Meeting Title: Daily AI Team Sync Date: 2025-02-13 Meeting participants: Janna Wong, Uttam Kumaran, Miguel De Veyra, Casie Aviles
WEBVTT
1 00:03:49.180 ⇒ 00:03:50.329 Miguel de Veyra: Hey, Jenna.
2 00:03:54.010 ⇒ 00:03:54.810 Janna Wong: There we go!
3 00:03:55.840 ⇒ 00:03:56.660 Miguel de Veyra: I don’t know.
4 00:03:57.270 ⇒ 00:03:59.349 Miguel de Veyra: I’ll just ping Daisy.
5 00:03:59.870 ⇒ 00:04:02.040 Miguel de Veyra: Okay, there you go.
6 00:04:02.640 ⇒ 00:04:08.420 Miguel de Veyra: Let’s wait for like 2 or 3 min, and then, if autumn doesn’t join. Let’s just start.
7 00:04:20.250 ⇒ 00:04:21.880 Miguel de Veyra: And then
8 00:04:30.490 ⇒ 00:04:31.570 Miguel de Veyra: or 2 1.
9 00:05:03.540 ⇒ 00:05:08.038 Miguel de Veyra: I don’t think Utam’s gonna join guys. He’s offline. So I guess let’s just
10 00:05:16.450 ⇒ 00:05:17.160 Miguel de Veyra: ugh!
11 00:05:18.170 ⇒ 00:05:21.390 Casie Aviles: But yeah, that I don’t discuss right now. Hold on a minute.
12 00:05:23.810 ⇒ 00:05:28.670 Miguel de Veyra: I guess we just loop Jana in on what happened what we had earlier. And then I think that’s pretty much it. Now.
13 00:05:29.660 ⇒ 00:05:30.290 Casie Aviles: Okay.
14 00:05:31.200 ⇒ 00:05:34.810 Miguel de Veyra: I’ll just share my have a good trip.
15 00:05:35.430 ⇒ 00:05:36.300 Casie Aviles: Yeah.
16 00:05:36.550 ⇒ 00:05:38.190 Casie Aviles: 5 min. Okay.
17 00:05:38.190 ⇒ 00:05:39.440 Miguel de Veyra: Then we just do it.
18 00:05:47.410 ⇒ 00:05:50.956 Miguel de Veyra: That usually turns out to be 20 min.
19 00:05:53.060 ⇒ 00:05:56.290 Miguel de Veyra: But yeah, I guess I’ll just show it to Jana anyways.
20 00:06:02.190 ⇒ 00:06:02.910 Miguel de Veyra: Oh, no.
21 00:06:05.070 ⇒ 00:06:07.220 Miguel de Veyra: So what we did, Janna, cause I
22 00:06:07.430 ⇒ 00:06:14.470 Miguel de Veyra: so far, we’re gonna stick with. And first, st right.
23 00:06:15.790 ⇒ 00:06:16.480 Janna Wong: Yep.
24 00:06:16.690 ⇒ 00:06:22.739 Miguel de Veyra: So in business, Google sheets, which we did before
25 00:06:23.500 ⇒ 00:06:27.100 Miguel de Veyra: right now connect the dimension to Snowflake.
26 00:06:27.890 ⇒ 00:06:30.819 Miguel de Veyra: So, for example, I don’t know about it.
27 00:06:32.020 ⇒ 00:06:37.279 Miguel de Veyra: Oh, or is it on me, and download all the minutes?
28 00:06:37.930 ⇒ 00:06:44.230 Miguel de Veyra: It shouldn’t know why arrow will see it here. It should appear here
29 00:06:53.630 ⇒ 00:06:58.419 Miguel de Veyra: so 23, or 3 or 2 11. Yeah, it should be this one.
30 00:06:59.640 ⇒ 00:07:04.030 Miguel de Veyra: So there you go, and then I guess we have to reload. No PC.
31 00:07:04.990 ⇒ 00:07:05.740 Casie Aviles: Have it!
32 00:07:06.280 ⇒ 00:07:07.229 Miguel de Veyra: Oh, yeah, yeah.
33 00:07:07.230 ⇒ 00:07:09.229 Miguel de Veyra: I refreshed. Oh, there you go!
34 00:07:12.690 ⇒ 00:07:14.680 Janna Wong: Damanga.
35 00:07:17.620 ⇒ 00:07:18.470 Janna Wong: Okay, okay.
36 00:07:18.470 ⇒ 00:07:19.890 Miguel de Veyra: So then these are sharing right
37 00:07:24.410 ⇒ 00:07:26.380 Miguel de Veyra: what Alano Border is right
38 00:07:41.830 ⇒ 00:07:43.310 Miguel de Veyra: then, I guess.
39 00:07:43.970 ⇒ 00:07:46.839 Miguel de Veyra: There you go. So what happens? There is
40 00:07:52.510 ⇒ 00:07:57.810 Miguel de Veyra: we’ll see right now you’ll see it. Here comes in, and then I
41 00:07:57.930 ⇒ 00:08:01.189 Miguel de Veyra: because it’s a snowflake. There’s like, No, you know.
42 00:08:01.680 ⇒ 00:08:04.629 Miguel de Veyra: like it’s so weird. Why, it has to be like this.
43 00:08:05.500 ⇒ 00:08:07.560 Miguel de Veyra: What the way it works is.
44 00:08:07.960 ⇒ 00:08:13.180 Miguel de Veyra: So edit Field, you have to name it after the fields in the actual SQL.
45 00:08:13.600 ⇒ 00:08:19.049 Miguel de Veyra: And then just map it there. I that’s why I named it Snowflake cleaner. And then
46 00:08:19.170 ⇒ 00:08:25.340 Miguel de Veyra: basically insert it like this, the columns. And then it just maps out automatically.
47 00:08:26.590 ⇒ 00:08:35.660 Miguel de Veyra: so yeah, and then, yeah, so I guess at the at the moment, in conversations, you think that we need to do with them.
48 00:08:36.510 ⇒ 00:08:37.470 Miguel de Veyra: No problem.
49 00:08:38.240 ⇒ 00:08:38.830 Casie Aviles: Nope.
50 00:08:39.770 ⇒ 00:08:42.990 Miguel de Veyra: Okay, use a taxonomy. Docs.
51 00:08:44.370 ⇒ 00:08:50.790 Miguel de Veyra: Fuck, am I doing looks? No Google books.
52 00:08:54.530 ⇒ 00:08:56.900 Miguel de Veyra: you know. So taxonomy, taxonomy?
53 00:09:06.360 ⇒ 00:09:08.326 Miguel de Veyra: Yeah. So it’s this one.
54 00:09:18.920 ⇒ 00:09:21.569 Miguel de Veyra: no, we’re destroying insect reports.
55 00:09:25.090 ⇒ 00:09:28.920 Miguel de Veyra: I know about this simarian or not yet
56 00:09:38.030 ⇒ 00:09:38.740 Miguel de Veyra: in the guitar.
57 00:09:38.740 ⇒ 00:09:40.270 Janna Wong: And zoom, coulomb.
58 00:09:42.540 ⇒ 00:09:49.110 Miguel de Veyra: Yeah, is Bubble again, like when we send it, like, basically.
59 00:09:53.730 ⇒ 00:09:58.260 Miguel de Veyra: So I’m not sure if necessary, I’m going. Please go.
60 00:10:02.810 ⇒ 00:10:03.869 Miguel de Veyra: I’m sorry.
61 00:10:07.210 ⇒ 00:10:09.330 Miguel de Veyra: I think I did this one.
