Meeting Title: Uttam_Mitchell: Stackblitz Data Needs Date: 2024-12-18 Meeting participants: Mitchell, Uttam Kumaran
WEBVTT
1 00:00:47.700 ⇒ 00:00:49.899 mitchell: What’s up? Bhutan? You there.
2 00:00:50.940 ⇒ 00:00:52.030 Uttam Kumaran: Hey, dude!
3 00:00:54.060 ⇒ 00:00:55.090 Uttam Kumaran: Can you see me.
4 00:00:55.380 ⇒ 00:00:56.020 mitchell: Yeah, yeah.
5 00:00:56.020 ⇒ 00:00:56.790 Uttam Kumaran: Up, dude.
6 00:00:57.330 ⇒ 00:00:59.039 mitchell: Much, man, how you doing? It’s been a while.
7 00:00:59.040 ⇒ 00:01:05.940 Uttam Kumaran: Good. It’s been a long time, I mean, we talk in slack. But yeah, how’s life?
8 00:01:06.500 ⇒ 00:01:15.003 mitchell: You know, it’s it’s pretty good. Yeah. Things are pretty good. Just yeah. Started the new job last week. And
9 00:01:17.130 ⇒ 00:01:27.029 mitchell: yeah, the originally it was supposed to be kind of pretty focused on like data. But yesterday just had a chat with the CEO. And he
10 00:01:27.740 ⇒ 00:01:37.789 mitchell: basically said, it’s like, actually like, why don’t we kind of roll up for now, like marketing and data, and maybe customer success under you? And we’ll just like
11 00:01:38.050 ⇒ 00:01:40.499 mitchell: kind of, I was like.
12 00:01:40.500 ⇒ 00:01:45.629 Uttam Kumaran: What? Wait? What did you come in? What did you come in? As like just data stuff, mainly.
13 00:01:46.235 ⇒ 00:01:53.710 mitchell: It was pretty vague, but the title it gave me was like, Go to market analytics. Engineer. So
14 00:01:53.830 ⇒ 00:02:08.440 mitchell: yeah, that was kind of what I was planning on but anyway, so I was like, okay, like, I talked with Hank. And Hank’s like, you know, if someone offers you the power like you should take it unless you really.
15 00:02:08.440 ⇒ 00:02:08.830 Uttam Kumaran: Totally.
16 00:02:08.870 ⇒ 00:02:09.440 mitchell: I think.
17 00:02:09.449 ⇒ 00:02:10.359 Uttam Kumaran: I agree.
18 00:02:10.509 ⇒ 00:02:15.932 mitchell: Yeah. So anyway. So that’s yeah. So that’s kind of that. But
19 00:02:16.559 ⇒ 00:02:30.519 mitchell: as part of that, like, we, we currently don’t really have like, basically right now, what they’re doing data wise is they have segment, and they’re piping some like random events
20 00:02:30.759 ⇒ 00:02:35.339 mitchell: through segment and into mix panel. And
21 00:02:35.979 ⇒ 00:02:41.229 mitchell: they’re using mixed panel and then bare metrics. For, like financial related stuff.
22 00:02:41.230 ⇒ 00:02:41.880 Uttam Kumaran: Okay.
23 00:02:42.197 ⇒ 00:02:50.502 mitchell: That’s kind of what they’re doing right now. I don’t love it like the eventing. There’s not really like any rhyme or reason to it. Really?
24 00:02:51.680 ⇒ 00:03:00.669 mitchell: so yeah, so just would love to like, actually get some decent data
25 00:03:01.050 ⇒ 00:03:08.950 mitchell: structure going. I’ve got like a decent idea of some initial metrics and
26 00:03:09.780 ⇒ 00:03:26.999 mitchell: what entities will look like data wise. Everything’s pretty much in like we’ve got like a couple of postgres databases. And then we’ve got some stuff in R. 2 on cloudflare. So there’s nothing like super difficult there.
27 00:03:28.210 ⇒ 00:03:29.080 mitchell: yeah.
28 00:03:29.240 ⇒ 00:03:54.030 mitchell: I don’t feel like the modeling is gonna be too bad. I have a pretty good idea of what I want the the end entities to look like. And I’ve got some documentation on that. So yeah, I mean, ideally, it would be like a pretty straightforward engagement. For you all. But essentially, we’d need to start from ground 0 of just like, yeah, setting up the back and then
29 00:03:54.270 ⇒ 00:03:58.123 mitchell: kind of like what you were saying with like, there’s gonna be some
30 00:03:58.520 ⇒ 00:04:05.700 mitchell: gonna need to do like some masking on Pii and setting up our back, and all that kind of.
