There’s duplicate detection, okay.
429 00:33:07.410 ⇒ 00:33:13.969 Samuel Roberts: So there’s duplicate and confidence. So it looks like, 2389 and 2383.
430 00:33:15.250 ⇒ 00:33:16.590 Casie Aviles: Yes, these are for…
431 00:33:16.590 ⇒ 00:33:17.590 Samuel Roberts: all of these. Oh, I see.
432 00:33:17.590 ⇒ 00:33:19.210 Casie Aviles: M4, yeah.
433 00:33:20.480 ⇒ 00:33:25.690 Samuel Roberts: So, 5… 1… So, yeah, I would say…
434 00:33:30.540 ⇒ 00:33:35.069 Samuel Roberts: I might need to look at these a little closer and make sure I remember all of them, but,
435 00:33:35.400 ⇒ 00:33:38.920 Samuel Roberts: I think 5… Has duplicate here…
436 00:33:39.380 ⇒ 00:33:41.790 Samuel Roberts: And then 6 has conflict, right?
437 00:33:42.970 ⇒ 00:33:44.550 Samuel Roberts: Contradiction. Okay.
438 00:33:45.140 ⇒ 00:33:46.149 Samuel Roberts: That seems right.
439 00:33:52.280 ⇒ 00:33:55.200 Samuel Roberts: So I would say, let’s… I would say, you guys…
440 00:33:55.200 ⇒ 00:34:01.689 Mustafa Raja: The suggestions we would be making, we want to make sure that those are no conflicts or duplicates.
441 00:34:02.100 ⇒ 00:34:09.900 Samuel Roberts: Right, so if they suggest a change that’s already there somewhere, and we… maybe Andy’s just not answering it right, we want to know… this would help us identify why.
442 00:34:10.290 ⇒ 00:34:11.230 Mustafa Raja: Right?
443 00:34:11.530 ⇒ 00:34:13.429 Samuel Roberts: And then,
444 00:34:15.040 ⇒ 00:34:20.500 Samuel Roberts: Yeah, conflict would be, like, oh, this literally is saying the, you know, something else somewhere else.
445 00:34:21.380 ⇒ 00:34:31.249 Samuel Roberts: And I think that would also be a good way to identify problems with ANDI, where it’s, maybe the RAG system is pulling duplicate, or conflicting information based on the vector lookup.
446 00:34:32.120 ⇒ 00:34:32.860 Mustafa Raja: Yeah.
447 00:34:32.989 ⇒ 00:34:36.019 Samuel Roberts: And same thing with duplicate, where if they’re gonna put something in there.
448 00:34:36.159 ⇒ 00:34:41.559 Samuel Roberts: That is already somewhere else, we should surface that, and then be able to take action.
449 00:34:41.689 ⇒ 00:34:45.149 Samuel Roberts: if it’s bad. You know, this is really just like a, kind of, a gate…
450 00:34:45.309 ⇒ 00:34:53.269 Samuel Roberts: To say, like, yes, this check is a safe… this, change is a safe thing to make, because it’s not duplicate anywhere, and it’s not conflicting anywhere.
451 00:34:54.219 ⇒ 00:34:59.469 Mustafa Raja: Okay, and this… this automation triggers when we see a comment on…
452 00:34:59.729 ⇒ 00:35:01.289 Mustafa Raja: On the central dock, right?
453 00:35:02.600 ⇒ 00:35:08.220 Samuel Roberts: Yeah, that’s a good thing we should think that through a little bit. Where do we want this to trigger from? So…
454 00:35:08.340 ⇒ 00:35:14.049 Samuel Roberts: If we have it… They make a comment.
455 00:35:14.340 ⇒ 00:35:18.349 Samuel Roberts: We take that comment, identify where in the dock, and then potentially
456 00:35:18.930 ⇒ 00:35:21.449 Samuel Roberts: I think we probably want to wait and make sure…
457 00:35:22.040 ⇒ 00:35:25.759 Samuel Roberts: We want to run this check first, before we make the suggestion.
458 00:35:26.530 ⇒ 00:35:27.640 Mustafa Raja: Yeah, of course.