62 00:10:09.870 ⇒ 00:10:11.219 Miguel de Veyra: Oh, not that one
63 00:10:18.580 ⇒ 00:10:25.130 Miguel de Veyra: you know about. I don’t know 6 days ago.
64 00:10:25.750 ⇒ 00:10:29.319 Miguel de Veyra: but the main message should I? I think I give it right now.
65 00:10:30.800 ⇒ 00:10:37.280 Miguel de Veyra: and I was ready when you went 2 days ago.
66 00:10:38.790 ⇒ 00:10:48.360 Miguel de Veyra: IMPR. D. Rich and and annual reviews.
67 00:10:48.730 ⇒ 00:10:57.330 Miguel de Veyra: So far we’ll happen among comments on both need or opposite place. So
68 00:10:57.810 ⇒ 00:10:59.779 Miguel de Veyra: where it’s like up in the air right now.
69 00:11:07.310 ⇒ 00:11:09.230 Miguel de Veyra: No, she’s good tomorrow.
70 00:11:17.790 ⇒ 00:11:19.900 Miguel de Veyra: Sorry, John Nigerian.
71 00:11:20.400 ⇒ 00:11:21.370 Janna Wong: Hello! Hello!
72 00:11:21.520 ⇒ 00:11:23.119 Miguel de Veyra: Yeah, there you go, much greater.
73 00:11:23.640 ⇒ 00:11:24.799 Janna Wong: No, you you’ll be.
74 00:11:26.110 ⇒ 00:11:32.990 Miguel de Veyra: Yeah, I guess. Yeah, like, leave it as is this one
75 00:11:33.090 ⇒ 00:11:36.880 Miguel de Veyra: project pages such as basically the tasks.
76 00:11:38.200 ⇒ 00:11:40.980 Miguel de Veyra: So yeah, I guess right now.
77 00:11:42.170 ⇒ 00:11:46.810 Miguel de Veyra: no need going detail. Shit new suggestions, no.
78 00:11:48.450 ⇒ 00:11:55.630 Janna Wong: I think, reply.
79 00:11:56.166 ⇒ 00:12:00.020 Miguel de Veyra: Yeah, it’s fine. It’s fine. We don’t. We don’t really care about Scott.
80 00:12:00.240 ⇒ 00:12:00.580 Janna Wong: Yeah.
81 00:12:05.570 ⇒ 00:12:23.130 Miguel de Veyra: Advice may go signal from me, or we don’t do it.
82 00:12:23.530 ⇒ 00:12:26.650 Miguel de Veyra: Oh, not really, you know, it’s not really a.
83 00:12:27.690 ⇒ 00:12:28.560 Miguel de Veyra: Hey? What’s up?
84 00:12:29.350 ⇒ 00:12:30.933 Miguel de Veyra: It’s not really like.
85 00:12:31.690 ⇒ 00:12:32.579 Uttam Kumaran: Hey, guys.
86 00:12:32.870 ⇒ 00:12:33.750 Miguel de Veyra: Hey? Autumn?
87 00:12:34.340 ⇒ 00:12:36.270 Miguel de Veyra: It’s more of like a need to know.
88 00:12:38.440 ⇒ 00:12:39.150 Janna Wong: Yeah.
89 00:12:40.050 ⇒ 00:12:41.490 Miguel de Veyra: I may need to know, like good.
90 00:12:41.490 ⇒ 00:12:42.520 Janna Wong: And nice to have. Yeah.
91 00:12:42.520 ⇒ 00:12:45.310 Miguel de Veyra: To have. Yeah, but not not necessary.
92 00:12:47.200 ⇒ 00:12:48.010 Janna Wong: Okay.
93 00:12:48.640 ⇒ 00:12:52.860 Miguel de Veyra: So hey with them? Yeah. So we were just basically discussing how
94 00:12:53.040 ⇒ 00:12:57.640 Miguel de Veyra: you know instead of sending it to Google sheets. Now, we’re sending it to Snowflake, and
95 00:12:58.030 ⇒ 00:12:59.519 Miguel de Veyra: is like the results.
96 00:13:05.610 ⇒ 00:13:10.270 Miguel de Veyra: That one. And then, yeah, I’m just monitoring my emails, basically in this one.
97 00:13:10.380 ⇒ 00:13:15.070 Miguel de Veyra: And then so far, every document we have. Jana put them into something like this.
98 00:13:15.620 ⇒ 00:13:23.759 Miguel de Veyra: But before we do, you know more stuff on this, we’re just waiting, basically on what they their feedback.
99 00:13:30.900 ⇒ 00:13:33.580 Miguel de Veyra: Yeah, I think that’s pretty much it on my end.
100 00:13:34.000 ⇒ 00:13:38.849 Miguel de Veyra: As for for ABC, I mean 14 seats.
101 00:13:38.850 ⇒ 00:13:43.900 Uttam Kumaran: On on ABC, so couple of things one, we’re just waiting for feedback from them.
102 00:13:43.900 ⇒ 00:13:44.480 Miguel de Veyra: Yeah, yeah.
103 00:13:45.012 ⇒ 00:13:51.719 Uttam Kumaran: In the meantime, like, can we just keep moving forward with, like our assumptions? Maybe on
104 00:13:51.910 ⇒ 00:13:57.729 Uttam Kumaran: building up the Bible documents and sort of like starting to
105 00:13:57.920 ⇒ 00:14:01.610 Uttam Kumaran: starting to almost like, deprecate those other documents.
106 00:14:02.470 ⇒ 00:14:04.080 Miguel de Veyra: Oh, okay. Yeah. Sure.
107 00:14:04.080 ⇒ 00:14:06.190 Uttam Kumaran: I guess, like we don’t have to delete them. But like
108 00:14:06.400 ⇒ 00:14:09.819 Uttam Kumaran: we can basically, I hope to just remove them from rag right?
109 00:14:09.930 ⇒ 00:14:11.400 Uttam Kumaran: Ultimately, like.
110 00:14:11.770 ⇒ 00:14:17.420 Uttam Kumaran: That’s that way. We know that the only source of truth is coming from that Bible.
111 00:14:17.650 ⇒ 00:14:24.789 Miguel de Veyra: Yeah, yeah, yeah, okay, yeah, basically, just put everything into, you know something, SIM, something like this
112 00:14:26.760 ⇒ 00:14:29.750 Miguel de Veyra: to a Bible and then base the rug after. Yeah, that makes sense.
113 00:14:29.750 ⇒ 00:14:31.440 Uttam Kumaran: Also the also. The other thing is.
114 00:14:31.800 ⇒ 00:14:37.050 Uttam Kumaran: I don’t know how we’re gonna handle the spreadsheets like
115 00:14:37.370 ⇒ 00:14:46.909 Uttam Kumaran: my question about the spreadsheet is like, how do you? How do you maintain the relationships? Are they are, is it able like, are you gonna convert it to Csv and bring it in
116 00:14:47.100 ⇒ 00:14:52.839 Uttam Kumaran: like you could just bring it in, not to the Bible, but like to the.
117 00:14:53.302 ⇒ 00:14:54.689 Miguel de Veyra: Yeah, we’ll probably.
118 00:14:54.690 ⇒ 00:14:56.319 Uttam Kumaran: Yeah, a. Csv.
119 00:14:56.320 ⇒ 00:15:12.849 Miguel de Veyra: The best format. There would be Json to be honest, not Csv, and then putting it to rag, we could do that. But yeah, definitely. But we we want to know first, st basically on their feedback on, you know, which ones to keep.
120 00:15:15.110 ⇒ 00:15:15.790 Uttam Kumaran: Okay.
