Meeting Title: All Engineering (AI Uses + Workshop) Date: 2025-08-28 Meeting participants: Samuel Roberts, Casie Aviles, Mustafa Raja, Demilade Agboola, Awaish Kumar, Annie Yu
WEBVTT
1 00:00:09.590 ⇒ 00:00:10.500 Samuel Roberts: P.
2 00:00:13.360 ⇒ 00:00:14.290 Casie Aviles: A sound.
3 00:00:15.720 ⇒ 00:00:19.919 Samuel Roberts: So, I’m gonna let everyone know, but I have another meeting I gotta get to in, like.
4 00:00:20.340 ⇒ 00:00:25.329 Samuel Roberts: 20 minutes, so this might not… I might not just call as long as it looks like, at least.
5 00:00:26.130 ⇒ 00:00:26.850 Casie Aviles: Oh.
6 00:00:31.750 ⇒ 00:00:33.900 Samuel Roberts: But, we’ll see. Is there an….
7 00:00:33.900 ⇒ 00:00:36.159 Casie Aviles: Is there an agenda for today? I wasn’t…
8 00:00:36.310 ⇒ 00:00:38.660 Casie Aviles: present last week, I believe, yeah.
9 00:00:38.660 ⇒ 00:00:43.249 Samuel Roberts: Oh, yeah, no, last week, last week, … U-Tam led, like, a…
10 00:00:43.680 ⇒ 00:00:46.449 Samuel Roberts: a Fig Jam session, and just, like.
11 00:00:47.040 ⇒ 00:00:50.999 Samuel Roberts: Areas that we think there are problems, and where solutions could be, and….
12 00:00:51.440 ⇒ 00:00:51.970 Casie Aviles: Okay.
13 00:00:53.260 ⇒ 00:00:58.030 Samuel Roberts: this week, I was just going to kind of leave a general discussion, because I’m… I’m looking to learn a little bit more about, like.
14 00:00:58.270 ⇒ 00:01:02.349 Samuel Roberts: what other people in the company do, because I know, like, what we do.
15 00:01:02.350 ⇒ 00:01:03.230 Casie Aviles: Yeah.
16 00:01:03.230 ⇒ 00:01:08.100 Samuel Roberts: I’m curious to learn a little bit more about what everyone else does, and maybe get a sense of, like, how they’re using AI and other…
17 00:01:08.420 ⇒ 00:01:09.450 Samuel Roberts: Other ways.
18 00:01:09.860 ⇒ 00:01:13.859 Samuel Roberts: … But like I said, I got added to a call that said.
19 00:01:14.610 ⇒ 00:01:17.079 Samuel Roberts: In, like, 20 minutes, and so I didn’t wanna…
20 00:01:17.230 ⇒ 00:01:19.749 Samuel Roberts: cancel this at the last second, but I figured we could at least….
21 00:01:19.750 ⇒ 00:01:20.090 Casie Aviles: Okay.
22 00:01:20.090 ⇒ 00:01:21.260 Samuel Roberts: But we’ll see.
23 00:01:21.790 ⇒ 00:01:22.840 Samuel Roberts: See how it goes.
24 00:01:31.790 ⇒ 00:01:32.770 Samuel Roberts: Big loss.
25 00:01:33.480 ⇒ 00:01:37.700 Samuel Roberts: Must joins. This might be a quick one today anyway, because… Sweet.
26 00:01:40.210 ⇒ 00:01:40.950 Mustafa Raja: Hey guys.
27 00:01:42.140 ⇒ 00:01:43.070 Samuel Roberts: Hey.
28 00:01:43.590 ⇒ 00:01:44.280 Mustafa Raja: How you doing?
29 00:01:47.160 ⇒ 00:01:47.870 Samuel Roberts: Okay.
30 00:01:50.900 ⇒ 00:01:53.539 Samuel Roberts: I was just telling Casey, but this might be, …
31 00:01:53.930 ⇒ 00:01:59.219 Samuel Roberts: quick call, just a little discussion, maybe, with some of the other engineering folks, because I’m…
32 00:02:00.190 ⇒ 00:02:01.790 Samuel Roberts: On that call it.
33 00:02:02.840 ⇒ 00:02:04.330 Samuel Roberts: In, like, 20 minutes.
34 00:02:05.550 ⇒ 00:02:07.010 Mustafa Raja: Oh, yeah, the default one.
35 00:02:07.010 ⇒ 00:02:13.069 Samuel Roberts: Yeah, and so I didn’t realize those were the same time and everything, till a little while ago, but…
36 00:02:14.070 ⇒ 00:02:16.879 Samuel Roberts: I figured I didn’t really wanna…
37 00:02:17.110 ⇒ 00:02:19.370 Samuel Roberts: lead too much of anything to that. I was just looking, actually.
38 00:02:19.500 ⇒ 00:02:23.830 Samuel Roberts: hopefully chat with other folks, but no one’s here yet, so maybe not today.
39 00:02:24.660 ⇒ 00:02:26.749 Mustafa Raja: Yeah, I guess we reach out.
40 00:02:26.750 ⇒ 00:02:29.790 Samuel Roberts: Yeah, exactly. I was just looking… I’m curious about how the…
41 00:02:29.960 ⇒ 00:02:32.449 Samuel Roberts: Rest of the team, most of the engineering guys are…
42 00:02:32.810 ⇒ 00:02:37.190 Samuel Roberts: using AI and what their day-to-day look like, and so… there we go.
43 00:02:38.120 ⇒ 00:02:43.130 Samuel Roberts: This might be a quick one in general, but… We’ll see.
44 00:02:47.410 ⇒ 00:02:48.360 Demilade Agboola: Hi, everyone.
45 00:02:48.770 ⇒ 00:02:49.520 Samuel Roberts: Hello?
46 00:02:50.090 ⇒ 00:02:50.740 Casie Aviles: A….
47 00:02:56.180 ⇒ 00:02:59.540 Samuel Roberts: We have everyone else a few more minutes, or a few more moments?
48 00:03:00.870 ⇒ 00:03:01.680 Samuel Roberts: Perfect.
49 00:03:06.150 ⇒ 00:03:09.559 Casie Aviles: spoken for. And that’s the transformation.
50 00:03:09.560 ⇒ 00:03:10.680 Samuel Roberts: Hello, everyone.
51 00:03:13.810 ⇒ 00:03:14.990 Samuel Roberts: How’s everyone going today?
52 00:03:14.990 ⇒ 00:03:15.540 Mustafa Raja: Hey.
53 00:03:20.720 ⇒ 00:03:22.259 Demilade Agboola: Pretty good, can’t complain.
54 00:03:22.500 ⇒ 00:03:23.570 Samuel Roberts: Alright, cool.
55 00:03:24.810 ⇒ 00:03:26.929 Samuel Roberts: Royce, Shanny, you guys doing alright?
56 00:03:32.470 ⇒ 00:03:33.880 Annie Yu: Yeah, not bad.
