Meeting Title: ABC Final Output Planning Date: 2025-02-26 Meeting participants: Miguel De Veyra, Uttam Kumaran
WEBVTT
1 00:00:14.130 ⇒ 00:00:18.489 Uttam Kumaran: How you doing, hey? Slack is down. So I I was just like.
2 00:00:18.490 ⇒ 00:00:19.240 Miguel de Veyra: Oh, yeah.
3 00:00:19.240 ⇒ 00:00:23.369 Uttam Kumaran: I thought you. I thought nice. I joined the meeting. Nobody’s there. So I was like, Okay, maybe at the wrong meeting.
4 00:00:24.000 ⇒ 00:00:30.870 Miguel de Veyra: Oh, okay, okay. I actually joined like the AI you meet with you. Okay?
5 00:00:30.870 ⇒ 00:00:31.520 Uttam Kumaran: Okay, cool.
6 00:00:32.150 ⇒ 00:00:41.700 Uttam Kumaran: Yeah, I’m just gonna be. I’ll just be listening. I’m going to go to a coffee shop to get some work done. But yeah, I I think, basically, this meeting is just going through that list right.
7 00:00:43.005 ⇒ 00:00:46.250 Miguel de Veyra: Isn’t it about discussing like the final output.
8 00:00:48.340 ⇒ 00:00:49.909 Uttam Kumaran: Oh, yeah, but I guess, like.
9 00:00:50.090 ⇒ 00:00:52.630 Uttam Kumaran: have we done the Eval data set yet?
10 00:00:52.930 ⇒ 00:00:55.219 Miguel de Veyra: No, no, it’s it’s later, like an hour from now.
11 00:00:55.650 ⇒ 00:00:58.679 Uttam Kumaran: Oh, okay, so so there’s just 2.
12 00:00:59.010 ⇒ 00:00:59.880 Miguel de Veyra: Yeah, yeah.
13 00:01:00.290 ⇒ 00:01:02.749 Uttam Kumaran: Okay. I don’t think I’m on that meeting, so I didn’t know.
14 00:01:04.319 ⇒ 00:01:06.859 Miguel de Veyra: Yvette is gonna be there in Janice.
15 00:01:08.630 ⇒ 00:01:11.010 Uttam Kumaran: Wait, then, what is this meeting just between me and you?
16 00:01:11.010 ⇒ 00:01:11.670 Miguel de Veyra: Yeah.
17 00:01:12.000 ⇒ 00:01:13.900 Uttam Kumaran: Oh, okay, sorry.
18 00:01:14.963 ⇒ 00:01:22.309 Uttam Kumaran: Okay. Okay, okay, all right. I’ll let you go. Sorry I I thought I. For some reason I thought this was that meeting. Okay.
19 00:01:23.020 ⇒ 00:01:24.090 Miguel de Veyra: Oh, okay.
20 00:01:25.570 ⇒ 00:01:28.340 Uttam Kumaran: Okay, cool, like, I’m yeah. However.
21 00:01:29.200 ⇒ 00:01:32.719 Uttam Kumaran: how best do you want to take this? I guess, like I’ll I’ll let you lead.
22 00:01:33.730 ⇒ 00:01:40.649 Miguel de Veyra: I think this was, I think, the cause you mentioned to book this meeting last Monday, and then it’s just about
23 00:01:41.500 ⇒ 00:01:46.039 Miguel de Veyra: what do you call it the dashboard where we’ll put their admin stuff.
24 00:01:47.450 ⇒ 00:01:56.209 Uttam Kumaran: Okay, okay, okay, so yeah, I guess there’s 2 questions. One is, I want to show the real dashboard. I think. Second is.
25 00:01:56.907 ⇒ 00:02:00.170 Uttam Kumaran: I think we want to have like.
26 00:02:00.510 ⇒ 00:02:08.959 Uttam Kumaran: I don’t know. I think we want to think about like what the final product is like. How hard do you think it would be for them to have a login behind Brainforge
27 00:02:09.880 ⇒ 00:02:11.679 Uttam Kumaran: to go access everything.
28 00:02:12.280 ⇒ 00:02:14.939 Uttam Kumaran: And like, I guess we should talk through what is everything. Basically.
29 00:02:14.940 ⇒ 00:02:20.660 Miguel de Veyra: Yeah, yeah, cause it’s gonna be all in Google anymore, anyway. Right?
30 00:02:21.270 ⇒ 00:02:24.219 Uttam Kumaran: Yeah, so what do we want to give them access to like.
31 00:02:25.340 ⇒ 00:02:26.070 Miguel de Veyra: Dashboard!
32 00:02:26.426 ⇒ 00:02:30.349 Uttam Kumaran: Are there any admin things that they need to do like.
33 00:02:31.440 ⇒ 00:02:32.010 Miguel de Veyra: Okay.
34 00:02:32.010 ⇒ 00:02:32.430 Uttam Kumaran: I mean, like.
35 00:02:32.430 ⇒ 00:02:36.439 Miguel de Veyra: Not really because they can just change the information and will always get it.
36 00:02:36.900 ⇒ 00:02:40.450 Uttam Kumaran: I guess it would be great to also share what the system prompts are.
37 00:02:41.700 ⇒ 00:02:45.730 Miguel de Veyra: Alright, yeah, I guess that could be good.
38 00:02:51.810 ⇒ 00:02:56.349 Miguel de Veyra: So like a react, is this the completely different app than the one we have.
39 00:03:00.233 ⇒ 00:03:06.410 Uttam Kumaran: No, I I just think it’s like probably similar. But I basically would like to have it
40 00:03:07.020 ⇒ 00:03:12.130 Uttam Kumaran: as a standalone thing. But for now we can just be displaying the dashboard.
41 00:03:12.680 ⇒ 00:03:13.290 Miguel de Veyra: Yep.
42 00:03:13.620 ⇒ 00:03:14.350 Miguel de Veyra: Okay.
43 00:03:14.350 ⇒ 00:03:16.394 Uttam Kumaran: This doesn’t have to be anything beyond that.
44 00:03:16.650 ⇒ 00:03:19.090 Miguel de Veyra: Yeah, it could be like, you know, after the
45 00:03:19.380 ⇒ 00:03:23.829 Miguel de Veyra: 1st month, we can probably create something like App, that the app that.
46 00:03:23.830 ⇒ 00:03:32.500 Uttam Kumaran: Is there anything else they like need to like? See there? I mean, we could also put like embed, our like, our task list
47 00:03:32.660 ⇒ 00:03:34.629 Uttam Kumaran: of like what we’re working on there.
48 00:03:35.990 ⇒ 00:03:37.529 Miguel de Veyra: Yeah. Embed the mural.
49 00:03:38.990 ⇒ 00:03:42.312 Uttam Kumaran: Dude. The mirror is ugly. I got it. Well, I’ll get you something that looks better.
50 00:03:45.268 ⇒ 00:03:53.190 Miguel de Veyra: Yeah, I mean, that’s pretty much it. I guess we’ll know this Friday, because this will be the last meeting before we deploy.
51 00:03:53.650 ⇒ 00:03:54.280 Uttam Kumaran: Okay.
52 00:03:54.280 ⇒ 00:03:56.299 Miguel de Veyra: I mean Tim hasn’t replied to me. By the way.
53 00:03:56.300 ⇒ 00:03:58.260 Uttam Kumaran: I’m gonna I’m gonna push on, Tim.
54 00:03:58.460 ⇒ 00:04:04.440 Uttam Kumaran: After this, like for the next hour, I’m I’m basically just
55 00:04:04.620 ⇒ 00:04:09.560 Uttam Kumaran: developing the next the rest of the day. It’s like a dream come true for me.
56 00:04:09.923 ⇒ 00:04:11.740 Miguel de Veyra: Oh, yeah, it’s Wednesday. No.
57 00:04:12.510 ⇒ 00:04:14.740 Uttam Kumaran: Yeah, dude. Finally.
58 00:04:14.970 ⇒ 00:04:15.950 Miguel de Veyra: Finally.
59 00:04:16.579 ⇒ 00:04:21.229 Uttam Kumaran: And then, okay, so yeah, I dude, I just really want us to finish that eval set.