31 00:04:07.070 ⇒ 00:04:10.180 Uttam Kumaran: Okay, that makes sense. I mean one. I think
32 00:04:10.690 ⇒ 00:04:23.370 Uttam Kumaran: I mean, it’s it’s I would say it’d be great working with you because you have a good idea. The end state a lot of folks we work with are at like the CEO or Co. Level, but I’ve never done it from scratch. So it’s mainly my job is like.
33 00:04:23.500 ⇒ 00:04:38.860 Uttam Kumaran: try to avoid the usual set of mistakes. The nice thing here is like, if you’re like, very clear. I’m like, yeah, I just need these like 10 metrics. I need them cut these different ways. Then everything around how we get there is.
34 00:04:39.090 ⇒ 00:04:58.250 Uttam Kumaran: you know, the usual, which is like we pick a good etl tool. We’ll pick. We’ll just do everything on Snowflake and Dbt, and then have everything through github and then we usually do. We can run stuff on dbt, core, or we, if you guys want to use Dbt, cloud, so we have like a sort of a de development environment.
35 00:04:58.810 ⇒ 00:05:01.040 Uttam Kumaran: it’s a little bit expensive, but it’s.
36 00:05:01.040 ⇒ 00:05:06.100 mitchell: Yeah, I I don’t necessarily need cloud. But okay.
37 00:05:06.430 ⇒ 00:05:10.560 mitchell: cause even even at Vercel, like, I didn’t use cloud for
38 00:05:10.830 ⇒ 00:05:16.728 mitchell: development. We just used it to. Yeah, run all our Ci basically. So
39 00:05:17.500 ⇒ 00:05:18.210 Uttam Kumaran: Okay.
40 00:05:18.210 ⇒ 00:05:26.379 mitchell: Ticd stuff. So I mean, I’m open to just running core. I don’t know like anything that we’re missing out on by doing it that way. But.
41 00:05:26.380 ⇒ 00:05:31.709 Uttam Kumaran: No, the unless people care about like the data dictionary, or like that sort of or like lineage.
42 00:05:31.820 ⇒ 00:05:41.909 Uttam Kumaran: that’s otherwise. It’s kind of useless. And the nice thing is we can have a 1 free cloud account that we can just share in case we need it. Basically but we usually run.
43 00:05:42.100 ⇒ 00:05:46.379 Uttam Kumaran: We’ll usually just run the job. We we for some clients. We even run the jobs in Github.
44 00:05:46.480 ⇒ 00:05:57.719 Uttam Kumaran: But if it’s if it’s mostly just running Dbt, then, yeah, we’ll just use the one cloud account to to just trigger all the the main jobs. But again, it’s really probably narrow. I think the biggest thing to think about is like.
45 00:05:58.020 ⇒ 00:06:03.659 Uttam Kumaran: are you going to be building towards just like a modeling layer? Are you also going to be building like reporting tables
46 00:06:03.790 ⇒ 00:06:05.979 Uttam Kumaran: or those like one in the same
47 00:06:06.840 ⇒ 00:06:07.530 mitchell: But yeah.
48 00:06:07.530 ⇒ 00:06:11.299 Uttam Kumaran: Or and also, are you starting into? Are you gonna bite off
49 00:06:11.440 ⇒ 00:06:17.749 Uttam Kumaran: once part of the business like you mentioned? Marketing? Cs finance like, what part are you gonna bite off.
50 00:06:18.140 ⇒ 00:06:19.540 Uttam Kumaran: you know. First? st
51 00:06:20.004 ⇒ 00:06:22.830 Uttam Kumaran: I think that would also be helpful to know.
52 00:06:23.070 ⇒ 00:06:42.569 mitchell: Yeah. So here’s here’s kind of what I’ve been working on. Let’s see, is that the right? Yeah, okay, great so basically, just I’ve kind of broken these down into like different steps of the customer journey. And the different metrics that that I think are kind of important for us to get started with
53 00:06:43.420 ⇒ 00:06:47.000 mitchell: so it it feels fairly.
54 00:06:47.000 ⇒ 00:06:47.440 Uttam Kumaran: Nice.
55 00:06:47.440 ⇒ 00:06:48.220 mitchell: Forward.
56 00:06:50.186 ⇒ 00:06:59.410 mitchell: there are, you know, an okay number of metrics here, and and probably like some of these like performance and reliability ones, we could probably knock those off and not even worry about those, for now
57 00:06:59.897 ⇒ 00:07:05.260 mitchell: and then, so based off those metrics like, I started working on the entities, and then with those.
58 00:07:05.260 ⇒ 00:07:05.820 Uttam Kumaran: Nice.
59 00:07:05.820 ⇒ 00:07:24.046 mitchell: The different columns that we’ll want on each of those entities. So my thought is like, I can just kind of hand this over to you guys, and it should be pretty straightforward for you all to build like there should be enough there that, like, I want a subscriptions model. And like. Here are the columns I want, and
60 00:07:24.660 ⇒ 00:07:32.038 mitchell: a anything you need, as far as like, if we don’t have data, then we can work on an instrumenting events. There.