57 00:03:34.040 ⇒ 00:03:35.340 Samuel Roberts: Alright, cool.
58 00:03:35.490 ⇒ 00:03:44.690 Samuel Roberts: Yeah, doing alright. I was just letting people know, as people are joining, that I got added to a meeting in about 20 minutes, so this is probably just going to be a kind of
59 00:03:44.880 ⇒ 00:03:51.189 Samuel Roberts: Quick little… discussion, I guess, more than anything else, because I didn’t want to jump into anything
60 00:03:51.360 ⇒ 00:03:54.970 Samuel Roberts: that I’ve… Excuse me, have to jump right out of, but, …
61 00:03:56.640 ⇒ 00:04:00.399 Samuel Roberts: after last week, I thought it might be good to just have, like, a
62 00:04:00.540 ⇒ 00:04:10.909 Samuel Roberts: I’m curious, as someone relatively new here and more on the AI team side, about what it’s like on the other side of engineering, you know, how
63 00:04:11.280 ⇒ 00:04:16.389 Samuel Roberts: how you guys are currently using AI, what you guys think could be, you know… I… and this is…
64 00:04:16.529 ⇒ 00:04:26.809 Samuel Roberts: you know, without Utam, because I… we were talking, and he was like, I feel like I fill up the air just by talking, and I was like, well, let’s… let’s maybe just have one without the CEO, and let, …
65 00:04:27.680 ⇒ 00:04:32.780 Samuel Roberts: Let us just have a chat as engineering team. … So…
66 00:04:32.980 ⇒ 00:04:35.240 Samuel Roberts: That’s sort of my thought for the next…
67 00:04:35.550 ⇒ 00:04:39.989 Samuel Roberts: 20ish minutes, maybe. We can just have a little chat, …
68 00:04:41.890 ⇒ 00:04:57.640 Samuel Roberts: And I’m looking to learn a little bit about what people do on a day-to-day basis, what, you know, I have a better sense on the AI side, but I’m looking to just see, how things are with where, you know, we can chat a little bit about AI in particular, and where we could
69 00:04:57.750 ⇒ 00:05:02.959 Samuel Roberts: add things, but I… I don’t have a great sense of, like, you know, what you guys, not on the AI
70 00:05:03.110 ⇒ 00:05:08.949 Samuel Roberts: team do on a daily basis yet. I’d love to learn a little bit more, if you guys are willing to…
71 00:05:09.310 ⇒ 00:05:12.689 Samuel Roberts: The chat, … So if someone wants to…
72 00:05:12.980 ⇒ 00:05:19.349 Samuel Roberts: jump in. I’m just looking to… yeah, kind of, like, create a little bit of space where we just all get to
73 00:05:19.680 ⇒ 00:05:26.039 Samuel Roberts: Shadow’s engineering team, you know, even though we’re kind of on different… different sides of it. So, …
74 00:05:26.810 ⇒ 00:05:30.020 Samuel Roberts: Awes, Demolade, Annie, either any of you wanna…
75 00:05:30.530 ⇒ 00:05:36.000 Samuel Roberts: You know, just kind of throw your hat in and… and, kind of….
76 00:05:37.250 ⇒ 00:05:43.610 Awaish Kumar: Yeah, I would encourage, … team members, like Jamala Rayani, to…
77 00:05:44.890 ⇒ 00:05:49.030 Awaish Kumar: To just share, like, what they have been doing, In the week.
78 00:05:49.530 ⇒ 00:05:51.100 Samuel Roberts: Yeah, totally, that’s a good way to do it.
79 00:05:51.590 ⇒ 00:05:54.819 Samuel Roberts: But yeah, what has your week been like? You know, guys? I mean, I’m happy to…
80 00:05:54.950 ⇒ 00:06:01.089 Samuel Roberts: My week’s been a little different because I’ve been… traveling, but I’m happy to… Have a conversation.
81 00:06:01.780 ⇒ 00:06:10.909 Demilade Agboola: I mean, for the most part, it’s just a lot of modeling and debugging and fixing issues with, like, data quality.
82 00:06:11.150 ⇒ 00:06:15.470 Demilade Agboola: Or, you know, whatever issues people have with their data.
83 00:06:15.830 ⇒ 00:06:22.390 Demilade Agboola: … I think with AI, basically, it’s more of, like, a lot of cursor.
84 00:06:22.480 ⇒ 00:06:24.120 Samuel Roberts: ….
85 00:06:24.740 ⇒ 00:06:29.850 Demilade Agboola: our cursor and, like, ChatGPT, in terms of…
86 00:06:30.420 ⇒ 00:06:41.929 Demilade Agboola: sometimes just using to get a first go at things, and then trying to test the output. If it makes sense, cool. If it doesn’t, which sometimes it doesn’t, I would need to
87 00:06:42.690 ⇒ 00:06:49.170 Demilade Agboola: Create modifications to it, … Basically, tell it explicitly what it needs to do.
88 00:06:49.630 ⇒ 00:06:56.319 Demilade Agboola: … Yeah, I think that’s largely a lot of what I… I tend to do.
89 00:06:56.660 ⇒ 00:06:58.050 Samuel Roberts: ….
90 00:06:59.070 ⇒ 00:07:08.669 Demilade Agboola: the modeling part is the pretty heavy part, and sometimes, just being able to use the context that I have
91 00:07:08.860 ⇒ 00:07:11.589 Demilade Agboola: Plus the context that the tickets have.
92 00:07:11.880 ⇒ 00:07:19.379 Demilade Agboola: To guide it on what it should do, or at least make certain processes faster before I then finish up the step.
93 00:07:22.210 ⇒ 00:07:26.350 Demilade Agboola: So, that’s kind of, like, the week in my life so far.
94 00:07:26.350 ⇒ 00:07:27.470 Awaish Kumar: Totally….
95 00:07:27.660 ⇒ 00:07:30.219 Samuel Roberts: Yeah, go ahead. How do you use, ….
96 00:07:30.720 ⇒ 00:07:35.870 Awaish Kumar: like, chat GPT or Cursor, During your modeling world.
97 00:07:37.520 ⇒ 00:07:43.829 Demilade Agboola: … It depends on the task, different tasks, different, like, mindsets to it.
98 00:07:44.190 ⇒ 00:07:56.220 Demilade Agboola: Some are pretty, like, straightforward in the sense of, if there’s a minor ticket where, like, things need to be changed, I will just straight up put it, like, hey, change this from this to this.
99 00:07:56.440 ⇒ 00:08:03.220 Demilade Agboola: Or… you know, for instance, like, the… Amy pointed out yesterday that
100 00:08:03.920 ⇒ 00:08:17.199 Demilade Agboola: the new customer account was low for Eden, and I kind of just remember that, yeah, we changed the name, so I literally just put it in cursor. The name of this column has changed from New Customer Account to this, make a call list of…
101 00:08:17.580 ⇒ 00:08:28.290 Demilade Agboola: new customer account, use the new name, which is now this, and coalesce it with new customer accounts for the parts where it’s not. And he did it. Like, so things like that are quick fixes.