459 00:35:27.640 ⇒ 00:35:32.230 Samuel Roberts: Yeah. I was thinking… I was just trying to think if it made sense to have the suggestion first, and then…
460 00:35:32.360 ⇒ 00:35:36.249 Samuel Roberts: check the dock and make sure everything works, but I think beforehand makes the most sense, so yeah.
461 00:35:36.800 ⇒ 00:35:40.000 Samuel Roberts: I think the order of the flow is…
462 00:35:40.120 ⇒ 00:35:46.430 Samuel Roberts: What we have so far, which is comment made, You know, change identified.
463 00:35:48.930 ⇒ 00:35:57.610 Samuel Roberts: And then we do a pass of the duplicate, kind of, is this anywhere else? And that might even be able to make use of the RAG lookup we already did.
464 00:35:58.220 ⇒ 00:35:59.180 Samuel Roberts: Yeah.
465 00:35:59.520 ⇒ 00:36:00.320 Samuel Roberts: So…
466 00:36:02.000 ⇒ 00:36:09.129 Samuel Roberts: Ticket created, ticket commented on, we… the location for the change, I think as part of the… what we don’t have in here is that…
467 00:36:09.370 ⇒ 00:36:14.319 Samuel Roberts: We will do a check for the stuff, duplicate or conflict in here.
468 00:36:15.110 ⇒ 00:36:16.110 Mustafa Raja: Hmm.
469 00:36:16.120 ⇒ 00:36:19.670 Samuel Roberts: So that’s not part of this over… higher level overview, so…
470 00:36:21.340 ⇒ 00:36:26.559 Mustafa Raja: Yeah. And then the client approval would be on a suggestion, right?
471 00:36:26.790 ⇒ 00:36:32.589 Samuel Roberts: I think so, yes, I think that’s the best way to do it. And then those suggestions will get batched and updated once a day.
472 00:36:32.890 ⇒ 00:36:35.990 Samuel Roberts: And if it gets rejected, maybe we reprocess it.
473 00:36:35.990 ⇒ 00:36:45.960 Mustafa Raja: By batched and updated, we want to… we’re trying to say that we will add the… add all of the suggestions once a day.
474 00:36:46.200 ⇒ 00:36:50.039 Samuel Roberts: That was the plan, yeah, rather than just do them, one at a time.
475 00:36:50.180 ⇒ 00:36:51.069 Samuel Roberts: The thought was.
476 00:36:51.070 ⇒ 00:36:52.150 Mustafa Raja: Yeah, medications.
477 00:36:52.150 ⇒ 00:36:52.470 Samuel Roberts: Yeah.
478 00:36:52.470 ⇒ 00:36:52.990 Mustafa Raja: Yeah.
479 00:36:53.000 ⇒ 00:36:58.890 Samuel Roberts: I mean, that could change depending on how many of these changes are coming through, I suppose, but…
480 00:36:59.700 ⇒ 00:37:00.960 Samuel Roberts: I think that works.
481 00:37:01.800 ⇒ 00:37:02.630 Samuel Roberts: For now.
482 00:37:06.330 ⇒ 00:37:10.859 Casie Aviles: I… I guess I have a question. So, we’re saying that…
483 00:37:11.490 ⇒ 00:37:15.030 Casie Aviles: The approval will happen here in the document.
484 00:37:15.030 ⇒ 00:37:20.329 Samuel Roberts: No, no, no, I think that the… that’s where we will identify the change. The approval will still happen in linear.
485 00:37:20.820 ⇒ 00:37:21.960 Casie Aviles: Oh, okay, okay.
486 00:37:21.960 ⇒ 00:37:23.280 Samuel Roberts: Yeah, yeah, yeah.
487 00:37:23.280 ⇒ 00:37:24.190 Casie Aviles: misunderstood.
488 00:37:24.190 ⇒ 00:37:26.649 Samuel Roberts: No, no, no, that’s a… I’m just trying to think of the best way to, like…
489 00:37:26.980 ⇒ 00:37:33.819 Samuel Roberts: set up the change that we’re gonna make in the document. And if it’s a suggestion, or if it’s just a, like, identification of
490 00:37:34.090 ⇒ 00:37:41.230 Samuel Roberts: where we change it, I’m not sure which one’s better, if we want to make use of the Google Doc features, or just have it ready to go.