60 00:04:21.800 ⇒ 00:04:25.409 Uttam Kumaran: And I’m gonna get you the I’m gonna get you the embed stuff for real
61 00:04:28.850 ⇒ 00:04:30.669 Miguel de Veyra: So do I create like App, that
62 00:04:31.140 ⇒ 00:04:33.820 Miguel de Veyra: brain forge where they could, you know, see it, for now.
63 00:04:33.820 ⇒ 00:04:42.290 Uttam Kumaran: Yeah, can we create like ABC, home and commercial.brainforge.ai, I I want them to be able to log in
64 00:04:42.430 ⇒ 00:04:44.399 Uttam Kumaran: with some credentials.
65 00:04:44.963 ⇒ 00:04:54.589 Uttam Kumaran: I mean ideally, we could do with Google like I would love for them to log in with their Google account. I don’t know if, if, like doing off that way is really like hard on your end.
66 00:04:54.900 ⇒ 00:05:00.030 Miguel de Veyra: Oh, it’s I’ve done it before. I just you know. It was a couple of years back. But yeah.
67 00:05:00.030 ⇒ 00:05:04.040 Uttam Kumaran: I mean dude. Just ask chat, gpt, or, yeah, you start honestly. Try to use bolt.
68 00:05:05.240 ⇒ 00:05:06.449 Miguel de Veyra: Okay. Okay. Sure.
69 00:05:07.370 ⇒ 00:05:08.599 Uttam Kumaran: Have you tried it yet.
70 00:05:08.600 ⇒ 00:05:09.860 Miguel de Veyra: No, no, not yet.
71 00:05:10.230 ⇒ 00:05:12.839 Uttam Kumaran: Should try it. Dude, it’s really good. Bolt dot new.
72 00:05:13.265 ⇒ 00:05:16.419 Uttam Kumaran: They’re one of our clients. You probably heard of lovable.
73 00:05:16.970 ⇒ 00:05:18.180 Miguel de Veyra: Lovable.
74 00:05:18.710 ⇒ 00:05:20.400 Uttam Kumaran: Oh, dude, you’re out of the game. Bro.
75 00:05:20.970 ⇒ 00:05:21.540 Miguel de Veyra: No.
76 00:05:21.940 ⇒ 00:05:25.010 Uttam Kumaran: You’re too busy now. Now you’re not like learning all the new tools.
77 00:05:25.715 ⇒ 00:05:26.420 Miguel de Veyra: Yeah.
78 00:05:29.828 ⇒ 00:05:35.970 Miguel de Veyra: But yeah, I mean, in terms of like progress, the Csv stuff, the sheets
79 00:05:36.300 ⇒ 00:05:41.150 Miguel de Veyra: that should be good. I’m just finalizing some stuff off. But the base logic is there.
80 00:05:41.330 ⇒ 00:05:43.300 Uttam Kumaran: Wait. What is so? What does that work.
81 00:05:43.751 ⇒ 00:05:46.170 Miguel de Veyra: Can I share my screen, or are you like on the road.
82 00:05:46.700 ⇒ 00:05:47.630 Uttam Kumaran: Yeah, that’s fine.
83 00:05:48.450 ⇒ 00:05:54.750 Miguel de Veyra: Okay, wait. Let me wait. Let me just get this. Put this here
84 00:05:59.680 ⇒ 00:06:01.430 Miguel de Veyra: so that looks something like this.
85 00:06:03.260 ⇒ 00:06:08.049 Miguel de Veyra: So it’s basically like just this, but like a more.
86 00:06:08.260 ⇒ 00:06:15.040 Miguel de Veyra: you know, it’s all it’s always up to date. Version calls directly from.
87 00:06:15.790 ⇒ 00:06:17.820 Uttam Kumaran: Oh, it go! It pulls directly from the spreadsheet.
88 00:06:17.820 ⇒ 00:06:24.470 Miguel de Veyra: Yep. And then the way, yeah. And then the way it works, cause it’s kinda messed up if you see it here. Right?
89 00:06:25.110 ⇒ 00:06:30.649 Miguel de Veyra: So we wrote the code, basically, that just runs it. So now we have access to it.
90 00:06:31.910 ⇒ 00:06:33.400 Miguel de Veyra: So here is, you know.
91 00:06:33.400 ⇒ 00:06:34.680 Miguel de Veyra: Oh, yeah.
92 00:06:34.680 ⇒ 00:06:35.770 Uttam Kumaran: Let’s go.
93 00:06:36.620 ⇒ 00:06:37.759 Miguel de Veyra: We’re rodent. Yeah, yeah.
94 00:06:37.760 ⇒ 00:06:41.500 Uttam Kumaran: This is sick. Okay, but the the only thing is the code.
95 00:06:42.370 ⇒ 00:06:44.759 Uttam Kumaran: Like it’s gonna be hard to maintain like.
96 00:06:45.030 ⇒ 00:06:48.940 Uttam Kumaran: I wonder if we can write something that just gets the columns like instead of
97 00:06:49.040 ⇒ 00:06:54.210 Uttam Kumaran: like, instead of having to look at all the column names, you should just programmatically get the columns.
98 00:06:56.140 ⇒ 00:06:57.120 Miguel de Veyra: Yeah.
99 00:06:57.120 ⇒ 00:07:03.420 Uttam Kumaran: Right like, how would we combine all those code into something that’s like? No matter what the spreadsheet is, you can get all the columns and rows.
100 00:07:06.500 ⇒ 00:07:09.850 Miguel de Veyra: Yeah, I mean, we could look something like that eventually. But
101 00:07:10.140 ⇒ 00:07:11.950 Miguel de Veyra: so this is like the 1st version.
102 00:07:12.150 ⇒ 00:07:13.769 Uttam Kumaran: Okay. Okay. Okay. Yeah.
103 00:07:14.510 ⇒ 00:07:19.699 Miguel de Veyra: But yeah, basically, you know, the Jason is here. And then the area.
104 00:07:19.810 ⇒ 00:07:25.330 Miguel de Veyra: It’s very similar to the the current structures. Just so, you know, there’s relations.
105 00:07:26.330 ⇒ 00:07:29.630 Miguel de Veyra: But yeah, and then the idea is, if
106 00:07:29.910 ⇒ 00:07:36.640 Miguel de Veyra: the idea here is, for example, you you’re given, for example, here the question was
107 00:07:36.910 ⇒ 00:07:40.520 Miguel de Veyra: for this specific zip code which is in Austin.
108 00:07:42.340 ⇒ 00:07:54.889 Miguel de Veyra: Oh, wait! Sorry I am here like this Zip code is in Austin. So now you don’t have to look at anywhere else. So now we only have to pass this context into here. So it’s not as big.
109 00:07:55.800 ⇒ 00:08:00.722 Miguel de Veyra: right? So it saves time. Because if we go here, you notice, like some of this.
110 00:08:01.030 ⇒ 00:08:02.530 Uttam Kumaran: Can you walk through that one more time?
111 00:08:02.820 ⇒ 00:08:09.099 Miguel de Veyra: Okay, cause right now in the current process, right here, everything is
112 00:08:09.440 ⇒ 00:08:18.730 Miguel de Veyra: like we pass everything either way. So even if it’s only an Austin Zip code we still look for, you know, we still pass. The these ones are like as like.
113 00:08:20.380 ⇒ 00:08:22.299 Miguel de Veyra: what do you call this context?
114 00:08:23.900 ⇒ 00:08:29.000 Miguel de Veyra: So now for this one. What happens is when they input something. So for example, here
115 00:08:30.060 ⇒ 00:08:34.350 Miguel de Veyra: we decide. Basically, I gave them, you know.
116 00:08:35.000 ⇒ 00:08:43.609 Miguel de Veyra: the zip codes, and then some just some prompt, and then, Hey, if it’s here, it’s Austin. So it just returned Atx.
117 00:08:43.730 ⇒ 00:08:48.659 Miguel de Veyra: We use switch. So hey? If this is, if this contains, you know.
