Meeting Title: Brainforge x Urbanstems: Project Review! Date: 2025-09-22 Meeting participants: Zack Gibbs, Justin Breshears, Uttam Kumaran, Amber Lin
WEBVTT
1 00:00:10.170 ⇒ 00:00:10.870 Uttam Kumaran: Hey, Zach.
2 00:00:12.310 ⇒ 00:00:13.889 Zack Gibbs: Hey, how’s it going?
3 00:00:13.890 ⇒ 00:00:15.239 Uttam Kumaran: How are you?
4 00:00:15.930 ⇒ 00:00:21.570 Zack Gibbs: Doing good. I just, we’ve got this big project going on right now where…
5 00:00:21.800 ⇒ 00:00:25.590 Zack Gibbs: We’re adding, wine sales to the site, and the.
6 00:00:25.590 ⇒ 00:00:27.860 Uttam Kumaran: Oh, okay, interesting.
7 00:00:28.580 ⇒ 00:00:30.150 Zack Gibbs: Under that we’re working with is…
8 00:00:30.400 ⇒ 00:00:40.249 Zack Gibbs: A dumpster fire of a group, and we’re at the… we’re near the end phases of development, and we’re just having a bunch of problems with, like.
9 00:00:40.520 ⇒ 00:00:45.760 Zack Gibbs: Explicit rules that we need to follow, because the problem that we have is that we want to
10 00:00:46.390 ⇒ 00:00:56.079 Zack Gibbs: You know, people are buying, selecting a date, all of our… all of our gifting is very date-specific, except for our subscriptions, and
11 00:00:56.570 ⇒ 00:01:06.620 Zack Gibbs: like, getting explicit rules for the delivery of the wine SKUs at the same day of our SKUs is… because they’re being fulfilled separately, being.
12 00:01:06.620 ⇒ 00:01:07.300 Uttam Kumaran: Yeah.
13 00:01:08.070 ⇒ 00:01:10.370 Zack Gibbs: Has been really challenging.
14 00:01:10.540 ⇒ 00:01:15.469 Uttam Kumaran: Are you going through, like, a distributor for all that? Or, like, I don’t even know how alcohol sales online…
15 00:01:15.830 ⇒ 00:01:17.220 Uttam Kumaran: Even worse.
16 00:01:17.220 ⇒ 00:01:22.940 Zack Gibbs: It’s complicated. Yeah, the distributor that we’re… the third party that we’re going through is Drink… Drinks.
17 00:01:23.100 ⇒ 00:01:28.279 Zack Gibbs: So if you look up, if you look up drinks.com, you’ll, you’ll find them.
18 00:01:28.690 ⇒ 00:01:33.519 Zack Gibbs: But they… I would not recommend them. I would not recommend working with them, ever.
19 00:01:33.610 ⇒ 00:01:46.129 Uttam Kumaran: Okay, yeah, I’ll make sure we don’t buy our alcohol through them. No, but that’s interesting. I don’t think we’ve worked with anyone. We’ve talked to some people in beverage.
20 00:01:46.290 ⇒ 00:01:50.780 Uttam Kumaran: an alcohol beverage, but we don’t… we never had a client in alcohol beverage, so…
21 00:01:51.600 ⇒ 00:01:52.340 Zack Gibbs: Yeah.
22 00:01:52.600 ⇒ 00:02:03.309 Zack Gibbs: I mean, the idea is… it’s a cool idea, it’s just implementing that idea with a vendor who’s a little shaky, and they’re like, API is really terrible, is not… is not… not the easiest thing to do.
23 00:02:05.010 ⇒ 00:02:05.890 Zack Gibbs: So…
24 00:02:05.890 ⇒ 00:02:17.339 Uttam Kumaran: Yeah, I mean, it’s, I feel like the story of… story of this project has been a lot of that, so… but we’ll end up… hopefully they’ll have an API, we can get their data in as well, so…
25 00:02:17.340 ⇒ 00:02:18.060 Zack Gibbs: Yeah.
26 00:02:18.250 ⇒ 00:02:20.240 Uttam Kumaran: Nice.
27 00:02:20.370 ⇒ 00:02:39.309 Uttam Kumaran: Yeah, so I wanted to sort of just maybe set the stage for this call, and kind of also introduce Justin, who’s joining our team. So one of the things that our delivery team, and I describe it as a big team, but really it’s just Amber, Justin, one other person, and myself, is sort of thinking a lot about how do we
28 00:02:39.310 ⇒ 00:02:42.880 Uttam Kumaran: You know, as external folks, just continue to keep
29 00:02:42.880 ⇒ 00:02:52.149 Uttam Kumaran: You, our core stakeholder, like, in the loop on how things are going day-to-day. I think you’re seeing a lot in Slack, and I think you probably have a little bit of a sense from
30 00:02:52.150 ⇒ 00:03:07.939 Uttam Kumaran: from folks internally how things are going, but we want to give you sort of an overview at least once a month on all of our workstream progress. I know, ideally, this would be best to have everything tracked in linear and have things, but, you know, I think this project in particular
31 00:03:07.940 ⇒ 00:03:28.750 Uttam Kumaran: there’s been a lot of ad hoc that we’ve been really crushing through, but I just want to highlight the work streams that we agreed upon initially, talk about progress, and then I know, you know, our existing contract is going till November, so starting to set the stage of, okay, what are things that may carry over, what are next priorities?
32 00:03:28.750 ⇒ 00:03:48.370 Uttam Kumaran: And then, yeah, I just wanted to introduce Justin. Justin also comes from, an agency background, previously at a company called Kalint. He’s coming on to project manage, several projects, and just included in here as we’re, like, starting to do this process with clients so he can observe. Maybe, Justin, if you just want to say hello,
33 00:03:48.570 ⇒ 00:03:56.209 Uttam Kumaran: I haven’t given a brief background on Zach, but Zach Leeds is our core stakeholder at Red STEMS. But yeah, if you just want to say, say hi.
34 00:03:56.210 ⇒ 00:03:59.270 Justin Breshears: Yeah, I appreciate it. Thanks, Uten. It’s… it’s…
35 00:03:59.270 ⇒ 00:04:20.709 Justin Breshears: Nice to be here and meet you, Zach. Thanks for letting me be a fly on the wall. Like you said, I’m coming in, you know, new to Brainforge here, so, just kind of observing processes at the moment right here, but you’ll probably hear from me or see me around as I get more integrated with everything, so nice to meet you. I’m not too far away from Utam in Texas as well, so central time zone for me.
36 00:04:20.709 ⇒ 00:04:23.240 Justin Breshears: What about you? Where are you located?
37 00:04:24.210 ⇒ 00:04:30.849 Zack Gibbs: I’m in… I’m in Denver, so I’m in Mountain… Mountain Time. Most of the team is in…
38 00:04:31.290 ⇒ 00:04:44.349 Zack Gibbs: either Central or, Eastern Time. The biggest cohort of folks that we have are in Eastern Time, for sure. We have an office in New York. We used to have an office in DC. That was, like, where the company started, but it kind of has been shifted over to New York, so…
39 00:04:44.560 ⇒ 00:04:47.599 Zack Gibbs: Mostly Eastern Time is kind of what we operate out of.
40 00:04:48.130 ⇒ 00:04:57.830 Justin Breshears: I’m, I’m jealous. Colorado is my escape place, when it gets too hot here in Texas. I like to go up there, so… jealous of your weather and your mountains.
41 00:04:57.830 ⇒ 00:05:00.060 Zack Gibbs: Lots of Texans. Lots of Texans.
42 00:05:00.710 ⇒ 00:05:03.749 Zack Gibbs: I’ll visit Colorado in the wintertime, in particular.
43 00:05:03.840 ⇒ 00:05:06.319 Uttam Kumaran: Yeah, we’re gonna always, I always, always…
44 00:05:06.350 ⇒ 00:05:10.230 Zack Gibbs: I always hear the… I always hear the draw on the… on the slopes,
45 00:05:10.570 ⇒ 00:05:12.830 Zack Gibbs: You’re from Texas, aren’t you?