61 00:07:32.890 ⇒ 00:07:40.309 mitchell: I’ve been kind of going through and thinking, rethinking cause right now. The events they have are like very much
62 00:07:42.290 ⇒ 00:07:49.369 mitchell: I don’t know. They’re just like random, like, whenever someone is like. Oh, I think I should have this event. They’ll put it in. There’s no like real structure.
63 00:07:49.370 ⇒ 00:07:53.160 Uttam Kumaran: There’s usually 2 models, one, it’s like track everything or.
64 00:07:53.160 ⇒ 00:07:53.740 mitchell: Okay.
65 00:07:53.740 ⇒ 00:07:55.670 Uttam Kumaran: Basically start tracking when we need it.
66 00:07:56.220 ⇒ 00:08:21.769 mitchell: Yeah. And I’m I’m kind of like, let’s let’s track stuff as we need it. So I kind of just started putting together this like. I, I kind of lean towards this actor action object model. So like, user views, user creates account. And then we can have, like a metadata column as well, with additional data if we need it. But basically, the idea is like, we’ll have, like the event with, like the actor with the user id, and then the object
67 00:08:21.790 ⇒ 00:08:34.089 mitchell: would be like the page URL, that they’re looking at right? So there’s just kind of the structure to think about the events through and then we can add additional metadata of like referring sources utm parameters whatever if we want on that
68 00:08:34.720 ⇒ 00:08:36.280 mitchell: anyway. So that’s.
69 00:08:36.289 ⇒ 00:08:41.769 Uttam Kumaran: Where’s the? Where’s the do you guys? You have product usage data like, where is that? Right now?
70 00:08:42.330 ⇒ 00:08:43.370 mitchell: Yeah, so we put.
71 00:08:43.370 ⇒ 00:08:45.829 Uttam Kumaran: All. Is that all? Okay? Okay, yeah. Go ahead.
72 00:08:45.830 ⇒ 00:08:49.610 mitchell: Yeah, so so I have.
73 00:08:50.400 ⇒ 00:08:52.260 mitchell: Was putting together.
74 00:08:53.810 ⇒ 00:09:00.160 mitchell: They they sent me over the like the database schemas, so I just dropped them in here to kind of look at them.
75 00:09:00.160 ⇒ 00:09:00.810 Uttam Kumaran: Oh, nice!
76 00:09:01.265 ⇒ 00:09:06.279 mitchell: But yeah, basically. So we’ve got like users, and organizations.
77 00:09:07.059 ⇒ 00:09:12.770 mitchell: So usage can happen at like a user level or an organization level
78 00:09:14.050 ⇒ 00:09:24.365 mitchell: and and it’s it’s kind of broken up, because originally their product was Stack Blitz, which was like a web id. Then they have bolt now, but they’re reusing a lot of the stack Blitz,
79 00:09:24.840 ⇒ 00:09:26.500 mitchell: back end and data model.
80 00:09:26.500 ⇒ 00:09:26.830 Uttam Kumaran: Yeah.
81 00:09:26.830 ⇒ 00:09:34.750 mitchell: So but mostly from here. It’s it’s like users, subscriptions, organizations and then
82 00:09:35.550 ⇒ 00:09:44.110 mitchell: on bolt. They’ve got it’s a lot simpler, but you have like a chat. So on bolt.
83 00:09:44.450 ⇒ 00:09:49.221 mitchell: and with, like most of the Llm chat tools. If this will load, there we go.
84 00:09:49.660 ⇒ 00:09:52.590 mitchell: You basically have like a chat, and then a chat.
85 00:09:52.690 ⇒ 00:09:54.050 Uttam Kumaran: Come on!
86 00:09:54.490 ⇒ 00:09:55.510 mitchell: There we go.
87 00:09:56.080 ⇒ 00:10:08.419 mitchell: and then a chat will have chat messages. So yeah, chat with chat messages, and then there’s some other stuff of like, once you create something, you can deploy it. So we have, like.
88 00:10:08.420 ⇒ 00:10:08.900 Uttam Kumaran: Like deploy.
89 00:10:08.900 ⇒ 00:10:15.260 mitchell: But those are happening elsewhere. Yeah. So anyway, it’s it’s I don’t. There’s like nothing in there.
90 00:10:15.260 ⇒ 00:10:18.150 Uttam Kumaran: Then in terms of in terms of metadata. Every everybody
91 00:10:19.260 ⇒ 00:10:23.230 Uttam Kumaran: is, that is, everybody signed in like to use it? Or is it all.