118 00:08:48.790 ⇒ 00:08:56.239 Miguel de Veyra: Atx, then just route into this one, pull this, turn this into code and then use this as context.
119 00:08:58.450 ⇒ 00:09:01.190 Miguel de Veyra: So it doesn’t have to, you know, check everything.
120 00:09:01.190 ⇒ 00:09:03.309 Uttam Kumaran: So does it know which one to pull from.
121 00:09:03.970 ⇒ 00:09:05.380 Miguel de Veyra: Yeah, yeah, cause this one is.
122 00:09:05.380 ⇒ 00:09:07.440 Uttam Kumaran: How does a switch logic work.
123 00:09:09.531 ⇒ 00:09:14.350 Miguel de Veyra: Basically, for example, there’s Atx, it will only go there.
124 00:09:15.850 ⇒ 00:09:17.780 Uttam Kumaran: But then, how does it know
125 00:09:18.280 ⇒ 00:09:26.740 Uttam Kumaran: like I guess I even take me one step back like, how do? What are all these like? What are all these zips like? How are they all different.
126 00:09:27.474 ⇒ 00:09:31.619 Miguel de Veyra: It’s this one. So for Austin Zips, basically their services.
127 00:09:34.040 ⇒ 00:09:36.000 Miguel de Veyra: and who the people in Austin?
128 00:09:38.000 ⇒ 00:09:42.679 Miguel de Veyra: So it’s there cause we have to pull by sheet. We can’t pull like the entire, you know.
129 00:09:44.210 ⇒ 00:09:45.470 Miguel de Veyra: Fantastic. So.
130 00:09:45.470 ⇒ 00:09:47.790 Uttam Kumaran: What is the Cf. What are? What are the other Zips?
131 00:09:47.980 ⇒ 00:09:51.390 Miguel de Veyra: This is Austin. This is not really location, this is scam free.
132 00:09:52.590 ⇒ 00:09:54.640 Miguel de Veyra: This is like another service that they will.
133 00:09:55.300 ⇒ 00:09:57.430 Uttam Kumaran: So, and then service.
134 00:09:58.310 ⇒ 00:10:02.139 Uttam Kumaran: So I guess, like, how do you know which sheet to pull from? That’s my question.
135 00:10:03.064 ⇒ 00:10:04.560 Miguel de Veyra: Yeah. So I gave.
136 00:10:05.440 ⇒ 00:10:08.560 Miguel de Veyra: So I gave the bot. If you see over here.
137 00:10:08.870 ⇒ 00:10:16.070 Miguel de Veyra: I gave the bot the what you call this, basically the zip codes. So it’s gonna match. So, for example.
138 00:10:16.310 ⇒ 00:10:19.719 Miguel de Veyra: you know, wait, what was the Zip code that I gave it again?
139 00:10:21.910 ⇒ 00:10:24.959 Miguel de Veyra: So for example, this one. If we go here.
140 00:10:26.550 ⇒ 00:10:28.710 Miguel de Veyra: and then we open the prompt.
141 00:10:30.230 ⇒ 00:10:35.019 Miguel de Veyra: There you go now. It’s in off now it knows in. It’s in Austin, because it’s here.
142 00:10:36.650 ⇒ 00:10:40.699 Miguel de Veyra: and that’s gonna return Atx, because that’s like the code of Austin.
143 00:10:42.350 ⇒ 00:10:43.180 Miguel de Veyra: And then that will.
144 00:10:43.180 ⇒ 00:10:43.979 Uttam Kumaran: Let me go!
145 00:10:43.980 ⇒ 00:10:46.410 Miguel de Veyra: Yes, and then that once, you know.
146 00:10:47.160 ⇒ 00:10:52.779 Uttam Kumaran: So you literally are like, based on the Zip code. You map it to one of the sheets. Can you open that one more time.
147 00:10:53.030 ⇒ 00:10:53.879 Miguel de Veyra: This one.
148 00:10:56.810 ⇒ 00:11:01.200 Uttam Kumaran: Okay, I see.
149 00:11:06.060 ⇒ 00:11:14.420 Uttam Kumaran: Okay, I think I have a lot of ideas on like, make this a little bit better on the data side. But it’s as long as it’s working for. Now I’m okay.
150 00:11:15.930 ⇒ 00:11:23.819 Uttam Kumaran: yeah. I guess my only question is, what are the response? Times looking and like? Have they gotten better at all?
151 00:11:24.160 ⇒ 00:11:24.760 Uttam Kumaran: One thing I
152 00:11:24.760 ⇒ 00:11:41.570 Uttam Kumaran: I guess 2 questions. There one is. I want to try to run like I want to almost run evals like every day. So we can like test. So we can kind of see as the team is making improvements like are things.
153 00:11:41.690 ⇒ 00:11:43.139 Miguel de Veyra: Getting better.
154 00:11:45.330 ⇒ 00:11:45.920 Miguel de Veyra: Yep.
155 00:11:46.863 ⇒ 00:11:50.039 Miguel de Veyra: Wait! Give me one moment. My my fridge is.
156 00:11:51.170 ⇒ 00:11:52.400 Uttam Kumaran: Your fridge is running.
157 00:11:53.800 ⇒ 00:11:54.480 Uttam Kumaran: Okay.
158 00:11:58.060 ⇒ 00:11:59.480 Miguel de Veyra: Okay, sorry about that.
159 00:12:01.450 ⇒ 00:12:03.210 Uttam Kumaran: Your fridge door was open. What happened?
160 00:12:03.420 ⇒ 00:12:08.379 Miguel de Veyra: Yeah, it was open. It was like ringing alarm bells. Yeah, I don’t know. What.
161 00:12:09.640 ⇒ 00:12:11.780 Miguel de Veyra: What do I? When I.
162 00:12:11.780 ⇒ 00:12:14.879 Uttam Kumaran: Should I try? Should I try vellum today? Cause I have another? I’m gonna
163 00:12:15.140 ⇒ 00:12:19.520 Uttam Kumaran: I’m gonna go spend an hour on ABC stuff. I’m gonna set up the Evals.
164 00:12:19.680 ⇒ 00:12:22.690 Uttam Kumaran: I could try to just see if I can run that on a cron.
165 00:12:23.920 ⇒ 00:12:24.979 Miguel de Veyra: Which one, this one.
166 00:12:25.730 ⇒ 00:12:28.800 Uttam Kumaran: The no. Just testing.
167 00:12:29.973 ⇒ 00:12:30.760 Miguel de Veyra: Okay, yeah.
168 00:12:31.430 ⇒ 00:12:35.220 Miguel de Veyra: I mean for the testing, we need the golden data sheet first.st Now.
169 00:12:35.690 ⇒ 00:12:36.300 Uttam Kumaran: Yeah.
170 00:12:36.500 ⇒ 00:12:37.690 Miguel de Veyra: Yeah. But I’ll just.
171 00:12:37.690 ⇒ 00:12:39.779 Uttam Kumaran: Create a couple of tests just to see that it works.
172 00:12:39.780 ⇒ 00:12:40.240 Miguel de Veyra: Okay.
173 00:12:42.180 ⇒ 00:12:43.320 Miguel de Veyra: And then.
174 00:12:43.940 ⇒ 00:12:44.620 Miguel de Veyra: Because I when.
175 00:12:44.620 ⇒ 00:12:55.259 Uttam Kumaran: The other. The other piece is like, Yeah, the other piece is like, is there? There’s no version control in an right like, how do we start to build a process where we like review the changes.
176 00:12:57.110 ⇒ 00:13:03.630 Miguel de Veyra: Wait. Sorry. I think it’s I think we can do it here, but it’s like it’s not really as strict as
177 00:13:03.760 ⇒ 00:13:07.660 Miguel de Veyra: Github, where you have the pr, it’s just like this.
178 00:13:10.440 ⇒ 00:13:11.560 Uttam Kumaran: Where is it? Oh.
179 00:13:11.760 ⇒ 00:13:14.540 Miguel de Veyra: Yeah, you know, it’s just basically every time you save.
180 00:13:15.810 ⇒ 00:13:17.146 Uttam Kumaran: But there’s no like.