46 00:05:13.430 ⇒ 00:05:14.810 Justin Breshears: Pretty easy to spot us.
47 00:05:15.470 ⇒ 00:05:35.339 Uttam Kumaran: Yeah, we’re gonna be visiting Boulder, I think, like, early next… early November, but again, just cause, like, my girlfriend and I both have loved, like, going to Fort Collins, going to Boulder, like, these small towns, and hanging out, and it’s cold. It’s here, it’s… I thought it was getting colder, it was 95, it was, like, 95 still, and…
48 00:05:35.830 ⇒ 00:05:38.640 Uttam Kumaran: Yeah, just not changing much, but…
49 00:05:38.640 ⇒ 00:05:39.460 Zack Gibbs: Yeah.
50 00:05:41.410 ⇒ 00:06:00.039 Uttam Kumaran: Cool, so let me, let me just walk through this document that we set up, and again, would love to hear, sort of, your feedback. I think, Zach, you’ve always been helpful at giving us feedback on how… on our processes, so this is sort of one of the first times we’re doing this. So I just wanted to kind of give a sense of, like, overall,
51 00:06:00.560 ⇒ 00:06:18.070 Uttam Kumaran: where the project is, and I want to talk kind of a high level of a couple, like, wins that we’ve seen. Overall, I think this has been… we sort of came in into a lot of systems and processes, but I think one of the things that’s been helpful for me is to hear your voice in the back of my head, where it’s like.
52 00:06:18.070 ⇒ 00:06:35.880 Uttam Kumaran: just break things and get things, like, into the new phase. I think that has been the theme of a lot of… especially the last 4 to 6 weeks. But overall, we started the project on a big undertaking of getting, you know, the new NetSuite’s pipeline set up.
53 00:06:35.880 ⇒ 00:06:47.039 Uttam Kumaran: and doing a full audit of everything in ETL, in Redshift, and in Looker. So, I think all of that work really enabled us to start making decisions.
54 00:06:47.040 ⇒ 00:07:05.080 Uttam Kumaran: On cutting things, and really reducing the decision fatigue that I think a lot of people are having when they’re using reports, as well as starting to actually get code into the system that is working, that is tested. And so, I think we’ve done a really great job at
55 00:07:05.220 ⇒ 00:07:22.120 Uttam Kumaran: You know, getting that NetSuite stuff set up and getting it piping into inventory. And then that kind of goes to the next theme, which is, like, inventory mart. I think we’ve done a great job at getting that to a better point. I think that is currently our most mature new mart that we’ve developed.
56 00:07:22.120 ⇒ 00:07:29.609 Uttam Kumaran: And we’re starting to move on to, in parallel, working through the revenue mart. I think through this whole process, we’ve…
57 00:07:29.710 ⇒ 00:07:34.219 Uttam Kumaran: we got some things right, and then we’ve underestimated some things. I think one is…
58 00:07:34.500 ⇒ 00:07:48.590 Uttam Kumaran: There is a lot of cleanup that we’re still continuing to do, and I think in addition to that, one of the things that we were hoping for is that we’d be able to sort of slow a little bit down on
59 00:07:48.590 ⇒ 00:07:56.730 Uttam Kumaran: new iterations, and things that are in flight versus just focusing on, you know, new marts, and I think that’s been a challenge.
60 00:07:56.740 ⇒ 00:08:12.919 Uttam Kumaran: The way we’ve tackled that is, one, trying to make sure that the development process for anyone that develops on data is pushing code that works into the platform. And so a lot of what we’ve done to enable Emily is having clear guidelines on how we push code, having review processes.
61 00:08:12.920 ⇒ 00:08:20.210 Uttam Kumaran: enabling her to use cursor, creating dbt staging jobs, and starting to clean up the way we actually test
62 00:08:20.220 ⇒ 00:08:31.860 Uttam Kumaran: So that’s both prevented a lot of issues, but also naturally caused us to change the way we develop on the infrastructure. Before, we were just pushing code straight to master.
63 00:08:31.870 ⇒ 00:08:44.449 Uttam Kumaran: just, like, randomly. So now we are doing our best to not do that, but of course, that is a transition in the way we work, and has certainly caused some friction, but I think
64 00:08:44.480 ⇒ 00:08:51.169 Uttam Kumaran: everybody will agree that I think we’re in a better spot there. And then the last piece, I think we’re still not yet
65 00:08:51.340 ⇒ 00:09:11.159 Uttam Kumaran: Two is a lot of significant looker cleanup. I think we’ve identified a lot of that work, so there is some backlog of work that is sitting into removing those dashboards, but I think a significant portion of unlock is going to happen when these marts actually come in to replace existing Explorers.
66 00:09:11.160 ⇒ 00:09:20.789 Uttam Kumaran: However, throughout the entire time, I’ve been chipping away at, like, cleaning up and making Looker a little bit better, which is cleaning up some of our connection details.
67 00:09:20.820 ⇒ 00:09:35.100 Uttam Kumaran: cleaning up, like, random errors or bad scheduled reports, things like that. And especially in the last 3-4 weeks, I think we’ve improved a lot of how Looker talks to the warehouse, the query times,
68 00:09:35.420 ⇒ 00:09:55.149 Uttam Kumaran: And made a lot of progress there. So kind of talking about work streams in progress, I think we had these roughly four things that we were sort of looking towards accomplishing. I think on the dbt modeling side, I think we’ve made really great progress, on… on inventory, and we’re moving on to revenue.
69 00:09:55.370 ⇒ 00:10:02.179 Uttam Kumaran: What I’ll talk a little bit about revenue is I think the timeline for revenue really got shot a little bit just based on the last
70 00:10:02.320 ⇒ 00:10:15.150 Uttam Kumaran: like, 4 weeks of infra work that we had to do. And I can talk a little bit about what are the wins there. One is we’ve made significant progress in reducing the job refresh times. We had jobs previously that were running
71 00:10:15.150 ⇒ 00:10:24.429 Uttam Kumaran: typically anywhere from 1 to 4 hours. Now, there’s no job that runs more than an hour, and core jobs are running less than 30 minutes.
72 00:10:24.430 ⇒ 00:10:35.360 Uttam Kumaran: That was our primary objective, and what I talked to Emily about, and Demolade, is there’s no way for us to mitigate errors if things are taking 2 or 3 hours to run entirely.
73 00:10:35.360 ⇒ 00:10:50.239 Uttam Kumaran: So we’ve cut a lot of that. That goes directly to one. As errors continue to happen, we can fix them faster. So as you can see, we’re much more responsive in Slack, and actually errors can get fixed within a few hours, typically, if not within the hour.
74 00:10:50.240 ⇒ 00:11:01.480 Uttam Kumaran: Second, again, we’re getting NetSuite in almost every, like, 10-15 minutes, so that data is getting more closer to real time for decision making.
75 00:11:01.520 ⇒ 00:11:02.759 Uttam Kumaran: Yeah, go ahead.
76 00:11:03.180 ⇒ 00:11:19.800 Zack Gibbs: I was gonna ask, I think that’s one of the things I’ve asked Emily for is more of, like, the specifics there of, like, let’s celebrate the wins and expose that to the leadership team of, like, here are some of the things that have been updated and changed. So, from a, you know, from a model run perspective, like.
77 00:11:19.800 ⇒ 00:11:26.629 Zack Gibbs: What’s the… doesn’t have to be, like, model by model, but the summary of here, you know, there were these 10 core models, and they…
78 00:11:26.630 ⇒ 00:11:26.960 Uttam Kumaran: Yeah, fair.
79 00:11:26.960 ⇒ 00:11:42.610 Zack Gibbs: were, you know, running for, whatever, 82 minutes on average, or, you know, maybe it’s model by model, and then now these are running on X, and that’s a X percentage improvement. Like, those are the types of things that I think
80 00:11:43.090 ⇒ 00:11:52.489 Zack Gibbs: would be better to communicate, because internally, it’s, it’s more of, like, it’s invisible to many, many folks, and so I think that…
81 00:11:52.690 ⇒ 00:11:59.699 Zack Gibbs: is a piece of feedback of how do we bubble that up in a more explicit way?