92 00:10:23.230 ⇒ 00:10:28.943 mitchell: So if you, if you try. And essentially the landing page is
93 00:10:29.770 ⇒ 00:10:52.079 mitchell: is just this, but then you type in your prompt, and when you try and enter your 1st prompt. If you’re not logged in, then it pops a login modal. So yeah. So so everyone’s gonna be signed in. We do have segments, and we’re doing identify calls. So even if you know when 1st lands they’ll have an anonymous id, and then we’re identifying them. So that should
94 00:10:52.650 ⇒ 00:10:53.270 mitchell: yeah.
95 00:10:53.270 ⇒ 00:10:54.350 Uttam Kumaran: That’s my question.
96 00:10:56.540 ⇒ 00:11:06.620 Uttam Kumaran: Okay? I mean, look a lot of it seems really straightforward. If can you go back to that sheet? One? If you can share that with me. That would be great. But looking at, it doesn’t actually seem
97 00:11:06.730 ⇒ 00:11:07.820 Uttam Kumaran: that bad.
98 00:11:08.230 ⇒ 00:11:19.756 mitchell: Yeah, I still need to finish just making sure that we have, like, basically all the data that that I think we’ll need to run all these metrics. But yeah, more or less, that’s what we’re doing.
99 00:11:20.250 ⇒ 00:11:21.520 mitchell: so let me just.
100 00:11:21.520 ⇒ 00:11:22.280 Uttam Kumaran: Cool.
101 00:11:23.760 ⇒ 00:11:27.029 mitchell: I mean, this just seems like mainly web analytics. And then.
102 00:11:27.435 ⇒ 00:11:37.424 Uttam Kumaran: Yeah, I think, basically, once we get the 1st set of data in front of you, you can. We can probably go back and say, Okay, we need this event restructured or something like that.
103 00:11:38.440 ⇒ 00:11:52.059 Uttam Kumaran: But I mean, yeah, the reason I’m I’m I’m asking questions is because yeah, most clients come with nothing. So this is perfect like this is actually a lot of the work that we would do. Already done. So this is like ideal. And then I I kind of mentioned
104 00:11:52.677 ⇒ 00:11:58.369 Uttam Kumaran: one, I I totally think, and even I can. I can show you a little bit of like what real looks like.
105 00:11:58.370 ⇒ 00:11:59.609 mitchell: Oh, yeah, that’s just.
106 00:11:59.990 ⇒ 00:12:17.090 Uttam Kumaran: Yeah, let me let me pull that up and give me one sec. But it’s the nice thing is one. I I went to the market. I was like, I need to buy. I want to get a bi as code solution, because doing drag and drop Bi and everything, you just one business logic. It’s stuck in the tool. And then it’s like I don’t want. I just didn’t want analysts
107 00:12:17.210 ⇒ 00:12:43.928 Uttam Kumaran: sort of spending a lot of time like changing colors and shit. I was like, you need to be answering questions, so I felt like part of it was like picking a tool that was purposely a little bit limited, but also like more opinionated. And I met the guys at real. And the product is really really nice. So let me just pull up. We have an internal version of real that I’ll just show you. We have some synthetic data that we that we basically built for like a fake company. That we use for Demos
108 00:12:44.370 ⇒ 00:12:47.610 Uttam Kumaran: but I’ll just show you sort of what the ui looks like, and
109 00:12:48.220 ⇒ 00:12:52.690 Uttam Kumaran: kind of like what the what the back end looks like. A bit perfect.
110 00:12:52.960 ⇒ 00:12:59.350 mitchell: Yeah, cause the thing is like, I wanna be, I want to be very opinionated about the metrics and everything that we’re tracking.
111 00:12:59.350 ⇒ 00:12:59.820 Uttam Kumaran: Yes.
112 00:12:59.820 ⇒ 00:13:14.909 mitchell: I wanna kind of push that up to the CEO and like have him push it down to the company as a whole, and we don’t have any analysts right now, but I just don’t want people to be going in and like making their own weird metrics and analytics.
113 00:13:14.910 ⇒ 00:13:39.309 Uttam Kumaran: This is a solution for you, because basically, you define metrics as code. So at least one, you need some sort of Ci or approval process. So it’s always code. The second thing is it pushes it a little bit down from like the analyst to the Aes basically make the decisions. And it’s very lightweight. Basically, you just run it over one of your Dvt models like your final table, define the metrics and definitions, and it gives you a ton of
114 00:13:39.570 ⇒ 00:13:46.179 Uttam Kumaran: custom like you can do parameterized filters and a ton of stuff. So here’s an example. I’m just gonna show
115 00:13:48.310 ⇒ 00:13:50.069 Uttam Kumaran: you can see this right.
116 00:13:50.290 ⇒ 00:13:51.139 mitchell: Yeah, yeah.