181 00:13:19.990 ⇒ 00:13:24.959 Uttam Kumaran: Okay, I’ll I’ll look at it now I may. I may try to see like one.
182 00:13:25.810 ⇒ 00:13:29.630 Uttam Kumaran: Can we start to move towards like once we have all the nodes.
183 00:13:29.830 ⇒ 00:13:30.480 Miguel de Veyra: Yeah.
184 00:13:30.480 ⇒ 00:13:34.850 Uttam Kumaran: If we make prompt changes and stuff, I wanna be able to measure like the impact.
185 00:13:35.840 ⇒ 00:13:41.740 Uttam Kumaran: But I’ll think about it a bit more like. I know this is a long conversation. We’ve had a hell of time, so.
186 00:13:42.290 ⇒ 00:13:42.770 Miguel de Veyra: Yeah.
187 00:13:42.770 ⇒ 00:13:47.820 Uttam Kumaran: I’ll I’ll look into it more. Okay. But like, what do you? What’s your sense for the response? Times right now.
188 00:13:47.820 ⇒ 00:13:49.400 Miguel de Veyra: It’s insanely long.
189 00:13:50.210 ⇒ 00:13:51.520 Miguel de Veyra: How long?
190 00:13:51.520 ⇒ 00:13:53.850 Miguel de Veyra: Like some of it are like 38 seconds.
191 00:13:54.710 ⇒ 00:13:55.330 Uttam Kumaran: Okay.
192 00:13:55.490 ⇒ 00:13:59.930 Miguel de Veyra: But the thing with here, that’s what this that’s why it’s so long is basically
193 00:14:00.260 ⇒ 00:14:05.060 Miguel de Veyra: I prioritized accuracy, accuracy at all costs. So, for example.
194 00:14:05.470 ⇒ 00:14:19.811 Miguel de Veyra: So, for example, here. Wait. Sorry I’m not. I’m not sure what the question is. What services do we have for mosquitoes? So, for example, he, he, the agent, gets an ambiguous, you know, response from this one. It’s gonna try the other tool.
195 00:14:21.790 ⇒ 00:14:25.069 Miguel de Veyra: So now you know, it runs. Cause this one is.
196 00:14:25.230 ⇒ 00:14:33.390 Miguel de Veyra: I think it’s this one. So this tool, like takes 26 seconds itself, and then the other one will probably take a long one, too. So that’s why, you know, it’s 30,
197 00:14:33.390 ⇒ 00:14:34.480 Miguel de Veyra: one second.
198 00:14:35.402 ⇒ 00:14:37.570 Uttam Kumaran: 26 seconds.
199 00:14:38.330 ⇒ 00:14:39.890 Miguel de Veyra: Sorry you’re breaking up a bit.
200 00:14:41.920 ⇒ 00:14:46.269 Uttam Kumaran: But like, okay, so can I, is there? Is there? Is there an easy place to look at, the tracing.
201 00:14:49.640 ⇒ 00:14:51.590 Miguel de Veyra: Like, what’s happening? Why is it taking so long.
202 00:14:54.750 ⇒ 00:14:55.816 Uttam Kumaran: Hey? Can you hear me?
203 00:14:57.130 ⇒ 00:15:02.720 Miguel de Veyra: Yeah, here, the the one with the most stuff, basically.
204 00:15:05.400 ⇒ 00:15:09.270 Uttam Kumaran: But like, can I? Is there? Is there a ranking like, can I look at like?
205 00:15:09.590 ⇒ 00:15:12.049 Uttam Kumaran: Okay for this? For this run?
206 00:15:12.370 ⇒ 00:15:16.200 Uttam Kumaran: All of these ones are taking the longest.
207 00:15:17.640 ⇒ 00:15:20.030 Miguel de Veyra: I don’t think there’s a way to easily see that.
208 00:15:20.860 ⇒ 00:15:21.480 Uttam Kumaran: Okay.
209 00:15:23.220 ⇒ 00:15:28.180 Miguel de Veyra: But yeah, that’s why I wanted to work on something like this, because the context is, you know.
210 00:15:28.620 ⇒ 00:15:29.829 Miguel de Veyra: getting too long.
211 00:15:31.480 ⇒ 00:15:35.150 Uttam Kumaran: Okay, so, but even with that change it didn’t help much.
212 00:15:35.450 ⇒ 00:15:39.380 Miguel de Veyra: No, no, this, we haven’t really implemented this. We start. I started this today.
213 00:15:39.770 ⇒ 00:15:44.500 Uttam Kumaran: Okay, I think that’s 1 thing. I think the other thing is like.
214 00:15:45.310 ⇒ 00:15:49.810 Uttam Kumaran: we may just have to chunk the Bible up into logical sections.
215 00:15:50.940 ⇒ 00:16:09.029 Uttam Kumaran: because if we chunk the Bible up, and then each section has like metadata, then it’s basically going and retrieving which one to pick right versus, because the chunking strategy needs. We now have a good, we can chunk based on sections and like types of questions.
216 00:16:09.380 ⇒ 00:16:13.800 Uttam Kumaran: So I feel like we can almost get a lot better at that, too.
217 00:16:14.120 ⇒ 00:16:20.830 Uttam Kumaran: Implement yours. And then we can keep trying new methods.
218 00:16:21.650 ⇒ 00:16:27.250 Miguel de Veyra: Yeah, I mean, eventually, what we can do is here. I’ve talked to Casey about this
219 00:16:27.580 ⇒ 00:16:34.699 Miguel de Veyra: is eventually cause. Right now, it’s general notes, which is basically we where we dumped all their notes
220 00:16:35.190 ⇒ 00:16:39.330 Miguel de Veyra: right? Ideally, we’re gonna have one for billing one for this one for that.
221 00:16:40.000 ⇒ 00:16:43.440 Miguel de Veyra: Just so, you know it doesn’t have to scan through everything.
222 00:16:47.390 ⇒ 00:16:48.960 Uttam Kumaran: Hmm, okay.
223 00:16:56.560 ⇒ 00:17:00.099 Uttam Kumaran: okay, I think one of the things that we’ll I’ll start to break out
224 00:17:00.300 ⇒ 00:17:07.159 Uttam Kumaran: and I’m gonna I’ll I’ll see if I can help clean up the mirror a little bit is like, how do we get from 30 seconds to to
225 00:17:08.200 ⇒ 00:17:16.889 Uttam Kumaran: 10 seconds like, what are the key pieces that we need to hit. That’ll give us a lot more clarity of like what actually, we should go spend our time on.
226 00:17:17.150 ⇒ 00:17:17.740 Miguel de Veyra: Yeah.
227 00:17:19.270 ⇒ 00:17:19.839 Uttam Kumaran: Okay.
228 00:17:24.540 ⇒ 00:17:31.020 Miguel de Veyra: Yeah. But yeah, I think a good start would be like to see some runs like, you know, these
229 00:17:31.130 ⇒ 00:17:36.729 Miguel de Veyra: cause this alone. The entire run took 45 seconds, but one tool took 26.
230 00:17:38.250 ⇒ 00:17:41.070 Uttam Kumaran: And which tool was that like to pull from the Bible.
231 00:17:42.698 ⇒ 00:17:44.440 Miguel de Veyra: The one for the sheets.
232 00:17:46.290 ⇒ 00:17:46.690 Uttam Kumaran: Oh!
233 00:17:46.690 ⇒ 00:17:47.360 Miguel de Veyra: This one?
234 00:17:47.860 ⇒ 00:17:49.284 Miguel de Veyra: Yeah. So, hey?
235 00:17:50.340 ⇒ 00:17:55.120 Miguel de Veyra: So this one took 26 seconds, and then the other one, I would assume, took the rest of it.
236 00:17:55.120 ⇒ 00:18:02.499 Uttam Kumaran: Can you send me a screenshot of that? And then you put in the ABC slack because I can go help to see if we can reformat the
237 00:18:02.850 ⇒ 00:18:05.909 Uttam Kumaran: the the structure of that, anyway, you know.