491 00:37:41.500 ⇒ 00:37:42.640 Samuel Roberts: Some other way?
492 00:37:43.480 ⇒ 00:37:44.160 Casie Aviles: Okay.
493 00:37:46.170 ⇒ 00:37:49.010 Samuel Roberts: That’s a little bit of an unanswered question at this point.
494 00:37:49.140 ⇒ 00:37:53.079 Samuel Roberts: So, I think that’s part of the, like, let’s see what works best.
495 00:37:55.250 ⇒ 00:38:03.679 Samuel Roberts: But either way, we’ll have those changes identified. Like, we already kind of have, like, the location identified, we then need to…
496 00:38:04.240 ⇒ 00:38:09.060 Samuel Roberts: Do the actual… Text substitution, or addition, or deletion, or whatever.
497 00:38:09.710 ⇒ 00:38:15.049 Samuel Roberts: So, yeah, I think, Casey, if you can…
498 00:38:15.230 ⇒ 00:38:17.980 Samuel Roberts: I don’t know which tickets these are specifically now, but…
499 00:38:21.060 ⇒ 00:38:26.329 Samuel Roberts: the placement automation, is there other more tickets still unassigned, or, uncompleted in here? What’s the…
500 00:38:27.530 ⇒ 00:38:29.060 Samuel Roberts: Yeah, so these are all…
501 00:38:29.060 ⇒ 00:38:30.440 Casie Aviles: everything and testing.
502 00:38:30.440 ⇒ 00:38:33.569 Samuel Roberts: Okay, okay, so we’re pretty good on all that, except for…
503 00:38:34.200 ⇒ 00:38:36.650 Samuel Roberts: Do we have the preview generation? What is this?
504 00:38:37.370 ⇒ 00:38:38.820 Casie Aviles: Yeah, this one, no.
505 00:38:39.550 ⇒ 00:38:42.710 Casie Aviles: So maybe this… this should be a different…
506 00:38:42.960 ⇒ 00:38:43.610 Samuel Roberts: Okay.
507 00:38:45.160 ⇒ 00:38:49.150 Samuel Roberts: Yeah, okay. I just want to make sure we’re… we’re aligned on what’s… what we’re…
508 00:38:49.930 ⇒ 00:38:54.130 Casie Aviles: Yeah, let me see if I actually touched on anything here.
509 00:38:56.300 ⇒ 00:39:01.930 Casie Aviles: Oh, okay, yeah, because I was testing something about… Something like this.
510 00:39:01.930 ⇒ 00:39:03.260 Samuel Roberts: Yes, okay, cool.
511 00:39:03.600 ⇒ 00:39:07.750 Casie Aviles: But I did… it didn’t make it to… Friday, so…
512 00:39:07.750 ⇒ 00:39:10.680 Samuel Roberts: Okay, alright, so we have some progress there, either way.
513 00:39:11.430 ⇒ 00:39:14.470 Casie Aviles: Yeah, that’s kind of why I put it there, but…
514 00:39:14.470 ⇒ 00:39:15.299 Samuel Roberts: Okay, that makes sense.
515 00:39:15.300 ⇒ 00:39:19.260 Casie Aviles: It’s not the best, definitely can be improved still.
516 00:39:20.210 ⇒ 00:39:27.899 Samuel Roberts: then I would say work on that, and work on updating the change type to not be required here. Oops, sorry, but yeah.
517 00:39:28.800 ⇒ 00:39:31.830 Samuel Roberts: So that will identify the change type as part of this process.
518 00:39:33.450 ⇒ 00:39:46.649 Samuel Roberts: And then, I mean, this might be a little bit of you guys working together, because it’s kind of linked, but the… the duplicate and the conflict, I guess, agents, as part of the step.