117 00:13:51.490 ⇒ 00:13:56.000 Uttam Kumaran: Okay, so here’s like an example of a sample, like all orders. Table
118 00:13:58.400 ⇒ 00:14:24.400 Uttam Kumaran: great. So this is like the ui. This is like one sort of dashboard. Basically. What you can do immediately. Here is, you could say, cool. I want to look at like last. 3 months. I then want to compare it to the previous 3 months. It does like a stripe. It’s basically like a lot of stripe somewhere to stripe dashboard. If you’ve seen that. So you can do really nice comparisons in built right like this is a, this will be a nightmare to basically build. And the bi tool.
119 00:14:24.920 ⇒ 00:14:44.339 Uttam Kumaran: The second thing is you. For example, if I’m like, I want to dig down to target, you just click. It’s all built on TV. So all of the so the the database is in the browser. Meaning it’s all like Super. It’s like the query runs in the browser, doesn’t it? Doesn’t do a round trip to Snowflake. So you can quickly filter here
120 00:14:44.580 ⇒ 00:15:09.570 Uttam Kumaran: quickly, filter by another thing. And then the other thing you can do here is you can basically say, cool. I want to look at total revenue as a column here, and then it’ll actually show you like the percentage change between periods. The Delta. They just do a lot of like the 1st layer and second layer questions you would have they kind of solve here. So I find this to be like
121 00:15:09.690 ⇒ 00:15:25.215 Uttam Kumaran: very strong in terms of getting from 0 to one on Bi, I think you may find that your executives need like, okay, we actually want, like a very like we want, like a weekly report, dashboard or something. But again, it depends. That’s more like it depends on the people.
122 00:15:25.620 ⇒ 00:15:34.069 Uttam Kumaran: and then, to kind of show you like what this looks like behind the scenes. So basically, you have these metrics. Everything here. You can also do pivots here.
123 00:15:34.571 ⇒ 00:15:59.480 Uttam Kumaran: So you can basically drag in columns, drag in rows, build pivots, export these schedule, these, you know all the normal stuff. And it to show you kind of like what this looks like in terms of code. So this would be like a sample. This is like our basically our real project. If I go into sources. You’ll see this. All your orders. Yaml, file. It’s a select
124 00:15:59.810 ⇒ 00:16:14.473 Uttam Kumaran: from all orders. This is it for the source table? Basically, what that does is it loads it into their in browser memory. And then, if I go to models, we’re actually not doing
125 00:16:14.910 ⇒ 00:16:32.760 Uttam Kumaran: all we’re doing here is just doing a straight select. But what this allows you to do is real. You can select something from S. 3. You could select something from Snowflake. You can select something from like a Google sheet. And then they provide you with like this middle layer. So you can actually do some modeling. For the most part we do all the modeling and dbt, so we.
126 00:16:32.760 ⇒ 00:16:34.399 mitchell: Yeah, yeah, typically we.
127 00:16:34.400 ⇒ 00:16:35.420 Uttam Kumaran: History, select.
128 00:16:35.420 ⇒ 00:16:37.490 mitchell: Yeah, I like more doing the.
129 00:16:37.490 ⇒ 00:16:39.240 Uttam Kumaran: Kind of get how it could be like.
130 00:16:39.410 ⇒ 00:16:40.110 mitchell: Yeah.
131 00:16:40.270 ⇒ 00:16:42.490 Uttam Kumaran: Yeah. And then this is the real like.
132 00:16:42.700 ⇒ 00:16:50.421 Uttam Kumaran: this is where really everything is. It’s just it’s just dimensions and and measures. And so we set up like measures for all the key.
133 00:16:51.080 ⇒ 00:17:15.550 Uttam Kumaran: sort of columns. Each of these is like counts. Or you know, you’re just basically setting up metrics. You have sort of these, like a visual sort of parameters, which is like is this, can you do so? You do a percent of total on these, for example, like, you don’t want to do a percent of total with Aov. But you want to do a percent total with total revenue and then you can kind of do like, you know, different formats and stuff like that. And then
134 00:17:15.923 ⇒ 00:17:21.500 Uttam Kumaran: the nice thing is, you all can also do some. Basically, you have very easily available time ranges. So
135 00:17:21.609 ⇒ 00:17:26.740 Uttam Kumaran: some cop some people they want. They don’t want all this. They just want a couple. They want specified time zones.
136 00:17:26.960 ⇒ 00:17:39.499 Uttam Kumaran: And the nice thing is all this is as code, and then it’s just easy to scale this to the new dashboards, and on our end. What we do is basically for every model. We also just directly create a real
137 00:17:39.600 ⇒ 00:17:57.220 Uttam Kumaran: model for it. And then for my team in particular, like this is so much easier to spot check models than to do like run selects it, like to see it visually like this. You could immediately be like, oh, there’s something wrong here. So we build this anyways. And then
138 00:17:57.350 ⇒ 00:18:12.309 Uttam Kumaran: you can run this locally for free. So frankly you could just you could run it for yourself. And then the cloud version, it’s a minimum, 2, 50 a month, for, like their cheapest version, which, depending on your data volume, will probably work. So it’s like dirt, cheap.