238 00:18:06.500 ⇒ 00:18:11.789 Miguel de Veyra: But the other thing, by the way, is cause for here. Wait, let me save this
239 00:18:13.740 ⇒ 00:18:19.200 Miguel de Veyra: cause. We’re pulling from a central dock right? Abc, welcome
240 00:18:21.500 ⇒ 00:18:28.340 Miguel de Veyra: cause we’re pulling from here. So it’s centralized. We’re just getting context, basically everything.
241 00:18:28.810 ⇒ 00:18:31.540 Miguel de Veyra: So we can really chop it up as we want to
242 00:18:31.680 ⇒ 00:18:33.850 Miguel de Veyra: cause. The idea is, we have a central dock.
243 00:18:35.870 ⇒ 00:18:38.079 Uttam Kumaran: No, no, but that’s what I’m saying. Like
244 00:18:38.470 ⇒ 00:18:43.509 Uttam Kumaran: we may move away from the central dock, if that’s what’s taking the time.
245 00:18:43.740 ⇒ 00:18:47.059 Miguel de Veyra: Yeah, yeah, like, I want to know, is it the sheets.
246 00:18:47.190 ⇒ 00:18:52.046 Uttam Kumaran: Like? Is it the retrieval? Is it? What part of it is taking the most time.
247 00:18:52.350 ⇒ 00:18:53.000 Miguel de Veyra: Context.
248 00:18:53.000 ⇒ 00:19:13.219 Uttam Kumaran: The Central Bible. The the point of that was like we need some way for them to update. The second thing is, I want some way. Like I, I basically like the idea was like, let’s load into context. But similarly, we may not need all that in context. So I just want to bring in what’s relevant. This is where, like
249 00:19:13.790 ⇒ 00:19:19.450 Uttam Kumaran: up, you’ll probably have to spend a week looking at how we do rag, basically like
250 00:19:19.570 ⇒ 00:19:28.010 Uttam Kumaran: how we should chunk this into the base effectively, to improve our ability to retrieve and summarize faster, because.
251 00:19:28.010 ⇒ 00:19:30.429 Miguel de Veyra: Moving back to rock? Or are we sticking to context.
252 00:19:31.750 ⇒ 00:19:35.880 Uttam Kumaran: I don’t. I? I want us to the goal. I want us to get down to 15 seconds.
253 00:19:36.380 ⇒ 00:19:37.400 Uttam Kumaran: So
254 00:19:37.670 ⇒ 00:19:42.620 Uttam Kumaran: I want, I want to know? Like, yeah, I mean, it’s up. I think it’s up to our team to figure that out.
255 00:19:42.760 ⇒ 00:19:44.530 Miguel de Veyra: Okay, okay, yeah. I guess we’ll.
256 00:19:44.530 ⇒ 00:19:46.320 Uttam Kumaran: Was get the 1st thing was, get it to work.
257 00:19:46.820 ⇒ 00:19:47.520 Miguel de Veyra: Yep. Yep.
258 00:19:47.520 ⇒ 00:19:50.450 Uttam Kumaran: Second piece is, get it to be perfect.
259 00:19:50.730 ⇒ 00:19:56.269 Miguel de Veyra: Yeah, yeah, I guess what we can do is, you know, we’ll create 2 versions, one with drag, one with this.
260 00:19:56.800 ⇒ 00:20:02.409 Miguel de Veyra: with context. And then let’s see, like cause. I think accuracy is still more important than speed. At the end of the day.
261 00:20:02.410 ⇒ 00:20:09.060 Uttam Kumaran: Yeah, no. I want to get the accuracy down. And I want to get the testing down. We also need to help them do the update agent.
262 00:20:09.820 ⇒ 00:20:13.829 Miguel de Veyra: Oh, yeah, yeah, yeah, are they gonna extend after this 1st month? Then.
263 00:20:14.910 ⇒ 00:20:22.999 Uttam Kumaran: Well, so like, let’s talk about that, too. So one is, yeah, I guess 1st thing is like, what do you think about the update agent like, do you think that’s like.
264 00:20:23.570 ⇒ 00:20:28.021 Uttam Kumaran: do you think that’s possible? Do you think it’s gonna be like way harder.
265 00:20:30.510 ⇒ 00:20:36.080 Miguel de Veyra: It’s it’s basically a documentation agent. So it would be a bit harder, especially.
266 00:20:36.940 ⇒ 00:20:40.680 Miguel de Veyra: I don’t even think it’s gonna be that helpful if it’s in Google, chat.
267 00:20:41.750 ⇒ 00:20:42.470 Uttam Kumaran: Why?
268 00:20:45.640 ⇒ 00:20:50.729 Miguel de Veyra: But it’s mainly like we wanted to go. We want to go. They’re like, okay, actually, I want to change this policy.
269 00:20:51.800 ⇒ 00:20:57.200 Uttam Kumaran: It’s been, basically get, ask them questions and then go make the update in Google docs.
270 00:20:58.040 ⇒ 00:21:00.970 Miguel de Veyra: Is the bot gonna be the one making the update? Or is it gonna be.
271 00:21:00.970 ⇒ 00:21:01.560 Uttam Kumaran: Yeah.
272 00:21:03.950 ⇒ 00:21:08.699 Miguel de Veyra: Yeah, I have to look into how it it can update
273 00:21:12.430 ⇒ 00:21:13.670 Miguel de Veyra: me. The document.
274 00:21:14.840 ⇒ 00:21:18.230 Uttam Kumaran: I’m pretty sure you can pull. You can write back to Google.
275 00:21:18.230 ⇒ 00:21:20.120 Miguel de Veyra: Yeah, yeah, there’s find another place.
276 00:21:21.270 ⇒ 00:21:24.449 Uttam Kumaran: No, but it’s not even find and replace, like I think there’s probably.
277 00:21:25.870 ⇒ 00:21:26.770 Miguel de Veyra: Insert.
278 00:21:27.420 ⇒ 00:21:29.090 Uttam Kumaran: Yeah, insert or.
279 00:21:29.300 ⇒ 00:21:30.329 Miguel de Veyra: Yeah, I mean, there’s.
280 00:21:30.330 ⇒ 00:21:31.120 Uttam Kumaran: Something
281 00:21:36.210 ⇒ 00:21:42.210 Miguel de Veyra: There’s probably a way to do it. But yeah, let’s let’s dive into it.
282 00:21:42.720 ⇒ 00:21:45.410 Uttam Kumaran: The second piece, yeah, in terms of renewal. So
283 00:21:45.830 ⇒ 00:21:54.799 Uttam Kumaran: we basically want to talk about like what we’re gonna charge them ideally, we’re charging them, based on
284 00:21:54.970 ⇒ 00:22:00.319 Uttam Kumaran: interactions with the agent and then based on like
285 00:22:04.770 ⇒ 00:22:07.545 Uttam Kumaran: their if their customers are.
286 00:22:09.910 ⇒ 00:22:14.160 Uttam Kumaran: if their customers are basically like closing, there.
287 00:22:15.080 ⇒ 00:22:16.370 Miguel de Veyra: Yeah, if it improves, basically.
288 00:22:16.370 ⇒ 00:22:18.179 Uttam Kumaran: So one of the things I want to.
289 00:22:18.880 ⇒ 00:22:29.939 Uttam Kumaran: Yeah. So one of the things is like, when we, when do you know what data we get back from the Google chat like in terms of bugs. Because I actually, I want to see who’s chatting with it right?
290 00:22:30.719 ⇒ 00:22:31.069 Miguel de Veyra: Yeah.
291 00:22:31.070 ⇒ 00:22:35.509 Uttam Kumaran: Right now. The only thing we’re putting in Snowflake. But right now the only thing we’re putting in snow
292 00:22:35.640 ⇒ 00:22:39.370 Uttam Kumaran: like just from any end. Right, we’re getting anything from the Google Api.
293 00:22:40.443 ⇒ 00:22:42.640 Miguel de Veyra: We’re getting the name of the person.
294 00:22:43.600 ⇒ 00:22:45.440 Uttam Kumaran: But I wanna get the I wanna get.
295 00:22:46.610 ⇒ 00:22:48.860 Uttam Kumaran: I want to get the name email.
296 00:22:49.020 ⇒ 00:22:50.370 Uttam Kumaran: Timestamp.