82 00:12:01.160 ⇒ 00:12:07.209 Zack Gibbs: And then I think the other piece of it is, okay, the inventory data mart.
83 00:12:07.530 ⇒ 00:12:08.230 Zack Gibbs: Piked.
84 00:12:08.750 ⇒ 00:12:12.450 Zack Gibbs: the conclusion… I know there’s still some work being done there, but…
85 00:12:12.660 ⇒ 00:12:27.919 Zack Gibbs: like, there should have been some line in the sand of, like, this is now done, and now we’re building on top of it. And, like, you know, showing that to the broader team, I feel like, was another, like, miss of, you know, communicating outward.
86 00:12:27.970 ⇒ 00:12:32.810 Zack Gibbs: And I’ve been asking Emily to kind of help with that, but I think she’s also been very…
87 00:12:33.180 ⇒ 00:12:35.450 Zack Gibbs: in the weeds, tactically.
88 00:12:35.450 ⇒ 00:12:47.689 Uttam Kumaran: Yeah. No, I think that’s certainly something we can take on. I think we presented a little bit of that, Amber, last time we sent an update, but I think presenting a cohesive update of all the core wins to date
89 00:12:47.840 ⇒ 00:12:57.629 Uttam Kumaran: Would be really, really helpful. We have a lot of significant wins on the dbt and the infra side, and in the last 2 weeks, a lot of wins on the Redshift side.
90 00:12:57.630 ⇒ 00:13:12.820 Uttam Kumaran: Some of these I was hoping to make a little bit slowly, but we got pushed to kind of do a lot, and we changed a lot. So things are running much faster, we’re seeing quite a bit less errors, and we have the data to sort of show that. And I think we’ve… we’ve…
91 00:13:12.960 ⇒ 00:13:24.630 Uttam Kumaran: done a lot of the fixes that I discussed with Alex originally on doing about how things are accessing Redshift, and those are now… will help us, kind of, like, forever while you use that cluster. So I think we could certainly highlight that.
92 00:13:24.920 ⇒ 00:13:29.709 Uttam Kumaran: Similarly, on the inventory mark, I think you’re right, is that there was never, like, a fixed
93 00:13:30.360 ⇒ 00:13:46.860 Uttam Kumaran: cut off, because we’ve moved most of it over, and a lot of things are not powered by that. There are still some things lingering that were just deprioed, because they weren’t, yeah, for one reason or another, but I do think that we should celebrate some, like.
94 00:13:46.920 ⇒ 00:13:58.800 Uttam Kumaran: some win around all the work there, because that is all now powering the old models. To give you a sense of the way we did it is there are these, like, almost customer-facing data models that are now
95 00:13:58.800 ⇒ 00:14:08.809 Uttam Kumaran: powered by the new inventory mark. So those haven’t necessarily changed, which is something to do. For example, Tableau items, except some of these, like, really large tables we want to break up.
96 00:14:08.810 ⇒ 00:14:25.820 Uttam Kumaran: But the easiest way was just to have those now powered from the new inventory mark versus existing. So, you know, the customer can now see, you know, faster update times, we’re now able to debug faster, and all that data is coming in from their new sources. So I think that’s something certainly we can highlight.
97 00:14:27.130 ⇒ 00:14:34.100 Uttam Kumaran: And I know that that’s been a huge amount of work. I think the main thing also, and this is where, again.
98 00:14:34.160 ⇒ 00:14:51.539 Uttam Kumaran: it’s… we may have been able to achieve these outcomes without all this governance and develop… sort of developer ecosystem work that we did, but we’ve made huge strides in… in having dbt staging, so we can actually stage changes and test them on actually how we make improvements.
99 00:14:51.540 ⇒ 00:15:05.330 Uttam Kumaran: And also leveraging branching logic, and I think there’s a lot of wins there on… any, now, data team member that wants to contribute is now have a clear path to do that. Before, it was really, really tough.
100 00:15:05.450 ⇒ 00:15:24.960 Uttam Kumaran: And we shot ourself in the foot, like, every week on… on issues. So, that is something also, Amber, one thing that we talked about reporting on as a delivery team is data issues per week, basically. So that’s something I know we’ve been tracking in linear, and we’ve been keeping track of the amount of issues that have come up.
101 00:15:24.960 ⇒ 00:15:28.530 Uttam Kumaran: And I know it’s commonly for our team, it may… we may just…
102 00:15:28.560 ⇒ 00:15:36.929 Uttam Kumaran: we get in the limelight if there is a problem, and otherwise we’re sort of in the background, so I do think you’re right in that I want to push up the wins that we’ve seen.
103 00:15:37.040 ⇒ 00:15:39.260 Uttam Kumaran: And really, I think that
104 00:15:39.310 ⇒ 00:15:51.649 Uttam Kumaran: what I want to hope to do is… the looker changes are going to be the things that people see very, very clearly in their day-to-day. But also kind of the riskiest part of this project.
105 00:15:51.650 ⇒ 00:16:05.179 Uttam Kumaran: And again, and, like, deprecating some of that stuff, switching to new marts, and so that is what I think the next, you know, two months are mainly going to be, is getting the inventory mart and the revenue mart into people’s hands in Looker.
106 00:16:05.200 ⇒ 00:16:08.900 Uttam Kumaran: So yeah, happy to… happy to flag that and get you something to share.
107 00:16:09.780 ⇒ 00:16:10.640 Zack Gibbs: Yeah.
108 00:16:12.720 ⇒ 00:16:19.129 Zack Gibbs: I think the other thing that, you know, is a goal of mine as a part of this was making sure that
109 00:16:19.250 ⇒ 00:16:28.679 Zack Gibbs: we built up and leveled up some of our own analysts internally, and I… so, my perspective is that, you know, Emily… Emily…
110 00:16:29.000 ⇒ 00:16:35.039 Zack Gibbs: she didn’t have, like, the background of a BI developer, right? She was put into that role,
111 00:16:35.130 ⇒ 00:16:50.979 Zack Gibbs: And she wanted that role, so I knew that there was going to be some, like, general challenges there, her getting up to speed. But there are a couple other folks internally, like PK is a good example of somebody that we wanted to be deeply involved in this, so that way he could be, you know.
112 00:16:50.980 ⇒ 00:16:59.749 Zack Gibbs: you could understand, you know, from, you know, from source to ETL to, you know, dashboards and,
113 00:17:00.790 ⇒ 00:17:10.139 Zack Gibbs: do you guys feel like we have done a good enough job there of making sure that there are folks that are coming alongside Emily?
114 00:17:10.890 ⇒ 00:17:12.260 Zack Gibbs: In this process.
115 00:17:13.240 ⇒ 00:17:21.110 Uttam Kumaran: Yeah, I think we could do a better job. I think the process that we’ve been running so far is that we meet… we’ve met with the analysts and sort of
116 00:17:21.329 ⇒ 00:17:32.989 Uttam Kumaran: shown them what’s going on, but we have not taken as much of a, like, like, Emily joins our stand-ups, and we talk every day, right? So we’ve not done that with any other analyst.
117 00:17:34.000 ⇒ 00:17:39.820 Uttam Kumaran: I think there’s been a couple of reasons. One, like, in trying to get Perry’s time and trying to get some time, people are busy.
118 00:17:39.950 ⇒ 00:17:59.360 Uttam Kumaran: Second is, I think it’s just gonna require quite a bit of, like, it’s gonna require a bear hug of getting them into the process and getting them to develop and use stuff in dbt and in Looker. So, my kind of alternative is instead of focusing on everybody, maybe we do pick…
119 00:17:59.390 ⇒ 00:18:23.450 Uttam Kumaran: one person a month, or we try to ramp them up. And again, my KPIs for that is I want to see them… because we get requests from those folks, right? So if they have a request for a new data model, or they can take on moving some of their PDTs into DBT, and they can be in the warehouse asking questions, like, that would be… those would be my KPIs. And additionally, we start to see that maybe they take on some tickets
120 00:18:23.570 ⇒ 00:18:37.889 Uttam Kumaran: You know, over time. But if you have one person that we can focus on, I’m really open to that. And we can talk about how that works, but ideally they can join our stand-ups, or we somehow mold around them and work to enable them.