139 00:18:12.780 ⇒ 00:18:18.290 mitchell: Yeah, that’s not very expensive at all. Yeah, like.
140 00:18:18.950 ⇒ 00:18:24.380 mitchell: yeah. So basically, we just closed around of funding
141 00:18:24.837 ⇒ 00:18:36.590 mitchell: so my, CEO is like, we’ve got money like, let’s spend it so. But I also don’t wanna just like, you know, spend money wantonly. So I think this will be a great starting point.
142 00:18:36.590 ⇒ 00:18:41.030 Uttam Kumaran: Yeah, on the. So on the Bi side, I think definitely figure out like what
143 00:18:41.200 ⇒ 00:18:50.349 Uttam Kumaran: your downstream customers needs are and then make the bi investment like, if cause there’s some people who are like I need a tableau dashboard.
144 00:18:50.620 ⇒ 00:18:54.550 Uttam Kumaran: Then, even no matter what you put in front of them, they’re gonna be like fuck this like I.
145 00:18:54.550 ⇒ 00:18:57.826 mitchell: Yeah, it’s not tablet paper that I can see.
146 00:18:58.190 ⇒ 00:19:22.389 Uttam Kumaran: It’s so subject, it’s so subjective that, like it’s such a tough game that for me, I I think about one like, how do I get my guys to work faster? And this helps. And then for people, for some leaders are very like operational focus where they’re like, I just want to see, like the period over period analysis, or quickly be able to do like 12 months, over 12 months, and then look at like the contribution between things and this tool, for that
147 00:19:22.530 ⇒ 00:19:30.424 Uttam Kumaran: is like sick. But of course they’re adding some more stuff. On terms of this, but definitely opinionated in that
148 00:19:31.220 ⇒ 00:19:34.699 Uttam Kumaran: it’s like it’s just serving with more operational use cases. But.
149 00:19:34.700 ⇒ 00:19:35.120 mitchell: Yeah.
150 00:19:35.120 ⇒ 00:19:48.380 Uttam Kumaran: Worth just doing off the rip, getting the feedback and then figuring out like whether sigma or looker or also again, like, if you’re gonna have analysts working. Then it also helps because you don’t want an analyst who’s like a tableau analyst to then come into environment where
151 00:19:48.610 ⇒ 00:19:53.260 Uttam Kumaran: they like may not have functionality. So this also probably helps with that hire as well. You know.
152 00:19:53.640 ⇒ 00:20:00.869 mitchell: Yeah, yeah. I mean, we’ve got. We’ve got the the mix panel dashboards that they’ve built out. And I think
153 00:20:01.511 ⇒ 00:20:29.294 mitchell: those will probably be like a decent approximation of like where we should start and what people want to see. But I think when we give them some of the like additional functionality. You know, the real is gonna bring like like this, drill down stuff is very nice. So I think they’ll like that, and and then we can kinda iterate a little bit from there. But I also kind of want to tell them like these are the things that we need to be looking at. So
154 00:20:29.930 ⇒ 00:20:32.349 mitchell: I’ve got a pretty good idea, those metrics of what.
155 00:20:32.350 ⇒ 00:20:38.840 Uttam Kumaran: Yeah, found, like execs like love, the people that we put this in front of one like the execs love that it’s like quick.
156 00:20:39.170 ⇒ 00:20:39.560 mitchell: Yeah.
157 00:20:39.560 ⇒ 00:20:46.620 Uttam Kumaran: A lot of times there’s like I I couldn’t pull something or something breaks. The nice thing is it’s really hard to break this because
158 00:20:46.950 ⇒ 00:20:58.240 Uttam Kumaran: the real compiles a bunch of stuff, and we’ll yell at you in the Ci process so you can’t really push stuff that doesn’t work into here. And then, second, it’s like very snappy. So this is the kind of thing that I try to tell
159 00:20:58.470 ⇒ 00:21:00.850 Uttam Kumaran: mar our companies that we work with is like
160 00:21:00.960 ⇒ 00:21:07.950 Uttam Kumaran: self service analytics and like looker and stuff is like very hard because it’s it’s pitch to self service. But it’s very complicated.
161 00:21:08.190 ⇒ 00:21:17.620 Uttam Kumaran: Instead, people are used to seeing stuff like this, especially operators are like, I just want to see comparisons. I want to like, see all the data in front of me. Something like this usually works really well for them.
162 00:21:17.620 ⇒ 00:21:17.940 mitchell: Yeah.