297 00:22:53.700 ⇒ 00:22:56.500 Uttam Kumaran: Yeah, for for every chat session.
298 00:22:56.500 ⇒ 00:23:01.529 Miguel de Veyra: Okay, name, name and email. Yeah, that should be possible because we’re we’re already getting the name.
299 00:23:03.950 ⇒ 00:23:05.110 Uttam Kumaran: Yeah, okay?
300 00:23:06.570 ⇒ 00:23:12.379 Uttam Kumaran: And then also, for from na, then how can we publish? So we’re publishing the execution time?
301 00:23:13.770 ⇒ 00:23:18.758 Uttam Kumaran: Can we add more logs about like the steps and things like.
302 00:23:20.100 ⇒ 00:23:24.610 Uttam Kumaran: or you think I should just like try to buy the N. 8 M. Logging thing to do that.
303 00:23:25.510 ⇒ 00:23:26.749 Miguel de Veyra: We should probably just.
304 00:23:26.750 ⇒ 00:23:28.860 Uttam Kumaran: Cause. I wanna, I want us. Okay, okay.
305 00:23:28.860 ⇒ 00:23:29.360 Miguel de Veyra: Just like.
306 00:23:29.360 ⇒ 00:23:30.170 Uttam Kumaran: Logs.
307 00:23:31.010 ⇒ 00:23:31.750 Uttam Kumaran: Edit.
308 00:23:32.150 ⇒ 00:23:33.540 Uttam Kumaran: Okay?
309 00:23:33.540 ⇒ 00:23:37.349 Miguel de Veyra: So wait. Sorry I need to note this. So we need to add email.
310 00:23:38.500 ⇒ 00:23:39.490 Uttam Kumaran: Yeah.
311 00:23:39.490 ⇒ 00:23:43.980 Miguel de Veyra: We already have execution execution time. We have. Ps. I believe we have.
312 00:23:45.150 ⇒ 00:23:47.410 Uttam Kumaran: We do have execution time.
313 00:23:47.720 ⇒ 00:23:51.629 Uttam Kumaran: We do have an event. Timestamp. Yeah, I wanna add A,
314 00:23:51.820 ⇒ 00:23:57.540 Uttam Kumaran: I want to add the username. And then if you can get their actual full name, too. That’d be great.
315 00:23:58.049 ⇒ 00:24:01.279 Miguel de Veyra: I mean, it’s here. We already have it, general.
316 00:24:01.280 ⇒ 00:24:01.890 Uttam Kumaran: Okay.
317 00:24:02.040 ⇒ 00:24:03.400 Miguel de Veyra: Miguel Guera.
318 00:24:04.520 ⇒ 00:24:06.359 Miguel de Veyra: Let me go.
319 00:24:06.360 ⇒ 00:24:07.210 Uttam Kumaran: From Google.
320 00:24:07.210 ⇒ 00:24:08.570 Miguel de Veyra: Yeah, yeah, this is from Google.
321 00:24:08.570 ⇒ 00:24:11.380 Uttam Kumaran: Oh, okay. 6. Alright. Then email would be great.
322 00:24:11.700 ⇒ 00:24:13.149 Miguel de Veyra: So only email, now, no.
323 00:24:13.950 ⇒ 00:24:15.300 Uttam Kumaran: Yeah, this email.
324 00:24:15.710 ⇒ 00:24:17.710 Miguel de Veyra: Okay, get from Google.
325 00:24:18.540 ⇒ 00:24:20.960 Uttam Kumaran: And I’ll ping Tim again right now.
326 00:24:20.960 ⇒ 00:24:23.719 Miguel de Veyra: Yeah, yeah, I don’t think you’re like doing
327 00:24:27.070 ⇒ 00:24:29.329 Miguel de Veyra: so. Yeah, I think that’s pretty much it. And then I guess.
328 00:24:29.732 ⇒ 00:24:33.357 Uttam Kumaran: Basically, what I want to try and do is
329 00:24:35.060 ⇒ 00:24:42.580 Uttam Kumaran: I’m gonna start to pull their chat logs and then try and figure out which which customer
330 00:24:42.840 ⇒ 00:24:48.409 Uttam Kumaran: which customers are calling, and and if we use the chat bot, and then, if there was a successful resolution.
331 00:24:48.530 ⇒ 00:24:50.110 Uttam Kumaran: and then we’ll get paid for that.
332 00:24:51.060 ⇒ 00:24:52.210 Miguel de Veyra: Okay, okay.
333 00:24:52.410 ⇒ 00:24:57.900 Miguel de Veyra: but we won’t really get the customer name. We’ll get the client, the the username of the chat. There.
334 00:24:57.900 ⇒ 00:25:02.069 Uttam Kumaran: But if I know, if I know if I know who the Csr. Was.
335 00:25:02.360 ⇒ 00:25:05.720 Uttam Kumaran: I from their phone system, I can pull who they’re talking to at the same time.
336 00:25:05.948 ⇒ 00:25:07.089 Miguel de Veyra: Yeah, yeah, that makes sense.
337 00:25:07.090 ⇒ 00:25:08.360 Uttam Kumaran: Given the timestamp.
338 00:25:09.590 ⇒ 00:25:10.610 Miguel de Veyra: Okay. Yes.
339 00:25:11.020 ⇒ 00:25:16.369 Uttam Kumaran: The only other way is like, I would be like, tell me which customer this is but like. If I can avoid that, I will.
340 00:25:17.720 ⇒ 00:25:18.480 Miguel de Veyra: Okay.
341 00:25:19.440 ⇒ 00:25:20.160 Miguel de Veyra: Yeah.
342 00:25:22.550 ⇒ 00:25:23.245 Uttam Kumaran: Okay.
343 00:25:24.840 ⇒ 00:25:28.740 Miguel de Veyra: And then for yeah, sorry, I think we got no, no, no.
344 00:25:28.890 ⇒ 00:25:31.869 Miguel de Veyra: yeah, I think the the original discussion was.
345 00:25:31.870 ⇒ 00:25:32.529 Uttam Kumaran: Yeah, yeah.
346 00:25:33.110 ⇒ 00:25:35.479 Miguel de Veyra: The plan for the extension. I guess.
347 00:25:37.420 ⇒ 00:25:44.990 Uttam Kumaran: Well, yeah, I mean, but there’s so our plan from the from the whole time was to build the 2 agents and charge them on a performance basis.
348 00:25:45.120 ⇒ 00:25:46.080 Miguel de Veyra: Okay. Okay.
349 00:25:46.550 ⇒ 00:25:51.459 Uttam Kumaran: So basically, we are. So there’s gonna be 2 things, one, if they want us to build agents for the other.
350 00:25:51.630 ⇒ 00:25:55.930 Uttam Kumaran: like for the other service businesses that we’re gonna charge a per agent fee.
351 00:25:57.190 ⇒ 00:26:01.870 Uttam Kumaran: But then I want to charge an ongoing price based on how it affects their business.
352 00:26:04.980 ⇒ 00:26:07.560 Miguel de Veyra: Results based. You don’t wanna use.
353 00:26:08.302 ⇒ 00:26:09.867 Uttam Kumaran: What do you call it?
354 00:26:10.430 ⇒ 00:26:13.070 Miguel de Veyra: Like usage rate instead of success, rate.
355 00:26:15.370 ⇒ 00:26:19.789 Uttam Kumaran: I we it’s a good fallback, but I wanna I wanna rather do it on success.
356 00:26:20.210 ⇒ 00:26:22.439 Miguel de Veyra: Okay, okay, cause, yeah.
357 00:26:24.560 ⇒ 00:26:25.479 Uttam Kumaran: But we’ll see
358 00:26:25.740 ⇒ 00:26:29.720 Uttam Kumaran: see what we can get. If I can get that data. Then I’m gonna email you that for
359 00:26:30.110 ⇒ 00:26:33.380 Uttam Kumaran: like, somehow, we can get access to that platform.
360 00:26:33.380 ⇒ 00:26:38.150 Miguel de Veyra: Yeah, just yeah. Getting them to do anything is very hard.