121 00:18:38.280 ⇒ 00:18:55.840 Zack Gibbs: Yeah, I mean, I think the person that has the highest, you know, potential aptitude is PK. I was hoping that Perry would be a part of that, because ultimately, if I look at the backlog of, like, general requests, many are coming from her, but how, you know, her and the forecasting team in general,
122 00:18:55.840 ⇒ 00:19:02.489 Zack Gibbs: I don’t think that there’s as much aptitude there to, like, learn, whereas PK has more of that.
123 00:19:03.220 ⇒ 00:19:12.740 Uttam Kumaran: Yeah, totally because, like, I think… I also feel the same way. I think, one, it’s availability, and then second, I think it’s, like, who’s interested in fishing for themselves?
124 00:19:12.860 ⇒ 00:19:16.800 Uttam Kumaran: And, like, finding those people to enable,
125 00:19:16.840 ⇒ 00:19:32.349 Uttam Kumaran: like, I think… I think Emily, we’ve… we’ve developed Emily into much better of a dbt… of a dbt and BI developer. Like, I think overall her skill set has moved from just pushing things into Looker, or pushing things into dbt without any sort of testing, or, like.
126 00:19:32.350 ⇒ 00:19:50.100 Uttam Kumaran: you know, review process. So I think she’s gone much more effective. Additionally, we’ve enabled her with our best practices on using cursor, how to model things effectively, how to test things locally. So, if we can find one more person to do that, I think that they can gain a lot.
127 00:19:50.260 ⇒ 00:20:03.660 Uttam Kumaran: So if it’s PK, then I think, Amber, we should discuss how we start to loop him in, and as long as we can communicate to him, and he’s open to it, I think he’d be… he could definitely start taking on some stuff and start to learn from us.
128 00:20:04.830 ⇒ 00:20:21.789 Amber Lin: Yeah, totally. We currently have him in all our working sessions and verifying our work, so he is involved in our processes, but not as involved in our daily cadences as Emily. But now that we have, say, Monday, Wednesday, Friday stand-ups, and each are just 15 minutes, I think it’s good to
129 00:20:21.790 ⇒ 00:20:25.800 Amber Lin: To ask if he’s available, and we can always adjust the time so that he can join.
130 00:20:27.100 ⇒ 00:20:44.390 Uttam Kumaran: Yeah, and then I want to see how, like, we can… again, part of it is getting people comfortable with pushing models, getting them comfortable with actually developing in our new ecosystem. Before, it would have been very, very challenging. It would have been impossible for you to actually do something and then, like, understand whether it was…
131 00:20:44.390 ⇒ 00:20:51.539 Uttam Kumaran: impacted and get reviews, so I think it’s a nice scenario right now for him to get involved. So let’s… let’s understand, Amber, what
132 00:20:51.600 ⇒ 00:21:06.389 Uttam Kumaran: tickets are coming from him, or what is on his plate, and what are things that he can start… I can… we can give guidance on what are things that he can fish for, and then we can guide him onto taking those on. And if he’s open to join our stand-ups, I’m… I’m happy to do that, and if we need…
133 00:21:06.390 ⇒ 00:21:12.290 Uttam Kumaran: I’m open to also booking one-on-one sessions with him to… to make sure that he can… he can ship those, so…
134 00:21:14.060 ⇒ 00:21:20.250 Zack Gibbs: Yeah, and my ideal state is that, you know, he would be involved more, and that he would be given
135 00:21:20.390 ⇒ 00:21:33.180 Zack Gibbs: you know, have direction internally from Emily. Like, Emily could help, you know, help give him some direction, but that it’d be in tandem, Emily and Emily and him working in tandem with you guys supporting.
136 00:21:33.770 ⇒ 00:21:34.410 Uttam Kumaran: Okay.
137 00:21:35.230 ⇒ 00:21:36.160 Uttam Kumaran: Okay, great.
138 00:21:36.650 ⇒ 00:21:43.529 Uttam Kumaran: So yeah, probably the biggest thing I wanted to call out is, like, I think a lot of the infrastructure changes we’ve made in the last few weeks
139 00:21:43.580 ⇒ 00:21:58.330 Uttam Kumaran: We change a lot in Redship, we change a lot in DBT, change a lot in the job side. Our things that we should call out wins on impacted the roadmap on revenue a little bit. I think, like, we just lost some momentum there, given all the issues.
140 00:21:58.520 ⇒ 00:22:04.299 Uttam Kumaran: But overall, like, I feel that things are in a lot better spot.
141 00:22:04.340 ⇒ 00:22:16.110 Uttam Kumaran: kind of the… one of the big, things I would love to run by you is we’ve been using Metaplane pretty heavily to understand job failures, to also start to implement monitors on
142 00:22:16.110 ⇒ 00:22:33.019 Uttam Kumaran: key values, to give you the skinny on that, is jobs may succeed, but, like, if the values are not accurate, we’re gonna get something like inventory is zero, or these look way off. Those are things that are, like, pretty tedious if we were to write tests
143 00:22:33.030 ⇒ 00:22:35.670 Uttam Kumaran: For every single one of those, and using a tool like…
144 00:22:35.780 ⇒ 00:22:46.569 Uttam Kumaran: Metaplane has been really helpful. We basically set standard deviation bands on the most high priority metrics and the most high priority models, and we’re alerted.
145 00:22:46.570 ⇒ 00:23:00.560 Uttam Kumaran: I would say most of the things now are, like, self-induced issues, where, like, we push something and it doesn’t work, and so we’re mitigating that, and then now we want to basically understand, those standard deviations. So the…
146 00:23:00.560 ⇒ 00:23:07.559 Uttam Kumaran: Cost for Metaplane is about $500 a month. If you sign for, like, month-to-month. If you sign longer, it goes down.
147 00:23:07.740 ⇒ 00:23:19.320 Uttam Kumaran: there… I’m happy to give a little bit of a spike on, like, other observability tools, but it is kind of the easiest and cheapest one to sort of get started with. There is a high…
148 00:23:19.440 ⇒ 00:23:32.789 Uttam Kumaran: noise-to-signal ratio right now, that is just something that for us to continue to… to, like, chip away at and make sure that the alerts that are coming are all P0 or P1. That is just…
149 00:23:33.330 ⇒ 00:23:48.630 Uttam Kumaran: It’s just something we have to keep spending time on, so… but, like, that would be my recommendation, is for us to leverage a tool like that. They’re… they’re under the Datadog family now, but I don’t think you guys use Datadog, so…
150 00:23:48.880 ⇒ 00:23:50.479 Zack Gibbs: Yeah, we do use Datadog.
151 00:23:50.480 ⇒ 00:23:51.539 Uttam Kumaran: Oh, really? Okay.
152 00:23:51.780 ⇒ 00:24:01.570 Zack Gibbs: Yeah, so Datadog is for our, you know, for logging, alerting on the core, you know, core website experience, is who we’ve used for a long time.
153 00:24:01.970 ⇒ 00:24:04.200 Uttam Kumaran: So they’re… they would come under that…
154 00:24:04.530 ⇒ 00:24:11.530 Uttam Kumaran: I mean, last I talked to them, after they got acquired, they would be coming under that paper, or, like, somehow get associated
155 00:24:11.660 ⇒ 00:24:13.620 Uttam Kumaran: With that, since you’re already a customer.
156 00:24:13.740 ⇒ 00:24:20.050 Uttam Kumaran: I can try to… if you have a contact there, I can go message them and see who’s the contact and try to get something…
157 00:24:21.460 ⇒ 00:24:22.800 Uttam Kumaran: In front of you.
158 00:24:22.800 ⇒ 00:24:26.040 Zack Gibbs: What are… what are some of the other Zervilly platform
159 00:24:26.190 ⇒ 00:24:28.040 Zack Gibbs: That are… that are out there.