163 00:21:20.520 ⇒ 00:21:42.239 mitchell: Yeah, I mean, that was, it’s funny you mentioned the like speed thing because we used preset at and like people hated it because they always just said it was low. But it’s like it’s not slow. 1st off, the analysts are making shitty queries, and that’s the problem. But like, if it’s fast they would have been like way way
164 00:21:42.360 ⇒ 00:21:45.510 mitchell: they would have been fine with it. But instead, now it was, yes.
165 00:21:45.510 ⇒ 00:21:50.189 mitchell: got a switch. So we’re switching the hash board is what they ended up doing, and they were
166 00:21:50.590 ⇒ 00:21:53.760 mitchell: just finishing up that transition when I was leaving, but.
167 00:21:54.270 ⇒ 00:22:15.400 Uttam Kumaran: But this is a problem. Dude is like for data. And this is where I talk to. People is like, this is the last mile. But this is the only thing that the end user interacts like, unless you have people querying tables which a lot of sometimes most people are just using the bi tool. It’s like 80% of the work happens before this. But then this is a, this is like 90% of the deliverable. It’s like this.
168 00:22:15.400 ⇒ 00:22:17.499 mitchell: That’s all we care about. Yeah.
169 00:22:17.500 ⇒ 00:22:29.180 Uttam Kumaran: So like a lot of see like they won’t, because I can’t sit there and be like, well, we we have like 5 train. We optimize that we have our back. We have like great naming conventions. We have stuff in schemas. We have great modeling. They’re like.
170 00:22:29.440 ⇒ 00:22:35.310 Uttam Kumaran: I don’t know what any of that is, and then they’re like, Hey, like, why is this? Why is this not adding up? Is like the question they have?
171 00:22:36.110 ⇒ 00:22:59.759 Uttam Kumaran: And that’s the tough part. But so for me, being able, I I never liked any of the bi tools on the market. These guys are fairly new over the past like 3 years. But the CEO, they worked with a lot of like ad companies that needed like real time data on like millions of data points to basically do like bidding. And they built a really great like operational dashboard for that use case and then sort of scale this out.
172 00:22:59.970 ⇒ 00:23:20.769 Uttam Kumaran: But like, I don’t know. It’s nice to. I love this tool. I mean, we and all of our clients have particularly enjoyed it over the other experience. The only thing we see is, some people are like, I want a tableau like, I want a fixed thing with, like, I want the number here, and they want to do like a design process. Then I’m like, Okay, that’s just subjective, like, you have to kind of listen sometimes. But.
173 00:23:20.950 ⇒ 00:23:31.440 mitchell: Yeah, and we may end up somewhere like that eventually. But I think this looks like a great starting point for us, and especially at the price point. It’s like you can’t really make a mistake there. So.
174 00:23:32.500 ⇒ 00:23:33.439 Uttam Kumaran: 100%.
175 00:23:33.440 ⇒ 00:23:45.959 mitchell: Yeah, cool. Okay, so what does it look like then? For us to like, get get a more official proposal. And like, have me run that through just whatever approvals I need on my side.
176 00:23:47.820 ⇒ 00:24:12.259 Uttam Kumaran: Yeah. So if the scope you sent me in that spreadsheet, you know, kind of contains everything. Then I’m gonna kind of just put that into a little document, and then basically plan for like what the timeline on that would look like. Typically, we do like minimum 3 months. That looks like at least 3 months of work, if not longer, for sure. And then, of course, like in this process as well. We would help you basically procure the like the Etl tools, snowflake and stuff.
177 00:24:12.576 ⇒ 00:24:34.460 Uttam Kumaran: I think we’ll give you options on, like, usually clients. Decide on whether they want to own that or not. I think in this case you guys would definitely probably want to own that infrastructure. But in terms of helping you facilitate that and everything like we can do that. And then usually the 1st 2 weeks is a lot of just like set up, you know, like getting everything basically getting 5 trend set up getting data in there.
178 00:24:34.800 ⇒ 00:24:36.470 mitchell: Yeah, Snowflake, and then it’s
179 00:24:36.470 ⇒ 00:24:41.489 mitchell: and I’ll probably want to go with polytomic just cause I’ve really enjoyed working with their team
180 00:24:41.720 ⇒ 00:24:53.300 mitchell: and pricing wise. I don’t feel like it’s gonna be a huge difference. So that’s the only like. But yeah, Snowflake polytomic. Dbt, that’s probably the stack that
181 00:24:54.220 ⇒ 00:24:55.320 mitchell: I want to go with.
182 00:24:55.320 ⇒ 00:24:55.980 Uttam Kumaran: Yeah.