361 00:26:38.150 ⇒ 00:26:39.110 Uttam Kumaran: Yeah.
362 00:26:39.110 ⇒ 00:26:39.770 Miguel de Veyra: It’s very.
363 00:26:40.360 ⇒ 00:26:42.620 Uttam Kumaran: I’m gonna send the yeah, I’m gonna
364 00:26:43.700 ⇒ 00:26:45.940 Uttam Kumaran: it’s a difference between being like
365 00:26:46.580 ⇒ 00:26:51.180 Uttam Kumaran: like, I’ll this is what I’m gonna tell the team, too, and slack that there’s a difference between being nice
366 00:26:51.330 ⇒ 00:26:52.680 Uttam Kumaran: and then being like
367 00:26:53.520 ⇒ 00:27:03.160 Uttam Kumaran: sort of kind. It’s like some people are nice, and they’re just like we don’t want to step on their toes. Dude. If they don’t give that to us, we can’t succeed. So we just had to push and push
368 00:27:03.350 ⇒ 00:27:05.170 Uttam Kumaran: and make sure we get what we need, you know.
369 00:27:05.660 ⇒ 00:27:06.220 Miguel de Veyra: Yep.
370 00:27:07.050 ⇒ 00:27:09.854 Uttam Kumaran: Okay, alright. So let me work on a couple of things, and then
371 00:27:10.210 ⇒ 00:27:14.270 Uttam Kumaran: As soon as Slack gets back on I’ll I’ll email you with stuff.
372 00:27:14.270 ⇒ 00:27:16.960 Miguel de Veyra: And then I saw sorry last. Do you have to go.
373 00:27:17.510 ⇒ 00:27:18.102 Uttam Kumaran: Not at all.
374 00:27:18.490 ⇒ 00:27:26.699 Miguel de Veyra: The last thing I think I mentioned I saw in one of your talks with Yvette was about the voice agent. Is that something you want to offer them soon or.
375 00:27:26.700 ⇒ 00:27:27.290 Uttam Kumaran: The
376 00:27:28.200 ⇒ 00:27:30.875 Uttam Kumaran: He wants to see a demo of it.
377 00:27:32.370 ⇒ 00:27:33.819 Miguel de Veyra: Who? Yvette? Okay.
378 00:27:42.833 ⇒ 00:27:43.769 Miguel de Veyra: utop. You’re breaking up.
379 00:27:43.770 ⇒ 00:27:46.320 Uttam Kumaran: How easy is it for you to put there it is.
380 00:27:47.150 ⇒ 00:27:48.640 Uttam Kumaran: Oh, sorry! Hold on one second
381 00:27:55.710 ⇒ 00:27:56.520 Uttam Kumaran: about.
382 00:27:57.080 ⇒ 00:27:58.799 Miguel de Veyra: Yeah. Yeah, a bit better.
383 00:28:04.810 ⇒ 00:28:08.389 Uttam Kumaran: Okay, no, no. What I was saying is like, how hard would this be to put behind Vappy.
384 00:28:12.310 ⇒ 00:28:16.390 Miguel de Veyra: Like, can you just connect that? Can you just connect our normal twilio number to this.
385 00:28:16.680 ⇒ 00:28:19.609 Uttam Kumaran: And then we could just be like, Go play around with it.
386 00:28:19.610 ⇒ 00:28:21.390 Miguel de Veyra: Yeah, yeah, definitely.
387 00:28:21.630 ⇒ 00:28:24.250 Miguel de Veyra: But can I not do that this week? Probably next week.
388 00:28:24.650 ⇒ 00:28:26.199 Uttam Kumaran: Yeah. Yeah. Next week is fine. Yeah.
389 00:28:26.200 ⇒ 00:28:34.879 Miguel de Veyra: Yeah. Yeah. The only thing I’m concerned about with that is like the the response time. Of course, right? Like, this is taking 30 seconds.
390 00:28:34.880 ⇒ 00:28:38.539 Uttam Kumaran: That’s fine. I mean again, as we improve the system that’ll get better. But like
391 00:28:38.680 ⇒ 00:28:42.349 Uttam Kumaran: she wanted to play around with that, we can also just send her a different demo.
392 00:28:43.080 ⇒ 00:28:46.320 Miguel de Veyra: Yeah, like, maybe, like, specific for San Antonio. Only right?
393 00:28:47.240 ⇒ 00:28:51.800 Uttam Kumaran: Yeah, I just want to give her something to call. She’s always wanted to play around with that.
394 00:28:51.800 ⇒ 00:28:53.050 Miguel de Veyra: Okay. Sure. Sure.
395 00:28:55.820 ⇒ 00:29:00.252 Uttam Kumaran: Okay, alright. So I’ll get notes on this. And then, yeah, just slack me with anything.
396 00:29:01.050 ⇒ 00:29:08.600 Uttam Kumaran: and good luck I’ll I’ll try to be on the meeting later. But I’m not sure. Okay, okay.
397 00:29:08.780 ⇒ 00:29:09.840 Miguel de Veyra: Okay. Yeah.
398 00:29:10.010 ⇒ 00:29:12.130 Uttam Kumaran: I don’t know. How does stuff going? Do you talk to Rico?
399 00:29:13.260 ⇒ 00:29:25.790 Miguel de Veyra: Yeah, he’s not really. Yeah. He’s not not right now, I guess, because they have. He doesn’t want any change in his life, I guess because someone died in their family. So yeah, I didn’t really push.
400 00:29:25.790 ⇒ 00:29:26.380 Uttam Kumaran: Okay.
401 00:29:26.380 ⇒ 00:29:31.380 Miguel de Veyra: I’m talking to a friend before he he works with Janna right now.
402 00:29:32.370 ⇒ 00:29:37.720 Miguel de Veyra: cause I think he’s very keen to leave, too, and he’s pretty weird.
403 00:29:37.720 ⇒ 00:29:41.379 Uttam Kumaran: I don’t. Why isn’t Jana working with us full time? I’ve tried tried to ask her.
404 00:29:42.390 ⇒ 00:29:43.859 Miguel de Veyra: Did you ask her? What did she say.
405 00:29:43.860 ⇒ 00:29:46.770 Uttam Kumaran: Yeah, I did. I think she said, like she was just like
406 00:29:47.630 ⇒ 00:29:49.280 Uttam Kumaran: it was a hundred percent sure.
407 00:29:50.254 ⇒ 00:29:54.900 Miguel de Veyra: I think the concern she she told me it because I asked her about it is the working hours.
408 00:29:57.050 ⇒ 00:29:58.859 Uttam Kumaran: Yeah. But dude, I don’t give it like.
409 00:29:59.090 ⇒ 00:30:01.370 Uttam Kumaran: okay, I mean, I guess that’s fair. But then
410 00:30:01.720 ⇒ 00:30:11.990 Uttam Kumaran: you should tell her that I’m that we’re gonna we’re gonna move away from part time people, I mean, I can tell her that, too. But like I’m on the data side dude, we’re we’re we’re sort of pushing all the part time people out.
411 00:30:12.260 ⇒ 00:30:13.480 Miguel de Veyra: Oh, okay. Okay.
412 00:30:14.160 ⇒ 00:30:17.959 Uttam Kumaran: Because it’s like, I don’t want part like the part time people stuff like.
413 00:30:17.960 ⇒ 00:30:18.650 Miguel de Veyra: Yeah, it’s.
414 00:30:18.650 ⇒ 00:30:20.370 Uttam Kumaran: A lot of accommodating for us.
415 00:30:20.370 ⇒ 00:30:21.460 Miguel de Veyra: Yeah, I don’t. Wanna. I don’t want.
416 00:30:21.460 ⇒ 00:30:24.589 Uttam Kumaran: People to be dealing with like hours, and like shit like that.
417 00:30:24.590 ⇒ 00:30:30.669 Miguel de Veyra: Yeah, actually, it’s part of why I’m a bit like not really struggling, but it’s a bit complicated, because I’m you know.
418 00:30:32.090 ⇒ 00:30:35.410 Uttam Kumaran: Oh, yeah, you’re balancing her during the day, and you and Casey work at night.