102 00:08:28.670 ⇒ 00:08:32.130 Demilade Agboola: Some are more, detailed.
103 00:08:32.659 ⇒ 00:08:35.049 Demilade Agboola: So, like, for instance, ChatGPT.
104 00:08:35.059 ⇒ 00:08:42.519 Awaish Kumar: Like, for example… How would you go about, like, for complex data modeling tickets?
105 00:08:42.669 ⇒ 00:08:48.599 Awaish Kumar: So, where we have a lot of business contacts as well, … So, like…
106 00:08:49.709 ⇒ 00:08:54.949 Awaish Kumar: like, there are a lot of things which are generic an LLM can…
107 00:08:55.119 ⇒ 00:08:57.739 Awaish Kumar: Can, like, find the pattern from…
108 00:08:58.249 ⇒ 00:09:01.329 Awaish Kumar: From what it has been trained on, but…
109 00:09:01.639 ⇒ 00:09:12.489 Awaish Kumar: For each use case, or for each company, the thing, like, the business context changes, how they name things, how they define metrics.
110 00:09:12.679 ⇒ 00:09:16.599 Awaish Kumar: So, if you have some complex model.
111 00:09:16.809 ⇒ 00:09:22.559 Awaish Kumar: How do you use any AI, or… and also, like.
112 00:09:23.389 ⇒ 00:09:25.939 Awaish Kumar: What is your process for that?
113 00:09:28.290 ⇒ 00:09:41.060 Demilade Agboola: It depends on the task, and what the task requires of me, because I don’t think there’s one fixed way to use AI. Well, sometimes the task might… the way I might use AI is if, for instance, I didn’t build the model.
114 00:09:41.890 ⇒ 00:09:54.750 Demilade Agboola: I might just literally use AI to explain what… because sometimes reading people’s code is very frustrating, or can be a bit complex, so I just go AI, just explain what’s going on in this model. That gives me an initial first
115 00:09:55.200 ⇒ 00:09:57.679 Demilade Agboola: High-level understanding of what’s going on there.
116 00:09:57.850 ⇒ 00:10:05.439 Demilade Agboola: And that allows me to also now start to think of how to improve or change things there in the model. That’s one step. For some things, I…
117 00:10:06.800 ⇒ 00:10:13.530 Demilade Agboola: if I know this schema, or I have an understanding of the schema, I can feed the schema to…
118 00:10:13.670 ⇒ 00:10:14.640 Demilade Agboola: the…
119 00:10:14.780 ⇒ 00:10:20.420 Demilade Agboola: tool, like ChatGPT, and go, okay, this is the schema for this model, this is the schema for that model.
120 00:10:20.660 ⇒ 00:10:37.860 Demilade Agboola: maybe even download a CSV, of some sample data, like 10 rows or whatever, and go, I need to do this, this, this to this, and I’m trying to create this. Give it the guidelines, these are the keys that join, this is what’s going on here.
121 00:10:38.820 ⇒ 00:10:56.839 Demilade Agboola: I do not want duplicates, I want, like, you know, just, like, basic rules about what I… the aspects I need, and then usually you give me a framework, which I go through, and I can then start to go, okay, no, I can change this instead. There’s a better way to do this, or maybe there’s some context I realize I didn’t tell the…
122 00:10:57.010 ⇒ 00:11:00.819 Demilade Agboola: LLM, so I’ll just, like, You know, add that context.
123 00:11:01.340 ⇒ 00:11:02.950 Demilade Agboola: …
124 00:11:04.510 ⇒ 00:11:09.760 Demilade Agboola: I don’t know, there are different ways, I guess. I think it’s just dependent on what I’m trying to get to.
125 00:11:11.550 ⇒ 00:11:13.040 Samuel Roberts: Yeah, go ahead.
126 00:11:13.040 ⇒ 00:11:16.779 Awaish Kumar: Awesome. I have a… now I have a question for the AI team.
127 00:11:16.780 ⇒ 00:11:17.270 Samuel Roberts: Yeah.
128 00:11:17.270 ⇒ 00:11:24.879 Awaish Kumar: Like, you have listened what Demolares said, how he approaches a data model, building using AI, and …
129 00:11:25.050 ⇒ 00:11:26.870 Awaish Kumar: I’m not sure…
130 00:11:27.050 ⇒ 00:11:40.559 Awaish Kumar: like, that’s what exactly we do, like, we provide maybe schemas, sample data, some calculation of metrics, what we want to build at the end, right? That’s just the context, right? But…
131 00:11:41.090 ⇒ 00:11:42.590 Awaish Kumar: What are some…
132 00:11:43.060 ⇒ 00:11:53.769 Awaish Kumar: ways you, like, an AI agent can help. For example, I see there are a lot of different, agents, like, with fixed prompts, we have already.
133 00:11:53.770 ⇒ 00:11:55.579 Samuel Roberts: In platform, right?
134 00:11:55.580 ⇒ 00:12:08.309 Awaish Kumar: So, we know for SOW, we have already something, and we are just pasting the… the context, and it generates SOW, but there is already some prompt already written there, right? Similarly, like.
135 00:12:08.580 ⇒ 00:12:17.180 Awaish Kumar: Could there be something where we provide the schemas, metric definitions, and things like that, and it learns, like.
136 00:12:17.910 ⇒ 00:12:25.449 Awaish Kumar: Like, if you have… like, we could, like, standardize some prompt which can be used across the models, or something like that.
137 00:12:28.740 ⇒ 00:12:39.020 Samuel Roberts: Yeah, I mean, I think, this is kind of why I’m curious about, you know, the work, the work you guys do, because I don’t have a great insight into, like, what would be the most useful prompts there, but I think we can
138 00:12:39.690 ⇒ 00:12:42.570 Samuel Roberts: You know, definitely… there’s a combination.
139 00:12:42.570 ⇒ 00:12:44.520 Awaish Kumar: Yeah, maybe Casey and Mustaf.
140 00:12:44.520 ⇒ 00:12:56.030 Samuel Roberts: Yeah, I think you guys probably have a better sense of it, definitely, but there’s a combination… I was just gonna say there’s a combination of, like, a prompt like that, like, on the platform, and also building, and I know Utam’s talked about this, but different cursor rules for different tasks.
141 00:12:56.210 ⇒ 00:13:02.720 Samuel Roberts: that would help Cursor have more insight into what you’re doing at different times, but yeah, Casey Mustafa, you guys have any thoughts?
142 00:13:06.630 ⇒ 00:13:14.119 Casie Aviles: Yeah, I mean, I think… We could definitely do something similar, like… depending on…
143 00:13:14.280 ⇒ 00:13:18.380 Casie Aviles: Like, what kind of usage patterns do you have with the…
144 00:13:19.040 ⇒ 00:13:23.619 Casie Aviles: AI, then we can craft some specialized prompts for that.