160 00:24:28.330 ⇒ 00:24:48.139 Uttam Kumaran: Yeah, so, I can send a couple over, like, a tool like, Atlin, is one, atlin.com, sort of just, like, usually called, yeah, like, Data Observability Layer or Metadata Analysis, so Atland is one of them.
161 00:24:48.240 ⇒ 00:24:53.550 Uttam Kumaran: And I’ll… I’ll just send some bullies in Zoom as we’re talking.
162 00:24:54.100 ⇒ 00:25:08.990 Uttam Kumaran: to give you, like, the most expensive and the… one of the, kind of, the higher Fortune 500 sort of option, Big Eye is really well known to be a great product, but pretty pricey. Data…
163 00:25:08.990 ⇒ 00:25:14.900 Uttam Kumaran: Diff is another one by… oh, DataFold. Sorry, product is Data Diff.
164 00:25:15.510 ⇒ 00:25:17.349 Uttam Kumaran: This is,
165 00:25:20.870 ⇒ 00:25:23.690 Uttam Kumaran: Data Fold is another one.
166 00:25:24.640 ⇒ 00:25:26.049 Uttam Kumaran: And then let me get you.
167 00:25:31.090 ⇒ 00:25:32.990 Uttam Kumaran: Okay, absolutely.
168 00:25:42.760 ⇒ 00:25:50.930 Zack Gibbs: Yeah, I mean, we have, we have Datadog sales, sales rep, you know, slash CSM engaged that we can always reach out to.
169 00:25:51.310 ⇒ 00:25:54.439 Zack Gibbs: Yeah, my feedback on this is.
170 00:25:56.200 ⇒ 00:26:00.760 Uttam Kumaran: This, like, many of these platforms in data observability
171 00:26:01.130 ⇒ 00:26:10.369 Uttam Kumaran: They… this is, again, just my, like, opinion about this market of tools, is they all raise a ton of money, and they try to become products, and they try to become a platform.
172 00:26:10.400 ⇒ 00:26:29.859 Uttam Kumaran: And they try to, like, increase costs that way. We need Metaplain for a very specific reason, which is monitoring tables, monitoring updated at times, and sending alerts effectively. Tools like Big Eye, Monte Carlo, some of these, they just start layering on more and more
173 00:26:30.210 ⇒ 00:26:33.880 Uttam Kumaran: products and services that… like.
174 00:26:34.180 ⇒ 00:26:43.460 Uttam Kumaran: I don’t know, for many of our clients, I don’t think are that important at this time. But it becomes sort of like the monitoring platform.
175 00:26:43.520 ⇒ 00:27:02.189 Uttam Kumaran: The reason I like… I like Metaplane is the Datadog thing makes a lot of sense, because Datadog is the leading edge in, sort of, log monitoring, and they just do a very specific thing very well, and the pricing is actually quite appropriate. Many of these tools will cost more than
176 00:27:02.390 ⇒ 00:27:10.079 Uttam Kumaran: most parts of the data stack, and I’m… don’t think that that’s entirely appropriate, like, all the time.
177 00:27:11.580 ⇒ 00:27:19.010 Zack Gibbs: So how… how are we… have you hooked up your Metaplane instance to our, how are we… how are we using it today?
178 00:27:19.250 ⇒ 00:27:38.009 Uttam Kumaran: Yeah, so we… I got access to a demo instance, that… just for Urban Stems. We… I contacted them because we’re starting to promote them for… for most of our engagements as… as our sort of data observability tool of choice, so they… they gave us a demo instance just for you, and I…
179 00:27:38.010 ⇒ 00:27:46.680 Uttam Kumaran: I told them that, like, hey, we’re gonna… we’ll put it in front of you, but they… that’s why they haven’t clamped down on, like, you only get a month or anything. We’ve been running it for about 2 months now.
180 00:27:46.680 ⇒ 00:27:55.380 Uttam Kumaran: And I’m happy to invite you there so you can poke around. But we have it all hooked up to Redshift, to dbt, and
181 00:27:55.550 ⇒ 00:27:59.239 Uttam Kumaran: And then the redshift job is what’s looking at the model values.
182 00:28:00.910 ⇒ 00:28:16.210 Uttam Kumaran: So the reason why this is also nice is dbt tests are things you have to do within code. These are all UI-based tests to enable. So it’s actually very easy for anyone on the team to come on and add new tests.
183 00:28:16.350 ⇒ 00:28:21.339 Uttam Kumaran: And actually… Sort of start to monitor and get assigned as issues happen.
184 00:28:21.420 ⇒ 00:28:38.190 Uttam Kumaran: what… for all of our teams, like, data issues happen, and it’s not our job to, like, get that to zero, although we do… we do want to see that reduced, the overall number, but most of our goal is to make sure that we can actually triage and mitigate those within a reasonable SLA.
185 00:28:38.310 ⇒ 00:28:42.500 Uttam Kumaran: And so, I’ll just share briefly, like, what the platform looks like.
186 00:28:42.780 ⇒ 00:28:56.649 Uttam Kumaran: And Emily’s in here already, and we’ve been sort of using it. And so what you can see is we have, sort of our entire redshift here. As you can see, we have 72 monitors on analytics.
187 00:28:56.670 ⇒ 00:29:12.249 Uttam Kumaran: And really, you can see we have several table-level monitors across some of our core tables, like Tableau items, like set, for example. So on this table, we have monitors on freshness and row count, and so we’re looking at created at, updated at, and user created at.
188 00:29:12.480 ⇒ 00:29:31.279 Uttam Kumaran: ideally, what we should see, and what Metaplane does, it starts creating a historical measure of how long these values have been updated, and, like, what is the expected amount. It sort of tunes it over time. So, like, what we’ve been doing is sort of understanding, okay, what are SLAs for these tables, and
189 00:29:31.390 ⇒ 00:29:42.159 Uttam Kumaran: again, this indicates to me what jobs to go attack, and how do I get this to start to run faster. And these are… we set these up all via the UI.
190 00:29:42.160 ⇒ 00:29:56.749 Uttam Kumaran: So the… we looked at the core, sort of, linchpin models in the entire, architecture, and we’re making sure that those are updated on time. And then ideally, we could start looking at, okay, the revenue per day should roughly be within this range.
191 00:29:56.750 ⇒ 00:30:03.940 Uttam Kumaran: Things like that, to make sure if it is out of bounds, like if there’s a duplicate that comes in, something’s delayed, we get flagged.
192 00:30:04.510 ⇒ 00:30:12.700 Uttam Kumaran: And our goal is to get flagged before the stakeholder finds out, right? That is the true… if we get a great monitoring system, we should know before…
193 00:30:12.960 ⇒ 00:30:23.559 Uttam Kumaran: the customer knows, and so that’s what we… this is a great platform to do that. And again, you can put in standard deviations for things. Sometimes we expect some deviation, sometimes
194 00:30:23.670 ⇒ 00:30:27.899 Uttam Kumaran: We want it to be pretty tight, and so you could do all that, set alerts, things like that, so…
195 00:30:29.070 ⇒ 00:30:29.670 Zack Gibbs: Gotcha.
196 00:30:32.030 ⇒ 00:30:34.930 Uttam Kumaran: And it does this one thing really well, like…
197 00:30:35.200 ⇒ 00:30:51.680 Uttam Kumaran: A lot of the other platforms, you’ll see if you get demos and in the pricing, they just try to consume so much, and this is not where our team is, like, lives. This needs to do its job right in order to get it to slack when something’s going wrong.
198 00:30:51.870 ⇒ 00:30:58.109 Uttam Kumaran: And it is sort of our insurance policy to make sure that the stuff we’re pushing, works.
199 00:30:58.320 ⇒ 00:31:04.540 Uttam Kumaran: Versus, again, currently the state is that typically the stakeholder, finds out first before us.
200 00:31:05.210 ⇒ 00:31:05.850 Zack Gibbs: Yeah.