519 00:39:48.450 ⇒ 00:39:54.620 Samuel Roberts: Maybe even after the proposed edit, I don’t know where we… where exactly it should go best, but…
520 00:39:55.130 ⇒ 00:40:01.470 Samuel Roberts: I would say, Mustafa, if you can grab that, which I think is a few other tickets in…
521 00:40:02.710 ⇒ 00:40:04.729 Samuel Roberts: the next milestone.
522 00:40:05.690 ⇒ 00:40:06.300 Mustafa Raja: Yep.
523 00:40:06.530 ⇒ 00:40:08.250 Samuel Roberts: Would be good, and then…
524 00:40:08.250 ⇒ 00:40:13.360 Mustafa Raja: So, I need to take care of the agents, for conflicts and, duplicates, right?
525 00:40:13.940 ⇒ 00:40:16.049 Samuel Roberts: Yeah, and I think those can kind of be…
526 00:40:16.620 ⇒ 00:40:20.529 Samuel Roberts: very similar, or kind of tied together, you know, like, we’re gonna be looking for…
527 00:40:20.810 ⇒ 00:40:26.730 Samuel Roberts: The content anyway, so whether or not we can identify if it’s a duplicate or conflict might be a similar kind of approach.
528 00:40:27.050 ⇒ 00:40:28.689 Samuel Roberts: Like, everything’s broken.
529 00:40:28.690 ⇒ 00:40:40.730 Mustafa Raja: Yeah, so the duplicate or contradiction that we are looking… we want to look for is the suggestion that we would be suggesting based on the comment?
530 00:40:41.520 ⇒ 00:40:47.999 Samuel Roberts: Yeah, so, like, if, for example, one of the, the tickets we were playing with was, like, add something about carpenter bees, right?
531 00:40:48.690 ⇒ 00:41:00.470 Samuel Roberts: And so if somewhere else it explicitly says, like, we don’t do carpenter bees, but we’re adding… yeah, so this was, like, remove carpenter bees. And so I think we want to identify, like, if we remove it in this place.
532 00:41:00.580 ⇒ 00:41:01.479 Samuel Roberts: Is that…
533 00:41:01.650 ⇒ 00:41:09.470 Samuel Roberts: Conflicting with anything else that specifically says carpenter bees are still okay, in which case that will identify another change that has to be made.
534 00:41:09.720 ⇒ 00:41:10.400 Samuel Roberts: Potentially.
535 00:41:11.040 ⇒ 00:41:13.889 Samuel Roberts: Or is it… they try to suggest something that is…
536 00:41:14.400 ⇒ 00:41:17.249 Samuel Roberts: Also, already in the dock somewhere.
537 00:41:18.020 ⇒ 00:41:22.919 Samuel Roberts: Which might mean that that change is not necessary, or we need to get rid of it somewhere else, you know?
538 00:41:23.310 ⇒ 00:41:26.080 Samuel Roberts: Like, the outcome of this could be a few different things.
539 00:41:26.290 ⇒ 00:41:28.800 Samuel Roberts: But the identification is what’s important right now.
540 00:41:29.820 ⇒ 00:41:31.390 Samuel Roberts: And surfacing that back.
541 00:41:33.560 ⇒ 00:41:34.190 Mustafa Raja: Yeah.
542 00:41:36.620 ⇒ 00:41:45.789 Samuel Roberts: And I think, much like the placement stuff, you might be able to make use of that RAG lookup, even… even as part of that, potentially, maybe it branches off from there using that.
543 00:41:47.770 ⇒ 00:41:52.970 Samuel Roberts: You know, using that initial lookup, the semantic vector search,
544 00:41:53.800 ⇒ 00:41:55.870 Samuel Roberts: We’re already pulling a bunch of stuff.
545 00:41:56.700 ⇒ 00:42:00.940 Samuel Roberts: So… You might be able to just branch off that.
546 00:42:02.960 ⇒ 00:42:07.060 Casie Aviles: We have it under here, yeah, this one, central dog placement workflow.
547 00:42:07.820 ⇒ 00:42:13.790 Samuel Roberts: So this is… it’s a fairly simple one where it just kind of, like, identifies… Roughly where.
548 00:42:15.640 ⇒ 00:42:18.410 Samuel Roberts: Sorry, I’m not feeling great right now, guys.