419 00:30:35.410 ⇒ 00:30:36.290 Miguel de Veyra: Yeah.
420 00:30:36.650 ⇒ 00:30:38.500 Miguel de Veyra: So like, I’m working around the clock.
421 00:30:39.240 ⇒ 00:30:42.419 Uttam Kumaran: I know. So I that’s why I wanna I mean.
422 00:30:42.420 ⇒ 00:30:43.590 Miguel de Veyra: But yeah, I’ll talk to you tomorrow.
423 00:30:44.060 ⇒ 00:30:48.939 Uttam Kumaran: There’s 2 things. One is like, Yeah, tell her that like, Hey, the company is making a shift towards
424 00:30:49.260 ⇒ 00:30:53.420 Uttam Kumaran: full time people. But like dude we are! We are a Us. Company like.
425 00:30:53.910 ⇒ 00:31:05.329 Miguel de Veyra: Yeah. And then I told her, like the meetings. I I told her the meetings, you know. It’s we’ve already talked to them to move it like an hour earlier. So it’s not like one am.
426 00:31:05.330 ⇒ 00:31:11.079 Uttam Kumaran: Yeah, it’s it’s also like, we don’t have any ait meetings like she’s not in any of the client meetings.
427 00:31:11.080 ⇒ 00:31:11.830 Miguel de Veyra: Yeah, yeah, I don’t know.
428 00:31:11.830 ⇒ 00:31:15.550 Uttam Kumaran: Like you guys can meet during the you guys can meet like sometime in the evening.
429 00:31:15.660 ⇒ 00:31:22.519 Uttam Kumaran: and then, like, you can talk to me and like night, I don’t care. But also I don’t want you to have people that are like, part time.
430 00:31:22.520 ⇒ 00:31:25.509 Uttam Kumaran: Yeah, yeah, kind of messing up your schedule. So
431 00:31:25.720 ⇒ 00:31:31.979 Uttam Kumaran: it’s fine, either way. The other thing I’m gonna probably kick off as soon as the month closes is we’re gonna bring on someone to help on recruiting.
432 00:31:32.290 ⇒ 00:31:37.040 Uttam Kumaran: So I’m gonna just make sure that the AI engineering resume is like, kind of finalized.
433 00:31:37.040 ⇒ 00:31:37.680 Miguel de Veyra: Oh yes!
434 00:31:37.680 ⇒ 00:31:44.750 Uttam Kumaran: He’ll go get us. No, that’s fine. Don’t don’t worry about that. I’ll fill it out and make sure that he can. Sky can go get us candidates like that.
435 00:31:44.910 ⇒ 00:31:47.049 Uttam Kumaran: I just wanna exhaust our network first.st
436 00:31:47.050 ⇒ 00:31:47.879 Miguel de Veyra: Just came in.
437 00:31:49.070 ⇒ 00:31:50.700 Uttam Kumaran: Yeah, I mean, we’re all
438 00:31:51.030 ⇒ 00:31:56.650 Uttam Kumaran: yeah, Michael Weinberg. I’ll have you connect with him. He’s really he’s like my mentor, and like a really good friend of mine.
439 00:31:56.920 ⇒ 00:31:58.899 Uttam Kumaran: extremely smart data person.
440 00:31:59.381 ⇒ 00:32:04.019 Miguel de Veyra: I’m bringing him on to sort of help with our data team and meet with our data folks and sort of coach people.
441 00:32:04.467 ⇒ 00:32:07.680 Miguel de Veyra: And then dem a lot is another data person like an engineer.
442 00:32:08.460 ⇒ 00:32:14.360 Uttam Kumaran: We’re basically we’re done hiring on the data side on maybe one more like, I may bring on one more data analyst. But.
443 00:32:14.360 ⇒ 00:32:17.679 Miguel de Veyra: Our data team like blew up. No like so much people. Now.
444 00:32:17.680 ⇒ 00:32:21.299 Uttam Kumaran: Well, dude look, that’s all. I was doing all their jobs by myself.
445 00:32:22.680 ⇒ 00:32:27.325 Uttam Kumaran: So yeah, that’s we must show it shows like how much we need it.
446 00:32:28.750 ⇒ 00:32:31.790 Miguel de Veyra: Okay, okay, is that ever gonna close? By the way.
447 00:32:32.950 ⇒ 00:32:37.230 Uttam Kumaran: But, say, I, I think our proposal is a little bit better. I told Scott to go ahead and send it so.
448 00:32:37.630 ⇒ 00:32:38.669 Miguel de Veyra: Oh, okay, okay.
449 00:32:39.570 ⇒ 00:32:43.149 Miguel de Veyra: Oh, by the way, a quick question, what happened to what’s his name?
450 00:32:43.810 ⇒ 00:32:44.470 Uttam Kumaran: Honor.
451 00:32:44.470 ⇒ 00:32:45.379 Miguel de Veyra: Don’t hear you.
452 00:32:45.880 ⇒ 00:32:52.480 Uttam Kumaran: Yeah, we initially had a we had a sort of 3 month goal for him to just try and sell for us. But it was wasn’t working out like
453 00:32:52.740 ⇒ 00:32:58.040 Uttam Kumaran: he wasn’t good sales, and I don’t think he was having. He was. He wasn’t sort of picking up
454 00:32:58.520 ⇒ 00:33:02.600 Uttam Kumaran: what we were selling really easily. And then we just decided to sort of split ways.
455 00:33:03.030 ⇒ 00:33:04.450 Miguel de Veyra: I see. Okay, okay.
456 00:33:04.920 ⇒ 00:33:07.080 Uttam Kumaran: It’s tough dude sales is hard, but like
457 00:33:07.340 ⇒ 00:33:09.539 Uttam Kumaran: I can’t afford to have people that.
458 00:33:09.730 ⇒ 00:33:12.249 Miguel de Veyra: Aren’t selling sales is like.
459 00:33:12.780 ⇒ 00:33:16.240 Uttam Kumaran: Extremely easy to understand whether you’re doing a good job or not. So
460 00:33:16.480 ⇒ 00:33:22.990 Uttam Kumaran: we got like a month and a half. And I was like, okay, it’s not working like we should. We should either do something or like this is probably not gonna work out.
461 00:33:23.550 ⇒ 00:33:24.510 Miguel de Veyra: Okay. Okay.
462 00:33:29.180 ⇒ 00:33:32.700 Miguel de Veyra: okay, yeah. I think that’s pretty much it. My slack is still down.
463 00:33:33.600 ⇒ 00:33:36.670 Miguel de Veyra: I’ll prepare for the meeting, for with the ABC people.
464 00:33:36.850 ⇒ 00:33:38.620 Miguel de Veyra: I’ll add you to invite them.
465 00:33:38.620 ⇒ 00:33:40.989 Uttam Kumaran: Yeah, if you can, you add me? I’ll I’ll try to hop on.
466 00:33:41.480 ⇒ 00:33:42.370 Miguel de Veyra: Okay. Okay.
467 00:33:42.940 ⇒ 00:33:45.489 Uttam Kumaran: Okay. And then just try to keep that meeting
468 00:33:45.880 ⇒ 00:33:50.940 Uttam Kumaran: specific. They like to. They’re from. They’re from here in Texas. So they’re like slow
469 00:33:51.391 ⇒ 00:33:55.190 Uttam Kumaran: sort of like methodical. Just push them to just get whatever shit we need.
470 00:33:55.760 ⇒ 00:33:56.920 Miguel de Veyra: Yep. Yep. Okay.
471 00:33:57.420 ⇒ 00:33:57.990 Uttam Kumaran: Okay.
472 00:33:58.930 ⇒ 00:34:00.729 Uttam Kumaran: Okay, alright, thanks to you. Good chat.
473 00:34:00.730 ⇒ 00:34:01.080 Miguel de Veyra: Thanks.
474 00:34:01.080 ⇒ 00:34:01.430 Uttam Kumaran: 36.
475 00:34:01.770 ⇒ 00:34:02.410 Miguel de Veyra: Bye-bye.
476 00:34:02.410 ⇒ 00:34:03.080 Uttam Kumaran: Bye, bye.