201 00:31:06.250 ⇒ 00:31:17.919 Zack Gibbs: Yeah, I mean, I’m all for tooling updates and changes if they’re, like, if the ROI is there, and they’re providing, you know, good… a good level of value. And so…
202 00:31:18.680 ⇒ 00:31:22.359 Zack Gibbs: But I think it’s just making sure that
203 00:31:22.930 ⇒ 00:31:27.220 Zack Gibbs: the value is there, the value prop is there. So…
204 00:31:27.470 ⇒ 00:31:29.730 Uttam Kumaran: Yeah, I think at the $500 a month.
205 00:31:30.210 ⇒ 00:31:42.469 Uttam Kumaran: this… it’s definitely there. I think in the time you’ll… that we’ll spend, and certainly, again, in the time that internal people will not… will be without reporting, I think it’s… it pays for itself, for sure.
206 00:31:43.140 ⇒ 00:31:43.680 Zack Gibbs: Yeah.
207 00:31:44.830 ⇒ 00:31:45.900 Zack Gibbs: Oh, I’m not.
208 00:31:45.900 ⇒ 00:31:50.020 Uttam Kumaran: You’ll get pricing on other tools as well, just for, sort of,
209 00:31:51.060 ⇒ 00:31:59.769 Uttam Kumaran: I mean, I mean, I think it’d be helpful even for us to just go have refresh pricing from everybody, so I’m happy to do that this week, and then just put something together. That’s totally fine.
210 00:32:00.460 ⇒ 00:32:06.669 Zack Gibbs: Yeah, so are they… so, are the Metaplane folks bugging you now for, like, a down.
211 00:32:06.670 ⇒ 00:32:12.190 Uttam Kumaran: No, they actually have not. They’ve actually been pretty chill,
212 00:32:12.190 ⇒ 00:32:13.150 Zack Gibbs: Yeah, I think it would…
213 00:32:13.150 ⇒ 00:32:13.550 Uttam Kumaran: Yeah.
214 00:32:13.550 ⇒ 00:32:22.989 Zack Gibbs: Yeah, I think… I guess the trigger point for me would be, you know, once there’s… that’s brought up, and as the team is getting more familiarity.
215 00:32:22.990 ⇒ 00:32:38.509 Zack Gibbs: And, like, the value add is there. Then I think getting pricing from your contacts, and then making sure that we understand, like, our usage, what we would be using it for, and then I can go back to our, Datadog rep and say.
216 00:32:38.520 ⇒ 00:32:52.149 Zack Gibbs: you know, give us your best pricing here, and see how it compares with what you’re seeing from them directly. Because the Datadog rep that I’ve dealt with has, has given us steeper discounts than.
217 00:32:52.410 ⇒ 00:33:03.899 Uttam Kumaran: Yeah, I would certainly, given you are a data customer, get it as part of that paper, and get it as, like, some add-on. Like, I would… the Metaplane folks are gonna…
218 00:33:04.350 ⇒ 00:33:15.800 Uttam Kumaran: they’re gonna think of themselves as, like, number one. I think you’re right in that, like, go through Datadog, and I’ll tell… so I’ll… I’ll get… tell you. They haven’t been yelling at me, so we can keep going. I just wanted to get your, like.
219 00:33:15.950 ⇒ 00:33:22.950 Uttam Kumaran: Okay, if, for example, if they come to me, they’re like, hey, we need to do something in the next 48 hours, like, okay, we can figure something out, but…
220 00:33:22.950 ⇒ 00:33:26.970 Zack Gibbs: Yeah. One of my core questions for you guys as a part of this review was just.
221 00:33:27.150 ⇒ 00:33:46.390 Zack Gibbs: in retrospect, did we make tooling mishaps? Like, should we have changed tooling earlier in the process, or the way that we’ve gone down the path, has that been optimal, or not? And so, I’m… my assumption as a part of this was that we would be changing tooling somewhat.
222 00:33:46.980 ⇒ 00:33:50.369 Zack Gibbs: And, you know, whether that’s add, removing, changing.
223 00:33:50.370 ⇒ 00:33:50.720 Uttam Kumaran: Sure.
224 00:33:50.720 ⇒ 00:33:57.119 Zack Gibbs: So, that was one of my core questions, was, in retrospect, from your points of view, should we have done
225 00:33:57.710 ⇒ 00:33:59.940 Zack Gibbs: Should we have started this with…
226 00:34:00.150 ⇒ 00:34:09.259 Zack Gibbs: Core tooling changes, or have we… has the process been reasonably fine or optimal, so far?
227 00:34:10.080 ⇒ 00:34:18.320 Uttam Kumaran: Yeah, we can… we can also put a piece on that in our… in our little write-up. I think across the stack, I think…
228 00:34:18.580 ⇒ 00:34:37.770 Uttam Kumaran: if I talk about what tooling that I think in retrospect is fine, dbt and Redshift are fine. I think Redshift, you’re not gonna find a better deal than what you’re getting, for the amount of users you’re supporting. So, if it’s cost-conscious there, it’s fine. I think what you will pay for is that, like, we have to do a little bit of custom stuff, but again, this stuff I’ve implemented now.
229 00:34:38.260 ⇒ 00:34:51.350 Uttam Kumaran: back when you implemented Redshift, the usage wasn’t that high, and so you didn’t… you just needed… you didn’t need any of that. Now, because there’s tons of queries hitting it from multiple different places, like ETL’s hitting it, dbt’s hitting it, and Looker’s hitting it.
230 00:34:51.480 ⇒ 00:35:02.889 Uttam Kumaran: we have to figure out a couple things, and scaling and concurrency. So that, I think, is fine. dbt is perfect. Now we’re doing everything in GitHub, that’s perfect. The two areas, I think, to poke at is, like.
231 00:35:03.210 ⇒ 00:35:06.120 Uttam Kumaran: Could… should we have… move stuff.
232 00:35:06.190 ⇒ 00:35:19.480 Uttam Kumaran: to one singular ETL. I think, again, given the deal you’re getting, and given that some stuff is working on Stitch, I feel fine, but I’m glad that we made the consideration to… to use a polyatomic, and I do think that
233 00:35:19.480 ⇒ 00:35:26.930 Uttam Kumaran: we should continue to leverage that for new sources, as needed. I think the support, hopefully, you’ve seen has been positive.
234 00:35:26.930 ⇒ 00:35:35.859 Uttam Kumaran: And again, like, Fivetrane is continuing, even since we started our engagement, Fivetrain has increased their prices further, so there… there isn’t a better…
235 00:35:35.910 ⇒ 00:35:45.019 Uttam Kumaran: vendor right now, and so I appreciate that we made that decision. And then the last piece is on BI. Like, I still think
236 00:35:45.190 ⇒ 00:35:52.240 Uttam Kumaran: There are alternatives to Looker, and I think you… we… there are…
237 00:35:52.420 ⇒ 00:36:00.169 Uttam Kumaran: alternatives that will actually lower your price and be a better replacement. The one tool that we found in the last month that’s
238 00:36:00.170 ⇒ 00:36:18.130 Uttam Kumaran: really, really been successful is Omni. It’s sort of a mix of Tableau and Looker, and out of the box, they’ve been adding a lot of these, like, AI features. One of the things that we’ve been thinking about a lot is a lot of people that use a tool like Looker are using it just for point lookups sometimes.
239 00:36:18.130 ⇒ 00:36:26.589 Uttam Kumaran: all questions that can be answered by an LLM versus, like, going into a dashboard, and so I’m really trying to look for
240 00:36:26.610 ⇒ 00:36:34.379 Uttam Kumaran: tools that enable a lot of the benefits that’s coming from these, like, AI solutions, especially in data, which is, like.
241 00:36:34.470 ⇒ 00:36:53.809 Uttam Kumaran: basic questions where if the data team can own the semantic layer and understand these are the objects that mean this, an AI data tool can certainly help users ask questions over. So Omni, is one of the only tools that I think, could be worth considering on the BI side. The lift is very high.