121 00:15:16.170 ⇒ 00:15:16.700 Miguel de Veyra: And we try.
122 00:15:16.700 ⇒ 00:15:17.250 Uttam Kumaran: Thank God!
123 00:15:17.250 ⇒ 00:15:17.690 Miguel de Veyra: Remove.
124 00:15:18.500 ⇒ 00:15:26.790 Uttam Kumaran: I think, in terms of rag as well. I guess I want to understand as we’re built as we’re testing like, what decisions we’re making on chunking
125 00:15:27.287 ⇒ 00:15:29.490 Uttam Kumaran: how we’re doing the actual retrieval.
126 00:15:31.590 ⇒ 00:15:36.640 Uttam Kumaran: you know, because I guess this this document is gonna be pretty big. But of course we’re gonna be.
127 00:15:36.740 ⇒ 00:15:41.270 Uttam Kumaran: we’re gonna be looking at, most likely several other other
128 00:15:41.520 ⇒ 00:15:48.169 Uttam Kumaran: sectors as well. So I just wanna make sure that, like whatever rag we choose.
129 00:15:48.720 ⇒ 00:15:52.410 Uttam Kumaran: we sort of have a good understanding on why we’re making decisions on chunking and things like that.
130 00:15:52.810 ⇒ 00:15:53.150 Miguel de Veyra: Yep.
131 00:15:55.410 ⇒ 00:15:59.709 Uttam Kumaran: You know, and and also like again, we should. Also, I mean, that document isn’t so big
132 00:15:59.860 ⇒ 00:16:03.960 Uttam Kumaran: like you could just bring a lot of it into context, right? Like
133 00:16:04.200 ⇒ 00:16:11.369 Uttam Kumaran: one thing that I was reading a lot about. It was like people using gemini flash because the context window is like a million tokens.
134 00:16:11.610 ⇒ 00:16:17.860 Uttam Kumaran: So if that document isn’t that big, you can probably just bring the entire thing into context right? Like.
135 00:16:18.470 ⇒ 00:16:22.130 Uttam Kumaran: I guess that that would be my question as well as like, why not just do that.
136 00:16:23.885 ⇒ 00:16:28.000 Miguel de Veyra: Yeah, I mean, we could. We could also try that out.
137 00:16:29.180 ⇒ 00:16:30.059 Miguel de Veyra: Why not? Right?
138 00:16:30.060 ⇒ 00:16:36.349 Uttam Kumaran: Okay, maybe we just have like different. Pro, I mean, probably once we get the email set up, we can just create different ports and try.
139 00:16:36.350 ⇒ 00:16:37.810 Miguel de Veyra: Yeah, yeah, see which one.
140 00:16:37.810 ⇒ 00:16:39.850 Uttam Kumaran: Yeah, try, different. Things. Yeah.
141 00:16:39.850 ⇒ 00:16:51.629 Miguel de Veyra: Yeah, yeah, that’s true. But yeah, I mean, so far, this this one is doing pretty well what we what we decided not really decided, but sort of like
142 00:16:52.897 ⇒ 00:16:56.959 Miguel de Veyra: sort of like. How do you say this
143 00:16:58.020 ⇒ 00:17:00.790 Miguel de Veyra: thought of to like, you know?
144 00:17:01.250 ⇒ 00:17:07.190 Miguel de Veyra: Wait, let me just show you super base to improve the rag is basically
145 00:17:07.980 ⇒ 00:17:11.490 Miguel de Veyra: to not search everything. But you know, it’s pretty small
146 00:17:12.210 ⇒ 00:17:18.719 Miguel de Veyra: right now, so it won’t really matter, because so, for example, we go into something like this. Right? Our notion. Sync.
147 00:17:19.490 ⇒ 00:17:23.200 Miguel de Veyra: this is our, oh, wait. Yeah, that’s loading.
148 00:17:23.430 ⇒ 00:17:36.100 Miguel de Veyra: It’s like, you know, it has a type, you know. So if you ask for Demos, or if it’s about leads, it’s not gonna ideal. The idea is that it’s not gonna look anymore. For
149 00:17:36.860 ⇒ 00:17:42.080 Miguel de Veyra: you know, records that has this type. It’s only gonna look for basically some sort of tagging. Right? You you know.
150 00:17:42.080 ⇒ 00:17:42.870 Uttam Kumaran: Need metadata.
151 00:17:42.870 ⇒ 00:17:45.030 Miguel de Veyra: Everywhere, yeah, yeah, something like that.
152 00:17:45.200 ⇒ 00:17:46.969 Uttam Kumaran: Okay, okay, that’s 1 of the things.
153 00:17:46.970 ⇒ 00:17:48.170 Miguel de Veyra: We’re looking into.
154 00:17:48.710 ⇒ 00:17:53.979 Uttam Kumaran: Totally. I think there’s 3 things. One, it’s like the chunking second is what metadata you need on.
155 00:17:54.200 ⇒ 00:17:56.080 Uttam Kumaran: like basically each chunk.
156 00:17:56.260 ⇒ 00:17:56.580 Miguel de Veyra: Yeah.
157 00:17:57.910 ⇒ 00:18:06.120 Uttam Kumaran: Because also, again, we, we may potentially want to bring in images. We may potentially want to bring in other things into.
158 00:18:06.670 ⇒ 00:18:10.929 Uttam Kumaran: you know the process. So if we need to bring images videos, we can bring that.
159 00:18:12.150 ⇒ 00:18:12.770 Miguel de Veyra: Yeah.
160 00:18:13.070 ⇒ 00:18:18.470 Miguel de Veyra: Embedding, embedding. What was the thing with the 1 million contacts? Was it.
161 00:18:19.000 ⇒ 00:18:20.690 Uttam Kumaran: Yeah, so Gemini class.
162 00:18:21.320 ⇒ 00:18:21.950 Miguel de Veyra: Sorry.
163 00:18:22.490 ⇒ 00:18:23.180 Miguel de Veyra: Yeah. Jim.
164 00:18:23.774 ⇒ 00:18:26.500 Uttam Kumaran: The context window is very large.
165 00:18:26.960 ⇒ 00:18:32.540 Uttam Kumaran: so you don’t really even need to chunk for some stuff.
166 00:18:32.540 ⇒ 00:18:33.160 Miguel de Veyra: Yeah.
167 00:18:33.160 ⇒ 00:18:38.100 Uttam Kumaran: Like, you can just load basically everything into contacts every time.
168 00:18:39.810 ⇒ 00:18:42.359 Uttam Kumaran: I don’t know how that affects like performance, but.
169 00:18:42.600 ⇒ 00:18:44.399 Miguel de Veyra: It’s gonna be a lot slower.
170 00:18:45.650 ⇒ 00:18:54.590 Uttam Kumaran: Yeah. But but also it’s like we we don’t. We can avoid retrieval right in some ways, or like the retrieval process can be broader. I don’t know, but we can consider.
171 00:18:55.080 ⇒ 00:18:56.819 Uttam Kumaran: We can consider that as well.
172 00:18:57.060 ⇒ 00:19:05.199 Miguel de Veyra: I mean for reference with them. This entire Doc. I think the only thing missing here is the sops. It’s only like 2,000 tokens.
173 00:19:06.360 ⇒ 00:19:08.729 Uttam Kumaran: Wait for for our notion, or for.
174 00:19:08.950 ⇒ 00:19:10.269 Miguel de Veyra: For ABC for ABC.
175 00:19:10.510 ⇒ 00:19:11.550 Uttam Kumaran: Okay. Okay. Okay. Okay.
176 00:19:11.550 ⇒ 00:19:12.900 Miguel de Veyra: So we can definitely.