183 00:24:57.090 ⇒ 00:25:07.251 mitchell: So. But yeah, okay, cool. Yeah, that makes total sense. So I’ll yeah. And I mean, obviously, like, we’re we, we’re taking the next 2 weeks off over here, so
184 00:25:07.930 ⇒ 00:25:28.050 mitchell: wouldn’t really move on anything till January, anyway. So like no rush on that. And I’m gonna try and finish or less locking in that document that I shared with you. So you know, if if you’re looking at it, feel free to like give some padding, because there’s gonna be a little bit more work there than is like fully
185 00:25:29.640 ⇒ 00:25:34.990 mitchell: fully on the on that sheet. But yeah, and then we can just kind of go.
186 00:25:35.269 ⇒ 00:25:45.900 Uttam Kumaran: We’ve been wanting to use. We’ve been wanting to use polytomic. By the way, I met with the CEO. A great guy we just haven’t had a client that’s like opinion enough for us to use it.
187 00:25:46.180 ⇒ 00:25:46.900 mitchell: It’s like, if you already
188 00:25:46.900 ⇒ 00:25:52.960 mitchell: we started using portable for some folks because their pricing was really good. But I think polytomic is the best. Yeah.
189 00:25:52.960 ⇒ 00:26:15.881 mitchell: yeah, yeah, I they were just really great to work with when we ever sell. And so, yeah, I just wanna kind of stick with them. And I like, yeah, like the CEO over there. So cool, alright man that sounds great. I don’t really feel like I need to shop this around. So like, send me the proposal. I’ll do what I need to do. Like
190 00:26:16.500 ⇒ 00:26:21.910 mitchell: approved internally. And yeah, hopefully, we can just kinda like.
191 00:26:22.640 ⇒ 00:26:26.779 mitchell: get something done early January, and then get.
192 00:26:26.780 ⇒ 00:26:27.470 Uttam Kumaran: But in January.
193 00:26:27.470 ⇒ 00:26:29.350 mitchell: And like start moving.
194 00:26:31.510 ⇒ 00:26:49.620 Uttam Kumaran: Okay. Cool dude I’ve been. I’m on a side note. Dude. It’s great that you’re a bull, I mean, I saw Bull blow up on Twitter like over the last, like 2 months. And I was. I was literally sitting. I’m at a coffee shop here. I was sitting here using bowl with a friend, showing him that it he wanted to build some like quick app and like dude use bowl. So that’s that’s dope.
195 00:26:50.040 ⇒ 00:26:59.250 mitchell: Yeah, no, that’s sweet. They yeah. They. So I invested a couple of years ago when they were doing their original product. And then they kind of pivoted, and
196 00:26:59.400 ⇒ 00:27:05.879 mitchell: it was mostly just supposed to be like a proof of concept of like. Here’s what you can do with the technology we’ve built. And then.
197 00:27:05.880 ⇒ 00:27:06.520 Uttam Kumaran: Wow!
198 00:27:06.650 ⇒ 00:27:12.689 mitchell: It like blew up. So they went from 0 to 26 million arr in 3 months. It’s just like.
199 00:27:13.159 ⇒ 00:27:13.549 Uttam Kumaran: Yeah.
200 00:27:13.550 ⇒ 00:27:14.190 mitchell: Crazy, gross.
201 00:27:14.190 ⇒ 00:27:14.610 Uttam Kumaran: Crazy.
202 00:27:15.004 ⇒ 00:27:26.430 mitchell: Yeah. So anyway. So the CEO, hey, like, are you interested? Like, here’s the situation. We’re gonna raise around like, let’s get you in before that. And
203 00:27:26.660 ⇒ 00:27:30.210 mitchell: so I was like, Okay, yeah, let’s do it because I wasn’t planning on leaving Brazil for
204 00:27:30.380 ⇒ 00:27:33.030 mitchell: probably till like March or April next year. But
205 00:27:33.810 ⇒ 00:27:36.909 mitchell: yeah, yeah, timing worked out. So we just went for it.
206 00:27:39.030 ⇒ 00:27:39.690 Uttam Kumaran: Dope.
207 00:27:40.400 ⇒ 00:27:45.390 Uttam Kumaran: Yeah, it’s a great product. I mean, it’s kind of like how Chat Gpt started right? They’re just like we need some sort of
208 00:27:45.520 ⇒ 00:27:48.709 Uttam Kumaran: consumer facing use case for the platform. But that’s dope.
209 00:27:48.890 ⇒ 00:28:03.309 mitchell: Yeah, yeah, exactly. Okay. Cool. Well, I will talk to you later, then. And yeah, I’ll just look, look for if you have any questions while you’re doing stuff. Just let me know proposal wise, and I can answer any questions as well.
210 00:28:05.620 ⇒ 00:28:06.590 Uttam Kumaran: Okay. Perfect.
211 00:28:06.820 ⇒ 00:28:07.960 Uttam Kumaran: Alright, thanks. Mitch.
212 00:28:07.960 ⇒ 00:28:09.779 mitchell: Talk to you later, man, bye.