145 00:13:24.650 ⇒ 00:13:31.180 Casie Aviles: I guess regardless of what kind of work, we could definitely create, like, you know, similar to how even other
146 00:13:32.510 ⇒ 00:13:36.980 Casie Aviles: Non-engineering work that we have, like, generating…
147 00:13:38.550 ⇒ 00:13:48.600 Casie Aviles: For, for, like, emails and stuff, we have, we have different prompts already, because it’s gonna, you know, like, these trends, these patterns are gonna surface when…
148 00:13:49.770 ⇒ 00:13:50.940 Casie Aviles: When you use it.
149 00:13:51.130 ⇒ 00:13:57.460 Casie Aviles: When we, yeah, depending on how we use it, so we could definitely create, like, a prompt for that to help.
150 00:13:57.970 ⇒ 00:14:02.040 Casie Aviles: things, … and much smoother, I guess.
151 00:14:02.440 ⇒ 00:14:03.000 Samuel Roberts: Yeah.
152 00:14:04.990 ⇒ 00:14:12.770 Samuel Roberts: I’m curious… I’m curious about, like, where that would be most helpful. Would it be something in cursor? Would it be something on the platform? Would it be something….
153 00:14:13.500 ⇒ 00:14:14.599 Awaish Kumar: I’m not sure, like.
154 00:14:14.600 ⇒ 00:14:14.950 Samuel Roberts: Yeah, you agree?
155 00:14:14.950 ⇒ 00:14:17.670 Awaish Kumar: something better than, cursor.
156 00:14:18.530 ⇒ 00:14:22.180 Samuel Roberts: Well, I think it depends on the use case. Yeah, I mean, like, that’s… like, we can add….
157 00:14:22.180 ⇒ 00:14:23.230 Awaish Kumar: Caso is….
158 00:14:23.230 ⇒ 00:14:23.860 Samuel Roberts: I am.
159 00:14:24.270 ⇒ 00:14:27.159 Awaish Kumar: Cursor utilizes the NLMs, right?
160 00:14:27.440 ⇒ 00:14:33.780 Demilade Agboola: Yeah, so a limitation with cursor is Cursor doesn’t necessarily accept, files that are not pictures.
161 00:14:34.360 ⇒ 00:14:39.519 Demilade Agboola: So, it’s… you can’t really integrate, like, CSV schemas or whatever, like…
162 00:14:39.710 ⇒ 00:14:41.959 Demilade Agboola: It has to be a picture.
163 00:14:42.410 ⇒ 00:14:51.440 Samuel Roberts: Okay, that’s good to know. So, this is sort of the stuff where I don’t fully know your workflows that well, but… so, if you get, like, CSV files, you want to be able to pull…
164 00:14:52.120 ⇒ 00:15:02.069 Samuel Roberts: like, the schema from that to then be able to pass into Cursor, but right now Cursor just takes images and wouldn’t take that whole file, so that might be something where we could build, yeah, something that would just…
165 00:15:02.850 ⇒ 00:15:07.349 Samuel Roberts: process that, pre-process that, basically, before cursor? Is that what you’re kind of thinking?
166 00:15:07.660 ⇒ 00:15:08.310 Demilade Agboola: like, firm.
167 00:15:08.310 ⇒ 00:15:14.790 Awaish Kumar: Yeah, I’m using, … I’m using a combination of, like, ChatGPT, Press Customer.
168 00:15:15.050 ⇒ 00:15:15.440 Samuel Roberts: Right.
169 00:15:15.560 ⇒ 00:15:21.400 Awaish Kumar: the CSV files and things, all of the context in the raw form to chat GPT.
170 00:15:21.530 ⇒ 00:15:24.500 Awaish Kumar: and ask it to generate a prompt for AI.
171 00:15:24.880 ⇒ 00:15:27.919 Awaish Kumar: And then I copy that and put it in Cursor.
172 00:15:28.140 ⇒ 00:15:29.640 Samuel Roberts: Mmm, okay.
173 00:15:29.830 ⇒ 00:15:38.120 Awaish Kumar: So, then cursor has the sample data tables and the context and, like, a better prop than I could have wrote.
174 00:15:38.960 ⇒ 00:15:43.029 Samuel Roberts: Right. So, like, what are the inputs you’re giving ChatGPT at that point?
175 00:15:43.670 ⇒ 00:15:45.450 Awaish Kumar: Charity really is just…
176 00:15:46.500 ⇒ 00:15:56.200 Awaish Kumar: like, like, what the Demolade said, like, I have schema, sample data, provide metric definitions, and then you say.
177 00:15:56.330 ⇒ 00:15:58.910 Awaish Kumar: … then I don’t ask…
178 00:15:59.200 ⇒ 00:16:05.780 Awaish Kumar: ChatGP to create a model for me. I asked ChatGP to help me create a prompt for an LLM.
179 00:16:06.690 ⇒ 00:16:08.160 Awaish Kumar: And it generates a prompt.
180 00:16:09.160 ⇒ 00:16:09.770 Samuel Roberts: Brain.
181 00:16:10.260 ⇒ 00:16:15.150 Samuel Roberts: That I… and okay. And that’s for, like, a prompt for doing…
182 00:16:15.340 ⇒ 00:16:16.469 Samuel Roberts: what with the LLM am I.
183 00:16:16.470 ⇒ 00:16:19.789 Awaish Kumar: For building a model, right? A data model.
184 00:16:19.790 ⇒ 00:16:20.490 Samuel Roberts: Okay.
185 00:16:20.590 ⇒ 00:16:22.560 Samuel Roberts: Cool, cool, that’s what I’m… that’s what I’m trying to follow along here.
186 00:16:22.560 ⇒ 00:16:26.000 Awaish Kumar: And what the data model is, is basically,
187 00:16:26.440 ⇒ 00:16:32.320 Awaish Kumar: all… all that I provided, like CSV files, schemas, as input, and the….
188 00:16:32.320 ⇒ 00:16:35.789 Samuel Roberts: metric definitions, like, that’s what I want at the end.
189 00:16:35.790 ⇒ 00:16:39.270 Awaish Kumar: These are… The context for building the model.
190 00:16:39.570 ⇒ 00:16:43.309 Awaish Kumar: And … and I asked him to help me write a prompt.
191 00:16:43.460 ⇒ 00:16:49.759 Awaish Kumar: for an LLM, so that it can generate this model, and then you can copy that and put it
192 00:16:50.430 ⇒ 00:16:55.929 Awaish Kumar: and cursor, like, you can give that as an input to another NLM, and then it will write the code.
193 00:16:57.420 ⇒ 00:16:58.260 Samuel Roberts: Okay.
194 00:16:58.840 ⇒ 00:16:59.920 Samuel Roberts: Interesting.
195 00:17:02.260 ⇒ 00:17:11.330 Samuel Roberts: So I’m just processing a little bit, because I’m trying to think, like, I feel like we could definitely put together some kind of… I mean, if this is a common pattern, especially, like, where it’s, like, a bunch of files.
196 00:17:11.450 ⇒ 00:17:12.380 Samuel Roberts: ….