177 00:19:12.900 ⇒ 00:19:15.220 Uttam Kumaran: Okay, okay, so it’s so small. Yeah, yeah, yeah.
178 00:19:15.220 ⇒ 00:19:17.720 Miguel de Veyra: Yeah, it’s so small. So yeah, it should be, you know.
179 00:19:18.110 ⇒ 00:19:24.569 Miguel de Veyra: But the thing is with this. I remember we did this before Casey, right. It consumes like a shit ton of tokens.
180 00:19:26.102 ⇒ 00:19:27.129 Casie Aviles: For which one.
181 00:19:27.440 ⇒ 00:19:29.870 Miguel de Veyra: Remember we did this for Lushka Holyce.
182 00:19:30.190 ⇒ 00:19:36.969 Miguel de Veyra: and then we were putting it there, and it’s like very expensive. But that was like almost a year ago. So let’s see.
183 00:19:36.990 ⇒ 00:19:37.660 Casie Aviles: Hmm.
184 00:19:40.020 ⇒ 00:19:43.420 Miguel de Veyra: Do we have an open AI key for this with them? I’d love to try this out.
185 00:19:44.725 ⇒ 00:19:46.379 Uttam Kumaran: Yeah, you can go generate one.
186 00:19:47.750 ⇒ 00:19:49.639 Miguel de Veyra: For Gemini. Oh, yeah, yeah.
187 00:19:49.948 ⇒ 00:19:53.340 Uttam Kumaran: Yeah, we have, we have Google, it’s just gonna it’s like.
188 00:19:53.340 ⇒ 00:19:54.110 Miguel de Veyra: Yeah, yeah.
189 00:19:54.110 ⇒ 00:20:03.290 Uttam Kumaran: I don’t even know where to go. But I think if you go into Google, Llm. Studio, create one, just create one and call it. ABC, so we can.
190 00:20:03.290 ⇒ 00:20:03.940 Miguel de Veyra: Yeah, it’s right.
191 00:20:03.940 ⇒ 00:20:05.489 Uttam Kumaran: Update it. And then, yeah.
192 00:20:05.490 ⇒ 00:20:08.090 Miguel de Veyra: Okay, yeah, sure. I’ll I’ll create that.
193 00:20:08.310 ⇒ 00:20:11.439 Miguel de Veyra: And yeah, I think for that. ABC,
194 00:20:11.880 ⇒ 00:20:14.889 Miguel de Veyra: we’re pretty much like on. Oh, shit sorry, my bad.
195 00:20:14.890 ⇒ 00:20:31.579 Uttam Kumaran: It’s the only other thing for them is like, I want to start working on the Eval data set. I know I took that on, but maybe I’ll try to see if I can put that together, and we can start working on how to how to run that I’m pretty sure we got. We got it working in vellum, right in terms of like being able to call.
196 00:20:32.677 ⇒ 00:20:33.930 Miguel de Veyra: Yeah, yeah, yeah.
197 00:20:33.930 ⇒ 00:20:38.080 Uttam Kumaran: Yeah, okay, okay, yeah. I know. Jay. And I saw the video. So okay, cool. So.
198 00:20:38.080 ⇒ 00:20:42.739 Miguel de Veyra: But it’s not the one connected by the way on their demo. Should I move this.
199 00:20:44.434 ⇒ 00:20:46.499 Uttam Kumaran: I guess. What do you mean?
200 00:20:47.250 ⇒ 00:20:50.280 Miguel de Veyra: And this one is, I’m sorry. Go, Jenna.
201 00:20:51.610 ⇒ 00:20:58.959 Janna Wong: Oh, that one is directly connected through Miguel’s 8. N, so yeah, sorry.
202 00:20:58.960 ⇒ 00:21:01.430 Miguel de Veyra: So should I move this to value.
203 00:21:01.870 ⇒ 00:21:04.000 Uttam Kumaran: Oh yes, yes.
204 00:21:04.000 ⇒ 00:21:05.109 Miguel de Veyra: Okay, yeah, sure.
205 00:21:05.250 ⇒ 00:21:08.409 Miguel de Veyra: I’ll I’ll work on it tonight and hopefully, tomorrow.
206 00:21:08.410 ⇒ 00:21:12.039 Uttam Kumaran: Just double check that like it’s it’s not. Doesn’t get like way slower. But yeah.
207 00:21:12.280 ⇒ 00:21:15.249 Miguel de Veyra: Yeah, yeah, I. That’s that’s the other thing. I wanna make sure.
208 00:21:17.190 ⇒ 00:21:23.069 Uttam Kumaran: Yeah, we we’ll have to figure some of these out. And then I’m gonna start working on data set as well.
209 00:21:23.070 ⇒ 00:21:23.560 Miguel de Veyra: And
210 00:21:25.320 ⇒ 00:21:42.579 Miguel de Veyra: and then cause one of the things we eventually wanna do is the quality score right? Do you know a way, for example, once a record is created over here that it you know that we can trigger somewhere, that hey? A record is here. Can you process this and then update the quality score of this record.
211 00:21:44.920 ⇒ 00:21:48.719 Uttam Kumaran: Oh, okay, yeah, yeah. So if it’s in
212 00:21:49.140 ⇒ 00:21:54.430 Uttam Kumaran: now, it’s like more my world, I can. I can help you on how we and it. But this is for.
213 00:21:54.700 ⇒ 00:21:58.719 Uttam Kumaran: oh, okay, this is for the logs from the agent. Okay,
214 00:22:01.270 ⇒ 00:22:04.290 Miguel de Veyra: Basically show you. I guess if you give me.
215 00:22:05.460 ⇒ 00:22:08.340 Uttam Kumaran: If you give me that.
216 00:22:08.712 ⇒ 00:22:10.200 Miguel de Veyra: You’re cutting off like.
217 00:22:10.200 ⇒ 00:22:11.210 Uttam Kumaran: My fun.
218 00:22:15.060 ⇒ 00:22:17.089 Uttam Kumaran: Okay, one second. Give me. Give me a second.
219 00:22:17.090 ⇒ 00:22:18.119 Miguel de Veyra: Yeah. No worries.
220 00:22:38.940 ⇒ 00:22:40.230 Miguel de Veyra: Sorry guys. Wait a minute.
221 00:23:26.870 ⇒ 00:23:27.950 Uttam Kumaran: Hey, guys, can you hear me now?
222 00:23:28.360 ⇒ 00:23:30.070 Miguel de Veyra: Oh, yeah. Way. Better. Now. Yeah.
223 00:23:30.070 ⇒ 00:23:33.497 Uttam Kumaran: Okay, okay. Yeah. What I was saying is that
224 00:23:34.960 ⇒ 00:23:51.839 Uttam Kumaran: one. Yeah. We can move the we can move this to vellum and see how it works. I think the other thing I just wanna start working on the Eval data set so that I can start to get feedback from them on that was there another item? Wasn’t.
225 00:23:52.169 ⇒ 00:23:58.769 Miguel de Veyra: This one, basically how we oh, shit what happened like, for example, our record is created. And then we want
226 00:23:59.100 ⇒ 00:24:03.907 Uttam Kumaran: Oh, yeah, yeah, okay, yeah. So there’s 2 ways to do this one. We can. We can.
227 00:24:04.790 ⇒ 00:24:08.630 Uttam Kumaran: we could run this as part of like.
228 00:24:09.290 ⇒ 00:24:14.789 Uttam Kumaran: basically like the actual vellum step. Or we could run this in post.