242 00:36:53.890 ⇒ 00:37:00.399 Uttam Kumaran: like… it’s a huge… it’s a huge undertaking to move out of there. Yeah. So, like…
243 00:37:01.050 ⇒ 00:37:08.739 Uttam Kumaran: Although, like, yes, I think that there are… is there… is there alpha? Is it worth the ROI? I’d have to think about it.
244 00:37:09.220 ⇒ 00:37:21.190 Uttam Kumaran: Could it be worth doing, like, someone in parallel, and actually, because they’re up and coming, would they probably give us a pretty steep discount, or even maybe pay for it? There’s opportunity to consider that.
245 00:37:23.010 ⇒ 00:37:35.530 Uttam Kumaran: But I do think Looker is a really, really tough tool to use these days. Like, there’s come a long way in BI from that tool, and so that’s probably, like, how I would
246 00:37:35.840 ⇒ 00:37:43.209 Uttam Kumaran: describe the BI ankle. So maybe something to think about, like, if I can go to them and say, hey, like, here’s our current price.
247 00:37:43.470 ⇒ 00:37:48.520 Uttam Kumaran: If you can cut that and maybe sponsor the migration or something, maybe it’s something we can consider.
248 00:37:48.900 ⇒ 00:37:49.660 Uttam Kumaran: Yeah.
249 00:37:49.660 ⇒ 00:37:57.749 Zack Gibbs: Yeah, I mean, I went back to… so we’re under a new 12-month term with Looker, and I had them…
250 00:37:57.970 ⇒ 00:38:06.980 Zack Gibbs: cut that pretty dramatically. So we’re under… we’re under term with them through July 21st of next year.
251 00:38:07.010 ⇒ 00:38:18.289 Zack Gibbs: So… and the reality is that the goal… part of the goal here was that we have a much smaller group that’s in Looker. We have…
252 00:38:18.290 ⇒ 00:38:32.669 Zack Gibbs: reports that are being generated out of whatever the system is, and are pushed to those users who are, you know, those business users. But the subset of users in Looker, we want to generally diminish, partly because we have a lot of the
253 00:38:32.670 ⇒ 00:38:38.799 Zack Gibbs: same data sitting in Shopify. That’s easier for those teams to pull and.
254 00:38:38.800 ⇒ 00:38:39.160 Uttam Kumaran: Yes.
255 00:38:39.160 ⇒ 00:38:40.899 Zack Gibbs: report on.
256 00:38:40.900 ⇒ 00:38:45.190 Uttam Kumaran: ID lookups, customer lookups, should all happen out of Shopify, for sure.
257 00:38:45.520 ⇒ 00:38:59.129 Zack Gibbs: Yeah. So, we’ve got a good deal, like, what I view as being a pretty good deal with Looker right now, so our annual subscription price is, like, $35,000, whereas before, I think it was, like, 50-something. Yes.
258 00:38:59.130 ⇒ 00:38:59.610 Uttam Kumaran: Nice.
259 00:38:59.610 ⇒ 00:39:01.390 Zack Gibbs: 60,000.
260 00:39:01.390 ⇒ 00:39:01.820 Uttam Kumaran: And how many.
261 00:39:02.490 ⇒ 00:39:04.050 Uttam Kumaran: Where’s that for? The new one?
262 00:39:05.250 ⇒ 00:39:15.800 Zack Gibbs: It’s broken down by, look, type, but it’s, So…
263 00:39:16.250 ⇒ 00:39:23.059 Zack Gibbs: we get… part of the platform, we get, like, a set number of seats as well, so I can’t remember the exact number. It’s.
264 00:39:24.150 ⇒ 00:39:26.210 Uttam Kumaran: Like, 4 or 5 developer seats, and then…
265 00:39:26.210 ⇒ 00:39:30.730 Zack Gibbs: Extra, extra seats, so this is the platform. So there’s,
266 00:39:31.270 ⇒ 00:39:37.130 Zack Gibbs: 5 extra standard users, which I think are the dev… no, no, no, there’s 2 extra developer users.
267 00:39:37.350 ⇒ 00:39:48.489 Zack Gibbs: 14 extra viewer users and 5 extra standard users, and then the platform comes with… comes with some quantity of users as platform access.
268 00:39:48.490 ⇒ 00:39:49.790 Uttam Kumaran: Yeah, okay, okay.
269 00:39:52.710 ⇒ 00:39:56.800 Zack Gibbs: So I’m not tied to Looker, nobody is tied to Looker, but…
270 00:39:58.170 ⇒ 00:40:00.460 Uttam Kumaran: It’s pretty cheap. I think,
271 00:40:01.720 ⇒ 00:40:13.129 Uttam Kumaran: Yeah, the only other thing I’m thinking about is, like, one of the things that’s been really cool in Omni is they’ve added, like, basically, like, spreadsheets within the platform that directly,
272 00:40:13.250 ⇒ 00:40:20.070 Uttam Kumaran: can hook up to the data model, so business users can get, like, a Basically, like a…
273 00:40:20.180 ⇒ 00:40:33.770 Uttam Kumaran: spreadsheet-esque experience directly in Omni, which for, I know, for our finance users can be really helpful, because they are currently copying and pasting into Excel or Google Sheets, and they basically immediately lose all
274 00:40:34.070 ⇒ 00:40:49.460 Uttam Kumaran: like, the cutoff happens there from any live data or any updated definitions. This feature itself, like, in my career, would have been so nice. They basically recreated, like, a Google Sheet, like, a basic Excel-type environment.
275 00:40:49.540 ⇒ 00:40:57.140 Uttam Kumaran: in Omni. So, there is added functionality. I think, like, I can… I basically can get a…
276 00:40:57.350 ⇒ 00:41:00.339 Uttam Kumaran: Like, a little bit of, like, a what’s the pricing these days?
277 00:41:00.600 ⇒ 00:41:13.849 Uttam Kumaran: We could try it for, like, maybe one user, or maybe, again, if it’s, like, PK comes in and just, like, attempts to use it for his, if it could be something worth considering. Because the LLM and the AI-based features you’re gonna get out of this.
278 00:41:13.960 ⇒ 00:41:17.590 Uttam Kumaran: It’s really, really quite amazing.
279 00:41:17.900 ⇒ 00:41:18.470 Zack Gibbs: Yeah.
280 00:41:18.470 ⇒ 00:41:22.419 Uttam Kumaran: But I want to make sure that it’s not, like, it’s ideally the same price.
281 00:41:23.600 ⇒ 00:41:28.460 Zack Gibbs: But high level, there’s no major missteps that we’ve made in tooling.
282 00:41:28.630 ⇒ 00:41:34.239 Zack Gibbs: Retrospectively, as is what I’m hearing. Like, Looker’s fine, it’s not ideal, but.
283 00:41:34.240 ⇒ 00:41:39.009 Uttam Kumaran: I think it’s a lot of pro- I think it’s a lot of process and, like, governance changes.
284 00:41:39.290 ⇒ 00:41:46.220 Uttam Kumaran: Especially on the way you develop dbt and the structure within dbt. Like, any tool you can…
285 00:41:46.610 ⇒ 00:41:54.660 Uttam Kumaran: sort of slap on and just use, but there is a way to use those tools that are best. Same with the version control. But overall.
286 00:41:55.020 ⇒ 00:41:56.090 Uttam Kumaran: Not really.
287 00:41:57.620 ⇒ 00:42:07.290 Uttam Kumaran: Yeah, I feel pretty good. Like, and you guys didn’t use, like, a Power BI, you’re not, like, on an Informatica, you’re not running queries directly on Postgres.
288 00:42:07.350 ⇒ 00:42:09.780 Uttam Kumaran: Those are usually the issues that we see.
289 00:42:09.810 ⇒ 00:42:13.249 Uttam Kumaran: Everyone here feels pretty comfortable. I think…
290 00:42:13.270 ⇒ 00:42:32.789 Uttam Kumaran: the thing to really make sure is that bad code doesn’t enter the system, and that we remove as much of the old legacy crap as possible. And then at that point, you’re like, okay, new stuff, there’s a checks and balance for it to get added. That was the only thing that wasn’t there, which is why every system kind of got bloated, because there was no, like.