197 00:17:13.300 ⇒ 00:17:15.560 Awaish Kumar: Yeah, that’s… that is… that’s very common.
198 00:17:15.560 ⇒ 00:17:21.399 Samuel Roberts: Yeah, so, because if Cursor can’t handle that, that’s definitely something we could probably streamline a little bit, …
199 00:17:23.970 ⇒ 00:17:25.180 Samuel Roberts: Yeah….
200 00:17:26.339 ⇒ 00:17:35.709 Mustafa Raja: Shouldn’t Cursor be able to, handle these files if we put them in a folder named ForCursor, and then throw those, and then add those files to the….
201 00:17:35.710 ⇒ 00:17:39.349 Samuel Roberts: Oh, sure. So, is it that they’re not part of the current workspace?
202 00:17:39.350 ⇒ 00:17:45.890 Mustafa Raja: Yeah, what I’m understanding is these CSVs that, are not part of codebase at all.
203 00:17:46.300 ⇒ 00:17:53.009 Samuel Roberts: Got it, yeah, I think that’s… okay, yeah, interesting. Are these coming from somewhere else and not part of the code? Is that what is happening here?
204 00:17:55.210 ⇒ 00:17:56.619 Awaish Kumar: What are you saying, Depending?
205 00:17:59.400 ⇒ 00:18:00.930 Demilade Agboola: Alright, I think that’s the question.
206 00:18:01.190 ⇒ 00:18:07.970 Samuel Roberts: Oh, I think, what Gustava was pointing out is that, like, you can usually add these kinds of files, but they have to be part of the current codebase.
207 00:18:08.250 ⇒ 00:18:11.379 Samuel Roberts: So these are external files, somehow, that we.
208 00:18:11.380 ⇒ 00:18:12.290 Demilade Agboola: Yeah, yeah.
209 00:18:12.290 ⇒ 00:18:13.180 Samuel Roberts: Okay.
210 00:18:13.180 ⇒ 00:18:20.600 Demilade Agboola: You can add them, but you have to be part of the codebase, but you don’t want a situation where you push a CLG into… yeah, so…
211 00:18:21.460 ⇒ 00:18:23.929 Demilade Agboola: That would be far from ideal.
212 00:18:24.330 ⇒ 00:18:25.990 Samuel Roberts: Yeah, of course. I’m wondering….
213 00:18:27.380 ⇒ 00:18:32.750 Mustafa Raja: Maybe, … .gitignore for that folder or something.
214 00:18:34.260 ⇒ 00:18:36.740 Demilade Agboola: Yeah, fair enough, that’s one way to go about him.
215 00:18:37.480 ⇒ 00:18:43.250 Samuel Roberts: Right, so if we could… if we… so we might be able to, like, kind of leapfrog ChatGPT if we could just…
216 00:18:43.590 ⇒ 00:18:47.259 Samuel Roberts: keep those in, like, a special folder that Git ignores.
217 00:18:47.600 ⇒ 00:18:51.820 Samuel Roberts: … And then point cursor to that folder.
218 00:18:55.060 ⇒ 00:19:08.880 Mustafa Raja: Cursor usually all, also, ignores those files that we put in, gitignore, but there’s a workaround for that, and, I’m not….
219 00:19:09.100 ⇒ 00:19:09.540 Samuel Roberts: Oh, what is it?
220 00:19:09.540 ⇒ 00:19:10.090 Mustafa Raja: really show.
221 00:19:10.090 ⇒ 00:19:12.150 Samuel Roberts: I know cursor has its own cursor ignored, does it also.
222 00:19:12.150 ⇒ 00:19:13.610 Mustafa Raja: Yeah, yeah, yeah.
223 00:19:13.610 ⇒ 00:19:14.360 Samuel Roberts: Oh, okay.
224 00:19:14.360 ⇒ 00:19:16.359 Mustafa Raja: We need to look into this.
225 00:19:16.480 ⇒ 00:19:18.649 Mustafa Raja: Yeah, okay, definitely. That is set up.
226 00:19:22.470 ⇒ 00:19:35.549 Samuel Roberts: Interesting. Okay, that’s good to know. Yeah, I mean, this is… I mean, this is exactly why I wanted to have this conversation, because I don’t have a great, you know, sense of how… how these workflows go for… for you guys on the other side here. So, would…
227 00:19:36.460 ⇒ 00:19:37.980 Samuel Roberts: Hmm. Okay.
228 00:19:38.180 ⇒ 00:19:47.249 Samuel Roberts: We’re kind of getting to the time when I’m going to have to hop off here, so I… I mean, this is actually very helpful, because I’m getting a better sense of how some of these workflows go.
229 00:19:47.360 ⇒ 00:19:52.770 Samuel Roberts: So… To do from here would definitely be figure out the gitignore vs. cursor ignore stuff.
230 00:19:53.100 ⇒ 00:20:02.179 Samuel Roberts: we can… … C… if there’s a good way for Cursor to be able to ingest those files.
231 00:20:02.400 ⇒ 00:20:04.990 Samuel Roberts: But not commit them.
232 00:20:06.020 ⇒ 00:20:08.869 Samuel Roberts: It should be doable, but it was probably a little bit of a dance with Curtis.
233 00:20:08.870 ⇒ 00:20:09.210 Mustafa Raja: Yeah.
234 00:20:09.210 ⇒ 00:20:12.489 Samuel Roberts: and get ignore. Okay, cool.
235 00:20:13.020 ⇒ 00:20:13.940 Samuel Roberts: Yeah.
236 00:20:14.140 ⇒ 00:20:17.450 Samuel Roberts: Don’t lie, do you mind if I, like, message you at some point, and we can try to, like…
237 00:20:18.150 ⇒ 00:20:24.369 Samuel Roberts: test something out, or if I do a little bit of research on the gitignore and cursor ignore kind of stuff, …
238 00:20:24.490 ⇒ 00:20:25.990 Samuel Roberts: And maybe we can just, like…
239 00:20:26.350 ⇒ 00:20:28.730 Samuel Roberts: Go back and forth a little bit about how something like that might work?
240 00:20:29.180 ⇒ 00:20:30.369 Demilade Agboola: Yeah, sure, no problem.
241 00:20:30.370 ⇒ 00:20:43.330 Samuel Roberts: All right, cool, cool. Yeah, sorry for being just so, like, I just have, you know, very little insight into anything but, like, kind of the, stuff I’ve been touching. So, if I’m asking super basic questions or anything, forgive me, but…
242 00:20:43.450 ⇒ 00:20:44.890 Samuel Roberts: Okay.
243 00:20:45.070 ⇒ 00:20:50.839 Samuel Roberts: Yeah, like I said, this would probably be, like I said, pretty quick today, because I have to hop off.
244 00:20:51.350 ⇒ 00:20:55.639 Samuel Roberts: But was there anything else anyone wanted to just, like, chat about as, like, an engineering team?
245 00:20:58.760 ⇒ 00:21:00.859 Awaish Kumar: No, I don’t let you….