229 00:24:15.540 ⇒ 00:24:20.980 Miguel de Veyra: Ye. Yeah, I mean, there’s there’s a way to do it honestly, that’s easy. Just add like a
230 00:24:21.260 ⇒ 00:24:25.289 Miguel de Veyra: a step here, but it would, of course, add to the execution time.
231 00:24:26.090 ⇒ 00:24:27.019 Miguel de Veyra: so I’m not sure we want.
232 00:24:27.020 ⇒ 00:24:32.319 Uttam Kumaran: Yeah. So I think what we can do is like in we can batch, score them.
233 00:24:32.570 ⇒ 00:24:34.430 Uttam Kumaran: probably in a separate process.
234 00:24:34.620 ⇒ 00:24:46.166 Uttam Kumaran: I think. 2 things. One, I guess we can decide whether we want to do that. Also in Windmill. I can show you how to do that in Snowflake, and we can use, I think, like
235 00:24:47.000 ⇒ 00:24:48.599 Uttam Kumaran: or what we can do.
236 00:24:48.600 ⇒ 00:24:49.820 Uttam Kumaran: Yeah, go ahead.
237 00:24:49.820 ⇒ 00:25:01.889 Miguel de Veyra: Since, since these are separate conversations, right? We might want to create, like another database like copy, all conversations. And then, at the end of the day relate, you know, that’s only because
238 00:25:02.100 ⇒ 00:25:17.280 Miguel de Veyra: this, of of course, the quality here is gonna the quality score is not really gonna be accurate, since it’s gonna be one back and forth. But if it’s like a set of, you know, back and forth, basically an entire conversation with this conversation. Id, I think that would be a better use of quality score.
239 00:25:17.750 ⇒ 00:25:18.580 Miguel de Veyra: Right?
240 00:25:22.650 ⇒ 00:25:26.449 Miguel de Veyra: So it’s like the entire conversation we’re grading, not just one back and forth.
241 00:25:26.590 ⇒ 00:25:29.040 Uttam Kumaran: Oh, so you’re saying, like, Yeah, I guess
242 00:25:29.470 ⇒ 00:25:35.830 Uttam Kumaran: I guess we could do both we could. So typically, this is described as like a Conversation versus the Messages Right.
243 00:25:35.830 ⇒ 00:25:36.970 Miguel de Veyra: Yes, yes.
244 00:25:37.261 ⇒ 00:25:54.729 Uttam Kumaran: So why don’t we record all messages? And then we can also have a conversation score in a separate table. So for now just write all back and forth messages and just make sure that there is a conversation perfect. Yeah, there’s a conversation. Id, there’s an input and then do you have the user like that sent it. And.
245 00:25:55.007 ⇒ 00:25:58.609 Miguel de Veyra: Well, there, there’s no way for us to track it, because it’s here.
246 00:25:59.770 ⇒ 00:26:05.019 Uttam Kumaran: Well, no, no, not like who but like is it the AI versus? Is it the user.
247 00:26:05.275 ⇒ 00:26:07.060 Miguel de Veyra: Input is always gonna be the user.
248 00:26:09.260 ⇒ 00:26:11.900 Uttam Kumaran: Oh, okay, great. Okay, okay, okay. Okay.
249 00:26:13.270 ⇒ 00:26:29.374 Uttam Kumaran: okay, cool. So then, yeah, I think I think ideally, we leave it at that. And then we just have messages and conversations. So just keep keep tracking conversation. Id, and then I’ll show you we can. We can make another process that basically scores the conversations as well, and then I’ll show you how to how we can bring this into
250 00:26:30.560 ⇒ 00:26:31.940 Uttam Kumaran: into a dashboard.
251 00:26:31.940 ⇒ 00:26:33.009 Miguel de Veyra: Okay, okay, sure.
252 00:26:33.800 ⇒ 00:26:36.970 Miguel de Veyra: Okay, yeah. Wait. Let me go here.
253 00:26:37.780 ⇒ 00:26:40.939 Miguel de Veyra: Yeah. So I guess yeah, we’re we’re on track.
254 00:26:41.700 ⇒ 00:26:44.530 Miguel de Veyra: I guess it depends on if they reply within the next week.
255 00:26:45.730 ⇒ 00:26:51.200 Uttam Kumaran: They’ll reply, I’m gonna try. I mean, I’m gonna try my best to do some Eval work today and get them that, too.
256 00:26:51.350 ⇒ 00:26:53.479 Miguel de Veyra: No, I mean regarding the taxonomy stuff.
257 00:26:53.820 ⇒ 00:26:54.940 Uttam Kumaran: Oh, yeah. Yeah.
258 00:26:54.940 ⇒ 00:26:55.510 Miguel de Veyra: Yeah.
259 00:26:55.510 ⇒ 00:27:02.020 Uttam Kumaran: We’ll see worst case we can do on Friday. But let’s keep pushing, cause what’s gonna happen with clients, anyway. So.
260 00:27:02.020 ⇒ 00:27:05.739 Miguel de Veyra: Yeah, especially, Yvette is doing her annual stuff right? But have you?
261 00:27:06.490 ⇒ 00:27:11.130 Miguel de Veyra: Yeah, yeah, gonna be a bugger. But whatever it’s, it is what it is.
262 00:27:11.270 ⇒ 00:27:16.450 Miguel de Veyra: So. Yeah, I guess this one utam. Do we wanna speak about anything else? Or do we wanna move to Junior.
263 00:27:18.496 ⇒ 00:27:24.889 Uttam Kumaran: Let’s let’s move to Junior. This is really great. Thank you for running. You know, this part like this. It’s really helpful.
264 00:27:24.890 ⇒ 00:27:25.650 Miguel de Veyra: Okay.
265 00:27:26.150 ⇒ 00:27:40.960 Miguel de Veyra: yeah. We actually talked about this earlier, like an hour ago, between me and Casey when she was up, when he was helping me set up we. There’s like an issue here. I’m not sure if you know already, Casey, do you want to expand on this more? Please.
266 00:27:41.510 ⇒ 00:27:43.730 Casie Aviles: Yeah, sure. And I can also share my screen.
267 00:27:43.730 ⇒ 00:27:44.989 Miguel de Veyra: Oh, yeah, yeah, I’ll handle.
268 00:27:47.510 ⇒ 00:27:53.120 Casie Aviles: So yeah, basically, I created like this script on windmill. So
269 00:27:54.050 ⇒ 00:28:01.190 Casie Aviles: yeah, the only thing is, I did not use dlt because I I don’t. I couldn’t get the messages for some reason, and
270 00:28:01.550 ⇒ 00:28:06.859 Casie Aviles: I was spending a little too much time, I think so. I just decided to just keep
271 00:28:07.200 ⇒ 00:28:08.510 Casie Aviles: to continue, but
272 00:28:08.920 ⇒ 00:28:16.100 Casie Aviles: without the Lt, so yeah, like I I mentioned before, it’s using like, just snowflake connector and stuff.
273 00:28:16.720 ⇒ 00:28:21.789 Casie Aviles: So yeah, this should trigger like every.
274 00:28:22.040 ⇒ 00:28:27.080 Casie Aviles: And the yeah, it should be scheduled like here, using this time zone
275 00:28:27.330 ⇒ 00:28:34.750 Casie Aviles: 4 pm, so what this does is it will use the slack Api and just get all the messages for a given day
276 00:28:35.669 ⇒ 00:28:42.690 Casie Aviles: and then it should send the messages. For example. It’s a little still, a little messy, like the tables I’ve set up.
277 00:28:43.010 ⇒ 00:28:46.679 Casie Aviles: But yeah, for example, it would look like this. And
278 00:28:47.580 ⇒ 00:28:50.940 Casie Aviles: you know each text over here, there’s an id.