549 00:42:20.090 ⇒ 00:42:23.369 Casie Aviles: Oh, it… Yeah, feel free if you need to.
550 00:42:23.370 ⇒ 00:42:23.890 Samuel Roberts: Yeah, no.
551 00:42:23.890 ⇒ 00:42:24.330 Mustafa Raja: Yes.
552 00:42:24.330 ⇒ 00:42:28.979 Samuel Roberts: I just… I don’t have any water up here, since everything was moved before I went away.
553 00:42:29.090 ⇒ 00:42:32.400 Samuel Roberts: No, let’s finish this up, and then I’ll go take care of myself.
554 00:42:33.720 ⇒ 00:42:36.570 Samuel Roberts: I just wanna make sure you guys have enough to run with.
555 00:42:38.570 ⇒ 00:42:41.620 Casie Aviles: Yeah, all good. I can sync with Mustafa later as well.
556 00:42:41.620 ⇒ 00:42:42.880 Mustafa Raja: Yeah, I think, yeah.
557 00:42:42.880 ⇒ 00:42:45.830 Samuel Roberts: I imagine it’ll be a bunch of, like, working together today to…
558 00:42:46.160 ⇒ 00:42:49.450 Mustafa Raja: Yeah, yeah, I think Casey and I can take it from here, then.
559 00:42:50.150 ⇒ 00:42:50.800 Samuel Roberts: Okay.
560 00:42:52.000 ⇒ 00:42:57.840 Samuel Roberts: Yeah, I would say, let me know if there’s any questions, but I think… I think that’s the idea. I would say, yeah, sign…
561 00:42:58.250 ⇒ 00:43:05.560 Samuel Roberts: the, like, duplicate conflict stuff, Mustafa, and then Casey, you finish up the other ones, but there’s gonna be a lot of, like, working together for that, so…
562 00:43:05.710 ⇒ 00:43:06.550 Mustafa Raja: No.
563 00:43:07.030 ⇒ 00:43:15.210 Samuel Roberts: they’re not… they’re not as, and the other thing is, like, I tried to make sure that anything that needed human decisions and stuff were… was tagged with human required.
564 00:43:16.590 ⇒ 00:43:22.080 Samuel Roberts: So… like, here. Like, this one…
565 00:43:22.280 ⇒ 00:43:26.570 Samuel Roberts: there might have to be some work, but the other ones, I was hoping to make sure that they were…
566 00:43:27.350 ⇒ 00:43:31.879 Samuel Roberts: scoped out enough that Cursor would have more context, and could run with it a little bit.
567 00:43:32.350 ⇒ 00:43:36.829 Samuel Roberts: If that makes sense. So, definitely try to make use of cursor as much as possible for the ones that are…
568 00:43:37.340 ⇒ 00:43:39.740 Samuel Roberts: Not tag human required, but.
569 00:43:41.940 ⇒ 00:43:42.860 Samuel Roberts: Alright.
570 00:43:43.170 ⇒ 00:43:45.329 Casie Aviles: Okay, yeah, rest, rest stop.
571 00:43:45.330 ⇒ 00:43:54.049 Samuel Roberts: Okay, thanks, guys. Yeah, I’ll see… I have some other stuff to do today, too, but I’m gonna see what I can make work, but thank you both. Let me know if I can help with anything.
572 00:43:54.450 ⇒ 00:43:55.010 Samuel Roberts: Yep.
573 00:43:55.010 ⇒ 00:43:55.700 Casie Aviles: Sure.
574 00:43:55.700 ⇒ 00:43:56.410 Samuel Roberts: Alright.
575 00:43:56.850 ⇒ 00:43:57.990 Samuel Roberts: Appreciate you both.
576 00:43:58.830 ⇒ 00:43:59.870 Casie Aviles: Thanks, everyone.
577 00:44:00.380 ⇒ 00:44:01.189 Mustafa Raja: Thank you.
578 00:44:01.460 ⇒ 00:44:02.410 Samuel Roberts: Cocktail later.
579 00:44:02.910 ⇒ 00:44:03.760 Mustafa Raja: Yeah, bye.