246 00:21:00.860 ⇒ 00:21:01.790 Samuel Roberts: Alright, cool.
247 00:21:02.250 ⇒ 00:21:03.270 Awaish Kumar: Pick it off, Wolf.
248 00:21:03.890 ⇒ 00:21:06.330 Awaish Kumar: Do you read the workspace?
249 00:21:07.810 ⇒ 00:21:08.570 Samuel Roberts: I’m sorry?
250 00:21:09.720 ⇒ 00:21:12.549 Awaish Kumar: Like, do you need a Zoom workspace as well, or…?
251 00:21:12.550 ⇒ 00:21:17.270 Samuel Roberts: I don’t think so, I think it’s someone else’s, but I’m good to just hop off, I just don’t want to, like, kill it for everyone if I’m… yeah, if I can just leave.
252 00:21:17.270 ⇒ 00:21:17.610 Awaish Kumar: Okay.
253 00:21:17.610 ⇒ 00:21:25.750 Samuel Roberts: But, yeah, besides that, thank you all. Hopefully… this will be a little less… I’m trying to try to figure it out, because I’m not sure if this meeting is….
254 00:21:26.050 ⇒ 00:21:29.410 Awaish Kumar: Is all the time or not? Yeah, like, we can stay and discuss, like.
255 00:21:29.970 ⇒ 00:21:33.279 Awaish Kumar: For, like, a few more minutes here, and ….
256 00:21:33.280 ⇒ 00:21:33.880 Samuel Roberts: Perfect.
257 00:21:34.000 ⇒ 00:21:34.820 Samuel Roberts: Great.
258 00:21:35.270 ⇒ 00:21:35.920 Awaish Kumar: Yep.
259 00:21:35.920 ⇒ 00:21:39.370 Samuel Roberts: Thank you all, sorry for having to run, but, appreciate the time.
260 00:21:40.140 ⇒ 00:21:40.940 Samuel Roberts: Right.
261 00:21:41.400 ⇒ 00:21:42.140 Samuel Roberts: Bye.
262 00:21:45.520 ⇒ 00:21:46.210 Awaish Kumar: Okay.
263 00:21:47.440 ⇒ 00:21:58.190 Awaish Kumar: Yeah, so… Is there anyone… Wanna discuss anything, like… Any, …
264 00:22:01.270 ⇒ 00:22:07.820 Awaish Kumar: Improvements, optimization, like, any suggestion in any kind of work, like, we are doing.
265 00:22:11.850 ⇒ 00:22:16.349 Awaish Kumar: I would love for, like, each of you to give some input, like.
266 00:22:22.040 ⇒ 00:22:24.869 Demilade Agboola: Is this AI-related, or just generally speaking?
267 00:22:24.870 ⇒ 00:22:32.359 Awaish Kumar: It can be, like, anything to optimize our workflows, right? Demolari, whatever you are involved in, you think, okay.
268 00:22:32.600 ⇒ 00:22:43.160 Awaish Kumar: some things can be optimized, which can, like, some… anything which can help, the team, right? Not just, like, you have figured out something, or…
269 00:22:43.280 ⇒ 00:22:49.099 Awaish Kumar: Some ways which can also, like, help me, … Then, like, things like that.
270 00:22:56.060 ⇒ 00:23:02.179 Demilade Agboola: I don’t know if this is something I’ve noticed, is if the ticket is really simple.
271 00:23:02.380 ⇒ 00:23:10.039 Demilade Agboola: like, there isn’t a lot of complexity. You can actually just put the ticket in cost of, like, the definition of, like, copy out the ticket.
272 00:23:10.200 ⇒ 00:23:11.730 Demilade Agboola: I’ll put it in cursor.
273 00:23:11.950 ⇒ 00:23:18.459 Demilade Agboola: And direct… tell it the models it needs to work on, and he will do a pretty decent first job on…
274 00:23:18.680 ⇒ 00:23:21.050 Demilade Agboola: The ticket, and then you can now finish it up.
275 00:23:21.750 ⇒ 00:23:22.989 Awaish Kumar: Oh, goodness.
276 00:23:24.380 ⇒ 00:23:28.010 Awaish Kumar: So, do you think, if the ticket is groomed well, right?
277 00:23:28.670 ⇒ 00:23:44.579 Demilade Agboola: Yes, yes, if the ticket is groomed well, things are properly explained and detailed there. You can copy it, tell it the models that it needs to make the changes on, and it will do it. Again, it’s not perfect. Sometimes you look at it and, like, no, you didn’t fully understand it.
278 00:23:44.680 ⇒ 00:23:54.420 Demilade Agboola: But usually, you can get about 50 to 60 to… maybe 50-some percent of the way there, and then you can do the rest by yourself.
279 00:23:56.360 ⇒ 00:23:57.070 Awaish Kumar: Okay.
280 00:23:57.720 ⇒ 00:24:01.289 Casie Aviles: Yeah, that’s… that’s interesting. We were actually…
281 00:24:01.540 ⇒ 00:24:07.970 Casie Aviles: We had, like, in our backlog, the ticket grooming… AI agent?
282 00:24:09.390 ⇒ 00:24:14.310 Casie Aviles: Although right now, it’s not… it’s not widely available across all the teams.
283 00:24:15.480 ⇒ 00:24:26.950 Casie Aviles: But yeah, like, I guess that would also help, like, optimize that process, because, the idea is you would go to a ticket, and then you can tag the bot, and then it should
284 00:24:27.270 ⇒ 00:24:33.890 Casie Aviles: Provide, like, a review of the ticket, and, like, what’s lacking, what’s… Yeah.
285 00:24:34.040 ⇒ 00:24:35.260 Casie Aviles: Yeah, provide.
286 00:24:35.260 ⇒ 00:24:37.320 Awaish Kumar: Part of, like, ticket creation?
287 00:24:39.740 ⇒ 00:24:40.550 Casie Aviles: …
288 00:24:40.880 ⇒ 00:24:54.520 Casie Aviles: Well, how it works right now is you have to create the ticket first, and then you have to manually tag the bot. But I think, yeah, I get… I do see where you’re coming from. I think, yeah, that would also be good, like, if it’s automatic, right, when upon ticket.
289 00:24:54.520 ⇒ 00:24:56.739 Awaish Kumar: Yeah, we have a meeting….
290 00:24:57.270 ⇒ 00:25:00.689 Casie Aviles: Script, transcript and everything.
291 00:25:00.960 ⇒ 00:25:09.789 Awaish Kumar: Maybe you will be also using the… some related meetings and the Slack messages to get the context for that, right?
292 00:25:10.500 ⇒ 00:25:11.519 Awaish Kumar: to groom it.
293 00:25:12.550 ⇒ 00:25:14.169 Casie Aviles: Oh, okay, yeah.
294 00:25:14.170 ⇒ 00:25:18.049 Awaish Kumar: Like, to groom the ticket, you obviously get information from somewhere.
295 00:25:18.180 ⇒ 00:25:22.410 Awaish Kumar: That would be either a Zoom meeting or a Slack message.