279 00:28:50.940 ⇒ 00:28:51.920 Uttam Kumaran: Nice.
280 00:28:52.380 ⇒ 00:28:53.449 Casie Aviles: Yeah user.
281 00:28:53.450 ⇒ 00:28:54.080 Uttam Kumaran: Solid.
282 00:28:54.760 ⇒ 00:28:55.090 Casie Aviles: Yeah.
283 00:28:55.090 ⇒ 00:28:59.380 Uttam Kumaran: And are you? How are you? How are you getting the messages in now? It’s just through the bot.
284 00:29:00.750 ⇒ 00:29:07.910 Casie Aviles: Yeah, it should be through the bot, because, you know, it’s already it can already see it. Now that it’s added to that channel.
285 00:29:09.510 ⇒ 00:29:11.380 Miguel de Veyra: Nice good workaround.
286 00:29:12.835 ⇒ 00:29:16.969 Casie Aviles: Yeah, I guess the next thing is also similar to
287 00:29:17.220 ⇒ 00:29:19.519 Casie Aviles: what Miguel was talking about with
288 00:29:19.660 ⇒ 00:29:25.139 Casie Aviles: ABC, so it’s going to be about scoring, or yeah, the qualities part for the AI
289 00:29:25.847 ⇒ 00:29:29.149 Casie Aviles: that one. I’ve set up a table already, but
290 00:29:29.630 ⇒ 00:29:35.130 Casie Aviles: yet it’s empty yet, so I guess the next step for me is to just, you know, how do I populate this
291 00:29:36.186 ⇒ 00:29:40.880 Casie Aviles: table? For, you know, like for the assessment and the score?
292 00:29:41.180 ⇒ 00:29:41.860 Casie Aviles: Yeah.
293 00:29:46.164 ⇒ 00:29:48.469 Casie Aviles: Yeah, and also for the alert.
294 00:29:48.790 ⇒ 00:29:53.749 Casie Aviles: I was thinking, something like this, so this is just on Zapier. So
295 00:29:53.930 ⇒ 00:29:56.400 Casie Aviles: it’s going to send something like this.
296 00:29:59.150 ⇒ 00:30:00.000 Casie Aviles: Yeah.
297 00:30:00.590 ⇒ 00:30:06.369 Casie Aviles: So ideally this should be. I mean this right now. It’s just looking at the one.
298 00:30:06.970 ⇒ 00:30:09.670 Casie Aviles: the temporary one that I’ve set up.
299 00:30:10.400 ⇒ 00:30:11.420 Casie Aviles: Hey? Sorry.
300 00:30:11.590 ⇒ 00:30:13.630 Casie Aviles: Yeah, I think. Yeah, it’s this one.
301 00:30:14.020 ⇒ 00:30:17.090 Casie Aviles: But this is the one that’s from Zapier side. So
302 00:30:17.926 ⇒ 00:30:26.360 Casie Aviles: the the, I guess. Also, the next thing is for me to read from Snowflake, and
303 00:30:26.560 ⇒ 00:30:29.570 Casie Aviles: that will trigger the alert. And also the
304 00:30:29.700 ⇒ 00:30:34.179 Casie Aviles: AI part that sends these scores and assessment.
305 00:30:37.470 ⇒ 00:30:41.489 Casie Aviles: Yeah, that yeah, that’s pretty much what I have so far with this.
306 00:30:43.490 ⇒ 00:30:46.577 Uttam Kumaran: Okay. Awesome. I think. Couple of things. One.
307 00:30:48.010 ⇒ 00:30:52.930 Uttam Kumaran: What do you think is the next step? Is it like.
308 00:30:53.510 ⇒ 00:31:00.650 Uttam Kumaran: I guess, one dude. This is great, like I want to. Now, I’m gonna add the brain forge bot to couple other channels. So you have access.
309 00:31:01.125 ⇒ 00:31:07.170 Uttam Kumaran: I think so you’re gonna you’re just basically, for example, let’s say, one client has several channels.
310 00:31:10.663 ⇒ 00:31:15.620 Uttam Kumaran: let’s say, one client has several channels. I think the biggest thing
311 00:31:15.780 ⇒ 00:31:18.990 Uttam Kumaran: is, are you? Gonna are we gonna merge it into just one table.
312 00:31:20.460 ⇒ 00:31:21.100 Casie Aviles: Hmm.
313 00:31:21.100 ⇒ 00:31:22.210 Uttam Kumaran: So like for Javi.
314 00:31:22.210 ⇒ 00:31:22.810 Casie Aviles: Can.
315 00:31:23.230 ⇒ 00:31:31.719 Uttam Kumaran: Okay, I think it’s fine just to have like one client. And then the Channel Id can be different. Because, for example, for Javi, I added the Brainforge bot to 2 channels.
316 00:31:34.350 ⇒ 00:31:36.920 Casie Aviles: Oh, 2 channels! Oh, I thought it was just one channel.
317 00:31:36.920 ⇒ 00:31:37.880 Uttam Kumaran: Pretty sure.
318 00:31:39.740 ⇒ 00:31:40.450 Casie Aviles: Okay.
319 00:31:47.695 ⇒ 00:31:49.195 Uttam Kumaran: But this is okay.
320 00:31:49.700 ⇒ 00:31:57.390 Uttam Kumaran: Second thing is can we is, is windmill hooked up to the
321 00:31:58.650 ⇒ 00:32:02.150 Uttam Kumaran: is windmill hook up to Github by chance, like for version control.
322 00:32:02.990 ⇒ 00:32:07.379 Casie Aviles: I, yeah, yeah, it should be like, I, I think I, yeah, I did set it up before. But
323 00:32:07.500 ⇒ 00:32:14.150 Casie Aviles: I guess the token was yeah, for the the token expired so I could.
324 00:32:14.510 ⇒ 00:32:18.700 Casie Aviles: At best I could do it like manually, so I have to pull up the terminal and
325 00:32:18.850 ⇒ 00:32:21.710 Casie Aviles: run a few commands to sync, get mail to.
326 00:32:21.710 ⇒ 00:32:25.040 Uttam Kumaran: And this is just like windmill Workspace, brain forge AI.
327 00:32:25.600 ⇒ 00:32:26.650 Casie Aviles: Yes.
328 00:32:27.100 ⇒ 00:32:30.690 Uttam Kumaran: Okay, so yeah, let’s make sure that’s automatic.
329 00:32:31.527 ⇒ 00:32:43.330 Uttam Kumaran: And then I would love to start doing like even a little bit of code review. And I’m happy to review any python code. Because now that you’re becoming a data engineer like
330 00:32:43.710 ⇒ 00:32:48.948 Uttam Kumaran: welcome. So you can start, I think it’s it’s gonna help for you to get some feedback and like
331 00:32:49.710 ⇒ 00:32:58.089 Uttam Kumaran: dude. What you’re doing right now is basically all we do on the data side. So it’s that, except like maybe a little bit harder. But I would love to.
332 00:32:58.090 ⇒ 00:32:58.520 Uttam Kumaran: Sorry to
333 00:32:58.520 ⇒ 00:33:16.020 Uttam Kumaran: review. I would love to help review that code that you’re pushing in and sort of get comments on on that, like the review of the python code. So let me know if if we can do that, and what you need from the from Github to make that happen.
334 00:33:16.920 ⇒ 00:33:17.600 Casie Aviles: Okay.
335 00:33:18.590 ⇒ 00:33:19.350 Casie Aviles: Sure.
336 00:33:20.270 ⇒ 00:33:22.583 Uttam Kumaran: I think the second thing also is, I’m gonna
337 00:33:23.550 ⇒ 00:33:29.196 Uttam Kumaran: I’m also gonna suggest that we start to visualize this data in real
338 00:33:30.193 ⇒ 00:33:56.620 Uttam Kumaran: and I’m going to send you some documents on how to initialize real real is our like data visualization tool of choice. We do have an internal instance as well, where you can visualize this data super super easily. I think if I just can get you that access and show you sort of how to develop on that it’ll you’ll you’ll be able to nail it. So I’m gonna try to grab some time so we can set that up.
339 00:33:57.110 ⇒ 00:33:58.835 Uttam Kumaran: And the 3rd thing is,
340 00:33:59.620 ⇒ 00:34:07.930 Uttam Kumaran: yeah, I guess I’m less concerned about the AI scoring more concerned about. Let’s just get make sure all the pipelines are running to to write this.
341 00:34:08.508 ⇒ 00:34:15.190 Uttam Kumaran: And so how does it work? Is it all web book based? Is it like on a is it on like a a script
342 00:34:15.350 ⇒ 00:34:17.140 Uttam Kumaran: for new messages.
343 00:34:18.508 ⇒ 00:34:22.989 Casie Aviles: Yeah, it’s scheduled. So it’s using Cron, I think. Yeah, here. So.
344 00:34:22.989 ⇒ 00:34:25.279 Uttam Kumaran: And then it how does it? How does it get it gets?
345 00:34:25.929 ⇒ 00:34:30.029 Uttam Kumaran: It gets all of the messages it just gets like one time.
346 00:34:30.409 ⇒ 00:34:32.749 Uttam Kumaran: like, how does it? How does it do the overlap.
347 00:34:33.848 ⇒ 00:34:39.949 Casie Aviles: For a given day, like I, I’ve set it to get the messages for like one day. So
348 00:34:41.146 ⇒ 00:34:43.600 Casie Aviles: yeah, I that’s how it works like, if
349 00:34:45.500 ⇒ 00:34:48.089 Casie Aviles: sorry. Did did I answer the question.
350 00:34:49.052 ⇒ 00:34:53.250 Uttam Kumaran: Yeah. So you just do on the day, meaning like, There, there shouldn’t be any overlap.
351 00:34:55.000 ⇒ 00:34:56.159 Casie Aviles: Yeah. Yeah.
352 00:34:56.860 ⇒ 00:34:57.530 Uttam Kumaran: Okay?
353 00:34:59.140 ⇒ 00:35:07.969 Uttam Kumaran: And then are you? Is it? Is it using? Are you using dlt by chance, or, how are you actually doing the right to Snowflake. It’s just using the the Snowflake connector.
354 00:35:08.400 ⇒ 00:35:18.010 Casie Aviles: Yeah, the thing with Dlt is, yeah, I I could. For some reason I couldn’t get the messages like I was able to get like data about channels and the users. But
355 00:35:18.948 ⇒ 00:35:27.280 Casie Aviles: I I don’t know. I couldn’t get it to work. So I I don’t know. I just skip it for now, because I just wanted to bring it to Snowflake already.
356 00:35:27.560 ⇒ 00:35:27.940 Miguel de Veyra: Of.
357 00:35:27.940 ⇒ 00:35:28.540 Uttam Kumaran: Okay.
358 00:35:28.540 ⇒ 00:35:30.249 Miguel de Veyra: Casey. Quick question. Sorry.
359 00:35:30.940 ⇒ 00:35:36.010 Miguel de Veyra: Can you get all the ids of of slack users.
360 00:35:37.500 ⇒ 00:35:42.940 Casie Aviles: Slack users. Yeah, I think should be able to, since I’m already passing that. Anyway, here.
361 00:35:43.650 ⇒ 00:35:47.710 Miguel de Veyra: Yeah, I mean, like, who who is it named to like, for example, you know.
362 00:35:47.980 ⇒ 00:35:49.749 Casie Aviles: Oh, yeah. Here the username.
363 00:35:50.150 ⇒ 00:35:51.590 Miguel de Veyra: Okay. Okay. Nice. Nice.
364 00:35:52.530 ⇒ 00:35:54.549 Miguel de Veyra: Okay. Yeah. No worries. Then. Thank you.
365 00:36:03.790 ⇒ 00:36:06.830 Casie Aviles: Yeah, I guess that’s pretty much it from my end.
366 00:36:13.654 ⇒ 00:36:14.000 Miguel de Veyra: Utah.
367 00:36:14.000 ⇒ 00:36:17.483 Uttam Kumaran: Okay? And then I guess my last, my last piece is
368 00:36:18.915 ⇒ 00:36:23.819 Uttam Kumaran: How are we going to do the alerting like to the
369 00:36:27.170 ⇒ 00:36:33.190 Uttam Kumaran: to the leadership team like, I guess, like, if they’re not message wasn’t gonna get sent like, how do we think we should do that?
370 00:36:34.626 ⇒ 00:36:39.589 Miguel de Veyra: Guess we could come up with like some sort of slack message that says, like, here’s all the clients. And here’s
371 00:36:40.470 ⇒ 00:36:42.310 Miguel de Veyra: I think Casey, earlier.
372 00:36:43.750 ⇒ 00:36:50.660 Casie Aviles: Yeah, it’s very. I guess it’s you know. It’s just a naive in solution right now. But.
373 00:36:50.660 ⇒ 00:36:58.580 Uttam Kumaran: Oh, nice! Can we do that? Can we do like literally, like a quick scorecard, like client
374 00:36:58.720 ⇒ 00:37:01.040 Uttam Kumaran: number of messages from our team.
375 00:37:01.560 ⇒ 00:37:02.729 Miguel de Veyra: Hi got it!
376 00:37:02.730 ⇒ 00:37:07.450 Uttam Kumaran: Like a green check, and then let’s just do that for now.
377 00:37:08.440 ⇒ 00:37:16.230 Casie Aviles: Oh, okay, okay, so like I, I’ll add that I will add to here, like the number of messages sent for that day
378 00:37:16.490 ⇒ 00:37:17.559 Casie Aviles: and channel.
379 00:37:26.800 ⇒ 00:37:33.049 Uttam Kumaran: Yeah, I think, just like, yeah, I’m trying to think about what the best kind of like. Let me send a version of this
380 00:37:33.890 ⇒ 00:37:44.409 Uttam Kumaran: like daily client communication report, and then it’ll be like
381 00:38:00.030 ⇒ 00:38:03.400 Uttam Kumaran: it’ll be like Javi coffee.
382 00:38:10.620 ⇒ 00:38:18.680 Uttam Kumaran: 10 messages from team, and then kind of like
383 00:38:22.050 ⇒ 00:38:25.720 Uttam Kumaran: something like that, easy to digest.
384 00:38:26.091 ⇒ 00:38:27.579 Casie Aviles: Okay, straight straight through.
385 00:38:27.580 ⇒ 00:38:31.819 Uttam Kumaran: Because, because, yeah, because, okay, well, yeah, I mean, you guys know me like, I
386 00:38:31.970 ⇒ 00:38:42.880 Uttam Kumaran: cool. I just, I’ll just look at it really quickly. The second thing is, yeah, we’re gonna have, like 5 or 6 clients. I want it to be really clear where we need to take action. Then this also. This framework gives us the ability to add quality score.
387 00:38:43.290 ⇒ 00:38:43.960 Uttam Kumaran: There.
388 00:38:46.860 ⇒ 00:38:51.669 Casie Aviles: Okay? And then we want to have this sent to the leadership channel.