296 00:25:22.860 ⇒ 00:25:25.889 Awaish Kumar: Or something on linear ticket itself, right?
297 00:25:26.920 ⇒ 00:25:28.009 Casie Aviles: Yeah, yeah, yeah.
298 00:25:28.220 ⇒ 00:25:28.780 Casie Aviles: Yeah.
299 00:25:28.780 ⇒ 00:25:44.870 Awaish Kumar: And … and, like, when we are on this platform, we already, like, have a meeting, and we say, okay, let’s create a ticket. So instead of just focusing on that meeting, we focus on that meeting to create the ticket, but for the context, we can, like.
300 00:25:45.380 ⇒ 00:25:54.609 Awaish Kumar: Find out relevant meetings and the relevant tickets and relevant text messages to add the descriptions as well.
301 00:25:56.170 ⇒ 00:25:56.950 Casie Aviles: I mean, like….
302 00:25:57.340 ⇒ 00:25:59.370 Awaish Kumar: Tickets are groomed on the spot.
303 00:26:00.560 ⇒ 00:26:05.259 Casie Aviles: Yeah, yeah, we could… Possibly leverage our client hub.
304 00:26:05.560 ⇒ 00:26:10.949 Casie Aviles: Similar to how we did the Client Hub, where it has contacts on the Zoom meetings.
305 00:26:11.360 ⇒ 00:26:13.660 Casie Aviles: And also the Slack messages.
306 00:26:14.600 ⇒ 00:26:17.480 Casie Aviles: Yeah, that’s also something we can do.
307 00:26:17.630 ⇒ 00:26:18.480 Casie Aviles: For sure.
308 00:26:19.830 ⇒ 00:26:25.579 Casie Aviles: It’s just going to be a matter of probably optimizing the responses, because…
309 00:26:26.280 ⇒ 00:26:28.370 Casie Aviles: It can be a challenge to, like.
310 00:26:29.690 ⇒ 00:26:35.079 Casie Aviles: What do you call this? To get, like, the relevant output, because you can dump a lot of
311 00:26:35.240 ⇒ 00:26:41.919 Casie Aviles: context, but… it’s a challenge to, like, get which one is actually the most relevant, I guess.
312 00:26:42.120 ⇒ 00:26:45.149 Awaish Kumar: But yeah, that’s something we have to optimize for.
313 00:26:47.610 ⇒ 00:26:48.460 Awaish Kumar: Okay.
314 00:26:50.220 ⇒ 00:26:53.450 Awaish Kumar: Yeah, great, like, that would be really helpful for…
315 00:26:53.730 ⇒ 00:26:56.959 Awaish Kumar: Like, that reduces our time to groom it.
316 00:26:57.330 ⇒ 00:26:59.959 Awaish Kumar: look for… look… ask AI for…
317 00:27:00.390 ⇒ 00:27:04.770 Awaish Kumar: Contacts and clarifications. And then also, like.
318 00:27:04.900 ⇒ 00:27:09.089 Awaish Kumar: we gather all that information and put it in AI,
319 00:27:09.200 ⇒ 00:27:11.520 Awaish Kumar: like, as Demolari said, we can just…
320 00:27:11.700 ⇒ 00:27:18.259 Awaish Kumar: Grab that ticket and copy-paste the description to the cursor, and it can… it will generate the model.
321 00:27:19.770 ⇒ 00:27:21.960 Casie Aviles: Yes, yes, yeah, that’s… that’s a good idea.
322 00:27:21.960 ⇒ 00:27:26.470 Awaish Kumar: Yeah, and then we are mostly, like, reviewer and…
323 00:27:26.640 ⇒ 00:27:29.259 Awaish Kumar: Of the work instead of a different work.
324 00:27:31.620 ⇒ 00:27:32.310 Casie Aviles: Hmm.
325 00:27:34.450 ⇒ 00:27:35.280 Awaish Kumar: Okay.
326 00:27:35.740 ⇒ 00:27:38.780 Awaish Kumar: What about you? Any… any thoughts?
327 00:27:42.840 ⇒ 00:27:49.230 Annie Yu: One thing I’ve… I’ve been using cursors sometimes to ask about models.
328 00:27:49.400 ⇒ 00:28:00.709 Annie Yu: let’s say, like, a primary key, and what each row represents, and I think there are at least two cases where it’s wrong, so I eventually have to copy and paste those
329 00:28:00.840 ⇒ 00:28:12.509 Annie Yu: And to ask ChatGPT, which, so far has always been right. So that’s just one observation I, I’ve noticed, at least from that use case.
330 00:28:12.830 ⇒ 00:28:19.360 Annie Yu: But I don’t know if that’s something that we can do anything about, but that’s just, like, for awareness.
331 00:28:20.950 ⇒ 00:28:24.760 Awaish Kumar: So, but, like, in your work, like, One…
332 00:28:25.000 ⇒ 00:28:27.300 Awaish Kumar: One thing I understand, which you also…
333 00:28:27.440 ⇒ 00:28:37.650 Awaish Kumar: said, many times that, like, you could use AI to maybe get context, or more clarification, or business context, whatever.
334 00:28:37.780 ⇒ 00:28:40.580 Awaish Kumar: But apart from that, like, …
335 00:28:41.170 ⇒ 00:28:45.880 Awaish Kumar: Do you leverage AI to help you with any development work?
336 00:28:47.190 ⇒ 00:28:56.370 Annie Yu: Mmm… I would say not so much with Tableau. Well, I think I use it to help
337 00:28:56.520 ⇒ 00:29:02.719 Annie Yu: troubleshoot my, error with my Tableau formula, but in terms of
338 00:29:03.010 ⇒ 00:29:11.580 Annie Yu: what to build, and things of that sort. I’ve tried to use it, but I don’t think it’s not… I think it’s not that smart yet.
339 00:29:14.030 ⇒ 00:29:17.690 Awaish Kumar: You mean what kind of charts we should build?
340 00:29:17.690 ⇒ 00:29:18.440 Annie Yu: I think it knows that.
341 00:29:18.440 ⇒ 00:29:19.650 Awaish Kumar: And I’ll answer that.
342 00:29:19.650 ⇒ 00:29:30.519 Annie Yu: Yeah, I think it knows the basics, but with Eden, we’ve done some pretty different charts than your, like, typical ones. That’s where…
343 00:29:30.690 ⇒ 00:29:34.019 Annie Yu: I think, ChatGPT is not that smart yet.
344 00:29:34.960 ⇒ 00:29:36.140 Annie Yu: Which makes sense.
345 00:29:37.250 ⇒ 00:29:43.920 Awaish Kumar: Okay, well, yeah, that is one thing. So, but Casey, are there any AI tools?
346 00:29:44.370 ⇒ 00:29:48.149 Awaish Kumar: help you with, like, development in Tableau?
347 00:29:50.180 ⇒ 00:29:58.600 Casie Aviles: I haven’t really… I’m not super knowledgeable in Tableau work, so honestly, I just use
348 00:29:59.020 ⇒ 00:30:02.550 Casie Aviles: the same tools that we… every engineer here uses, like…
349 00:30:03.720 ⇒ 00:30:06.549 Casie Aviles: So far, like Courser and also ChatGPT.
350 00:30:07.950 ⇒ 00:30:09.440 Awaish Kumar: Yeah, I bet.
351 00:30:12.210 ⇒ 00:30:25.810 Awaish Kumar: Because, like, Tableau is more, like, UI-driven, tone, right? So it’s… It’ll be hard to… like, …
352 00:30:25.940 ⇒ 00:30:28.040 Awaish Kumar: Become an extension for that.
353 00:30:28.280 ⇒ 00:30:31.999 Awaish Kumar: Tableau itself can create some AI features, but…
354 00:30:32.180 ⇒ 00:30:34.680 Awaish Kumar: For external tools, it will be harder.
355 00:30:34.900 ⇒ 00:30:39.110 Awaish Kumar: like, compared to, like, real data, for example, that’s port-based.
356 00:30:39.440 ⇒ 00:30:43.089 Awaish Kumar: And, like, It’s easier to, like, …
357 00:30:43.450 ⇒ 00:30:46.359 Awaish Kumar: get… get development help from culture.
358 00:30:46.780 ⇒ 00:30:48.680 Awaish Kumar: From, from AI, right?
359 00:30:49.460 ⇒ 00:31:00.289 Casie Aviles: Yes, yes, if… yeah, for real, there’s, like, there… you have code, so you can just give it as context to the AI, and it can help you. And if we… unless we could do that with
360 00:31:00.690 ⇒ 00:31:08.249 Casie Aviles: Tableau, then I guess it’s going to be harder if it’s, you know, visual… if it relies on a visual interface.
361 00:31:10.890 ⇒ 00:31:15.300 Casie Aviles: I guess we could… I mean, a workaround is to have, like.
362 00:31:15.410 ⇒ 00:31:18.789 Casie Aviles: Screenshots or images, but yeah, that’s just okay.
363 00:31:19.000 ⇒ 00:31:20.719 Casie Aviles: One way we could do it, but…
364 00:31:20.870 ⇒ 00:31:22.910 Casie Aviles: Not the most elegant, I guess.
365 00:31:24.340 ⇒ 00:31:27.660 Awaish Kumar: And that does not help with development, right? That’s just…
366 00:31:28.040 ⇒ 00:31:32.439 Awaish Kumar: you can grab the data from Tableau to ChatGPT.
367 00:31:32.800 ⇒ 00:31:34.770 Casie Aviles: Yeah, yeah, yeah. It’s marital.
368 00:31:34.770 ⇒ 00:31:38.780 Awaish Kumar: Yeah, it won’t create anything in Tableau.
369 00:31:40.000 ⇒ 00:31:40.810 Casie Aviles: No.
370 00:31:42.710 ⇒ 00:31:44.710 Awaish Kumar: Okay, yeah.
371 00:31:45.050 ⇒ 00:31:53.870 Awaish Kumar: That, like, Tableau itself can build, like, some AI features. Okay, like, help me write a… Instead of, like…
372 00:31:55.230 ⇒ 00:31:58.699 Awaish Kumar: Reselecting the charts and things, like…
373 00:31:58.810 ⇒ 00:32:02.959 Awaish Kumar: AI can, like, Tableau can get the context and create chart, right?
374 00:32:04.680 ⇒ 00:32:11.750 Awaish Kumar: Because they… they have the UI, and in the backend, it’s all code, right?
375 00:32:12.010 ⇒ 00:32:17.080 Awaish Kumar: For us, it’s UI, so they can implement it internally, if they want.
376 00:32:17.080 ⇒ 00:32:17.630 Casie Aviles: jump in.
377 00:32:17.630 ⇒ 00:32:18.720 Awaish Kumar: I suggest….
378 00:32:20.170 ⇒ 00:32:29.060 Casie Aviles: Yeah, if… so I think, yeah, that’s what… yeah, they would have to do that. What… what most people do with, you know, how… how they…
379 00:32:29.470 ⇒ 00:32:31.620 Casie Aviles: allow the AI to interface with
380 00:32:31.770 ⇒ 00:32:40.089 Casie Aviles: different kinds of software is if, you know, if there’s, like… before it was API-based, then now there’s also MCP. That’s also another…
381 00:32:40.710 ⇒ 00:32:48.579 Casie Aviles: Potential… option, if Tableau builds out an MCP for… Yeah, for their tools, so…
382 00:32:48.940 ⇒ 00:32:53.599 Casie Aviles: AI can definitely use that to interface with Tableau, but…
383 00:32:53.830 ⇒ 00:32:58.379 Casie Aviles: Until then, if they… if there’s… if that doesn’t exist, then yeah, it’ll be difficult.
384 00:32:59.630 ⇒ 00:33:02.699 Casie Aviles: The other option is API, which is…
385 00:33:03.750 ⇒ 00:33:08.069 Casie Aviles: how tools were created before, before MCP, so…
386 00:33:08.430 ⇒ 00:33:13.970 Casie Aviles: We would connect, like, tools, and then it’s gonna send the AIs… you’re just gonna prompt the AI with
387 00:33:14.110 ⇒ 00:33:18.029 Casie Aviles: you know, when to use these tools, and then they would perform actions. So that’s how…
388 00:33:18.300 ⇒ 00:33:21.690 Casie Aviles: a lot of our N8N workflows actually work,
389 00:33:22.000 ⇒ 00:33:29.650 Casie Aviles: the AI agents can interface with, for example, Google Sheets or Google Docs using their API.
390 00:33:30.700 ⇒ 00:33:34.429 Casie Aviles: Yeah, that’s… that’s… One way to do it.
391 00:33:36.050 ⇒ 00:33:36.800 Awaish Kumar: Okay.
392 00:33:43.580 ⇒ 00:33:49.880 Awaish Kumar: Yeah, it was a really nice conversation. Anyone wants to, like, fill in anything, or…
393 00:33:50.150 ⇒ 00:33:52.110 Awaish Kumar: Otherwise, we can hop off.
394 00:33:55.440 ⇒ 00:33:57.379 Demilade Agboola: Nothing for me right now.
395 00:33:58.850 ⇒ 00:33:59.240 Awaish Kumar: Okay.
396 00:33:59.260 ⇒ 00:34:00.870 Casie Aviles: All good, all good friend, dude.
397 00:34:00.870 ⇒ 00:34:03.549 Awaish Kumar: Alright, thank you, thank you so much.
398 00:34:03.910 ⇒ 00:34:05.330 Awaish Kumar: Yeah, see you.
399 00:34:06.360 ⇒ 00:34:07.780 Demilade Agboola: Thank you. Bye.
400 00:34:07.780 ⇒ 00:34:09.400 Awaish Kumar: Next meetings. Bye.
401 00:34:09.800 ⇒ 00:34:10.519 Casie Aviles: Thank you.