389 00:38:52.130 ⇒ 00:38:52.680 Miguel de Veyra: Oh!
390 00:38:52.680 ⇒ 00:38:55.680 Uttam Kumaran: Let’s just do. Let no, let’s just do it to test channel.
391 00:38:56.643 ⇒ 00:38:57.356 Miguel de Veyra: Okay.
392 00:38:58.070 ⇒ 00:38:58.580 Casie Aviles: Okay. Yeah.
393 00:38:59.249 ⇒ 00:39:05.179 Uttam Kumaran: Yeah. And then also can do I do you need Nico to be in this test channel? Because.
394 00:39:05.180 ⇒ 00:39:05.729 Miguel de Veyra: I think he is.
395 00:39:05.730 ⇒ 00:39:06.060 Casie Aviles: Here.
396 00:39:06.060 ⇒ 00:39:06.499 Uttam Kumaran: I don’t know.
397 00:39:06.880 ⇒ 00:39:07.439 Casie Aviles: Hey! Johnny!
398 00:39:07.440 ⇒ 00:39:16.610 Uttam Kumaran: He is there but like, do like. Can I remove him? Because I don’t want anybody outside of AI team to be here because we’re going to be testing stuff that like I kind of, I don’t know.
399 00:39:16.730 ⇒ 00:39:20.030 Uttam Kumaran: I don’t want people to be bothered reading this and sort of until we we make.
400 00:39:20.338 ⇒ 00:39:21.570 Miguel de Veyra: Yeah, it’s a distraction.
401 00:39:21.570 ⇒ 00:39:25.480 Casie Aviles: Oh, okay, yeah, I understand. I mean, yeah, I I guess if we could.
402 00:39:26.232 ⇒ 00:39:27.649 Casie Aviles: Yeah, remove them.
403 00:39:28.300 ⇒ 00:39:30.779 Uttam Kumaran: And then I’m gonna I’ll add, I’ll add Jana there, too.
404 00:39:31.100 ⇒ 00:39:33.679 Miguel de Veyra: Oh, yeah, I can. I can do that with her. She don’t have to.
405 00:39:34.350 ⇒ 00:39:35.000 Uttam Kumaran: I was already.
406 00:39:35.000 ⇒ 00:39:35.830 Uttam Kumaran: Oh, there!
407 00:39:35.830 ⇒ 00:39:37.740 Uttam Kumaran: Cool, alright, cool, alright! Great!
408 00:39:41.620 ⇒ 00:39:43.479 Miguel de Veyra: Yeah, I think that’s pretty much it.
409 00:39:44.020 ⇒ 00:39:51.869 Miguel de Veyra: Autumn, I know, running out of time. I’ll just. I’m not sure I explained this to you. But we discuss I discussed with Casey like a day ago.
410 00:39:53.114 ⇒ 00:39:57.560 Miguel de Veyra: Basically, here we kinda this is the timeline, right?
411 00:39:58.590 ⇒ 00:40:19.290 Miguel de Veyra: So this is done. Initiative 3 is done. We already set the meeting with Nico. So we, you know it’s about the criteria he’s most interested in when we spoke to him, and it’s part of the this feature that we’re gonna look into like a while now, I like in the coming weeks. So right now, we just want to finish this too. First.st
412 00:40:19.990 ⇒ 00:40:20.600 Uttam Kumaran: Okay.
413 00:40:20.600 ⇒ 00:40:21.700 Miguel de Veyra: So ideally.
414 00:40:21.700 ⇒ 00:40:29.440 Uttam Kumaran: I like having everything here, too. This is actually like, I mean, again, we don’t need to use tickets. If you guys feel like this is pretty good order for organization.
415 00:40:30.722 ⇒ 00:40:36.410 Miguel de Veyra: Yeah, yeah, I’m a visual person. So I guess so. Yeah.
416 00:40:36.640 ⇒ 00:40:41.589 Miguel de Veyra: so, yeah, I mean, pm, 1, 0, 1. What you taught me was that, you know, even if cause.
417 00:40:42.040 ⇒ 00:40:45.049 Miguel de Veyra: Nico was pushing very hard on this, but the timeline was.
418 00:40:45.770 ⇒ 00:40:46.280 Uttam Kumaran: Let’s do first.st
419 00:40:46.280 ⇒ 00:40:48.569 Miguel de Veyra: So we want to get this done 1st and then.
420 00:40:49.180 ⇒ 00:40:49.790 Uttam Kumaran: Yes.
421 00:40:49.790 ⇒ 00:40:52.110 Miguel de Veyra: Yeah. And then this one. So yeah.
422 00:40:53.310 ⇒ 00:40:53.889 Miguel de Veyra: And then I
423 00:40:54.530 ⇒ 00:41:06.400 Miguel de Veyra: created tickets for each initiative. Just so there’s still tickets for it. But yeah, generally speaking, you know, everything is here. And then we’re just task done. Yeah, I think that’s pretty much it.
424 00:41:07.400 ⇒ 00:41:08.110 Uttam Kumaran: Okay.
425 00:41:09.163 ⇒ 00:41:09.870 Miguel de Veyra: And then.
426 00:41:09.870 ⇒ 00:41:10.210 Uttam Kumaran: Understood.
427 00:41:10.210 ⇒ 00:41:13.349 Miguel de Veyra: Question tomorrow. We have a meeting with them. Right?
428 00:41:13.540 ⇒ 00:41:14.679 Miguel de Veyra: ABC, people.
429 00:41:14.680 ⇒ 00:41:18.380 Uttam Kumaran: We do have a meeting with them. The meeting is at.
430 00:41:18.740 ⇒ 00:41:19.990 Miguel de Veyra: 1230, yeah, yeah.
431 00:41:20.320 ⇒ 00:41:21.250 Uttam Kumaran: Yes.
432 00:41:22.361 ⇒ 00:41:29.109 Uttam Kumaran: Yeah, if you if I know Janet, it’s kind of late. But, Miguel, if you can be there, that would be really really great.
433 00:41:29.110 ⇒ 00:41:29.470 Miguel de Veyra: Now.
434 00:41:29.886 ⇒ 00:41:34.880 Uttam Kumaran: Again. I think at minimum, we’re gonna I wanna discuss the Bible.
435 00:41:35.290 ⇒ 00:41:36.240 Miguel de Veyra: Yeah, yeah.
436 00:41:36.240 ⇒ 00:41:38.540 Uttam Kumaran: And talk about eval questions.
437 00:41:38.940 ⇒ 00:41:46.790 Uttam Kumaran: And then also, I want to show sort of like the vellum setup. And so how we’re getting logs. That would be really, really great.
438 00:41:47.470 ⇒ 00:41:48.500 Miguel de Veyra: Yep, okay.
439 00:41:49.070 ⇒ 00:41:55.730 Miguel de Veyra: So yeah, I’ll have to move the basically this one tonight to value.
440 00:41:57.530 ⇒ 00:42:00.789 Miguel de Veyra: Okay, yeah, I think that’s pretty much it on my side. Luton.
441 00:42:01.450 ⇒ 00:42:02.200 Uttam Kumaran: Okay.
442 00:42:03.954 ⇒ 00:42:07.159 Uttam Kumaran: Okay, alright, thanks. Guys. Appreciate it.
443 00:42:07.160 ⇒ 00:42:08.670 Miguel de Veyra: Thanks. Everyone have a good day. Bye, bye.
444 00:42:08.670 ⇒ 00:42:09.430 Janna Wong: Thank you.
445 00:42:09.430 ⇒ 00:42:10.570 Uttam Kumaran: Thank you. Bye.