291 00:42:33.390 ⇒ 00:42:35.249 Uttam Kumaran: Rhyme or reason for doing things.
292 00:42:36.810 ⇒ 00:42:37.750 Uttam Kumaran: So…
293 00:42:40.290 ⇒ 00:42:40.880 Zack Gibbs: Gotcha.
294 00:42:42.590 ⇒ 00:42:56.730 Uttam Kumaran: The other thing I wanted to chat about was Northbeam. So the feedback we got from the Northbeam folks is because it’s all Shopify orders, it’s not like Northbeam native data. They’re not able to give us
295 00:42:57.260 ⇒ 00:42:59.399 Uttam Kumaran: stuff via the API.
296 00:42:59.620 ⇒ 00:43:06.630 Uttam Kumaran: I know it… we… it sort of fell by the wayside, but I have those emails from them to kind of refresh.
297 00:43:06.820 ⇒ 00:43:09.519 Uttam Kumaran: So I kind of wanted to get your sense on, like.
298 00:43:10.430 ⇒ 00:43:16.650 Uttam Kumaran: path forward. We basically wanted to get attribution from them, what the North Beam
299 00:43:16.940 ⇒ 00:43:19.930 Uttam Kumaran: team said, is that because it’s…
300 00:43:20.150 ⇒ 00:43:24.559 Uttam Kumaran: you’re using the orders API, because the order… their orders API
301 00:43:24.780 ⇒ 00:43:30.100 Uttam Kumaran: is only able to show the orders from their orders API, not Shopify orders.
302 00:43:30.440 ⇒ 00:43:32.989 Uttam Kumaran: They said we have to manually go export.
303 00:43:33.170 ⇒ 00:43:34.210 Uttam Kumaran: from them.
304 00:43:36.610 ⇒ 00:43:41.649 Zack Gibbs: You’re talking about the… so… the… using the data export API?
305 00:43:42.430 ⇒ 00:43:42.990 Zack Gibbs: Or…
306 00:43:42.990 ⇒ 00:43:47.979 Uttam Kumaran: They didn’t… like, we even asked them if we could do that, and they didn’t,
307 00:43:48.550 ⇒ 00:43:59.670 Uttam Kumaran: say that was possible. Like, we basically said, can we get this through the API? And they said, you can’t get any of these orders through their API, you have to basically export manually.
308 00:44:08.360 ⇒ 00:44:18.350 Zack Gibbs: Yeah, maybe… why don’t you forward that thread to myself and Kristen Sampat, and I’ll throw her…
309 00:44:18.610 ⇒ 00:44:19.440 Uttam Kumaran: Okay.
310 00:44:19.440 ⇒ 00:44:21.810 Zack Gibbs: She’s the… she’s the relationship owner.
311 00:44:21.960 ⇒ 00:44:24.310 Uttam Kumaran: Okay. With them on our side. Okay.
312 00:44:24.730 ⇒ 00:44:28.950 Zack Gibbs: I mean… Yeah, I just wanna read through it. There’s gotta be a way…
313 00:44:28.950 ⇒ 00:44:32.019 Uttam Kumaran: I feel like… I mean, I’m… hopefully we’re just, like.
314 00:44:32.420 ⇒ 00:44:38.429 Uttam Kumaran: not saying the right thing, but I was very surprised, because I was like, Okay, yeah, so…
315 00:44:38.800 ⇒ 00:44:41.879 Zack Gibbs: Alright, let’s do that. I’ll… Because it’s not just…
316 00:44:42.580 ⇒ 00:44:54.470 Zack Gibbs: it’s not just orders, it’s all, like, the session… the session attribution is important as well, right? The order… the order tie-in from Shopify to theirs, there’s got to be some way to key, you know, key those two together.
317 00:44:54.470 ⇒ 00:44:57.180 Uttam Kumaran: Yeah, that’s why I was surprised,
318 00:44:57.480 ⇒ 00:44:59.639 Uttam Kumaran: But also, it’s… it’s just like a…
319 00:45:00.850 ⇒ 00:45:05.749 Uttam Kumaran: CSM on their side, so maybe we need to get looped up with somebody, or it needs to get escalated on their side.
320 00:45:06.640 ⇒ 00:45:08.159 Uttam Kumaran: So, I can forward you that.
321 00:45:08.800 ⇒ 00:45:09.460 Zack Gibbs: Okay.
322 00:45:13.040 ⇒ 00:45:20.800 Uttam Kumaran: Okay, cool. So I think takeaways on our side… I think we should get…
323 00:45:20.920 ⇒ 00:45:29.080 Uttam Kumaran: Amber, we can work on, sort of, that quick wins. We can work on a little bit of write-up that we can share. I think as part of this monthly process.
324 00:45:29.290 ⇒ 00:45:48.979 Uttam Kumaran: And I don’t know, Justin, we can all talk about it, is having that something as, like, that Zach can surface up to the rest of the business, or we can send and we can all approve, I think is really, really helpful, just a really clear wins that are measurable, which we have a bunch of those. I think we want to come to a little bit of decision on Metaplane, and then,
325 00:45:49.480 ⇒ 00:46:08.639 Uttam Kumaran: you know, I think we want to figure out this North Beam issue. We want to talk about getting PK involved in all things in our world. I think he’s a good analyst to focus on. I think Amber, it’d be helpful to see, like, what tickets he has on his plate, what tickets he may have submitted for our team to do that maybe we can show him how to fish for himself.
326 00:46:09.170 ⇒ 00:46:13.500 Uttam Kumaran: And, yeah, I think we can… Go from there.
327 00:46:14.010 ⇒ 00:46:17.729 Zack Gibbs: And PK is also another one that, if we’re having
328 00:46:17.870 ⇒ 00:46:24.930 Zack Gibbs: North Beam integration discussions and or issues, like, he should… he could help be a driver there.
329 00:46:24.930 ⇒ 00:46:25.810 Uttam Kumaran: Okay, cool.
330 00:46:25.810 ⇒ 00:46:29.629 Zack Gibbs: He’s on the marketing… he’s on the marketing team. He reports to Kristen.
331 00:46:29.630 ⇒ 00:46:34.450 Uttam Kumaran: Okay, he’s on this thread, but it was mainly, like, me, Galeb, and the North Beam, and I was like.
332 00:46:34.560 ⇒ 00:46:36.709 Uttam Kumaran: maybe I’m missing something here…
333 00:46:36.850 ⇒ 00:46:45.640 Uttam Kumaran: And so maybe I think that’s a good thing, Amber, too. If he’s on our team and we can review this together, then he can also help push or basically isolate what we need.
334 00:46:47.710 ⇒ 00:46:50.570 Zack Gibbs: Yeah, I mean, they’re the stakeholder group that…
335 00:46:51.100 ⇒ 00:46:56.729 Zack Gibbs: You know, once this integration’s set up, and, you know, ingesting this data, so…
336 00:46:56.900 ⇒ 00:47:01.580 Zack Gibbs: He can be a… he can be a driver, and then, you know, trouble… troubleshooter.
337 00:47:01.580 ⇒ 00:47:16.969 Uttam Kumaran: It’s actually great because, you know, even as part of our roadmap, we said we’re gonna deprive some of the marketing stuff, so it would be great for him to start taking that stuff on. Like, there are basic things, like, he… we’ve helped him in small ways where he’s wanted things refreshed faster, or us to create some simple models.
338 00:47:17.060 ⇒ 00:47:23.939 Uttam Kumaran: But I can show him how to do all that, and he can start to get those wins for that team, versus waiting for us to get to it, so…
339 00:47:24.720 ⇒ 00:47:25.250 Zack Gibbs: Yep.
340 00:47:27.030 ⇒ 00:47:32.769 Zack Gibbs: And then what about… what about loop data?
341 00:47:33.070 ⇒ 00:47:34.039 Zack Gibbs: Where does that stand?
342 00:47:34.040 ⇒ 00:47:52.320 Uttam Kumaran: We do have looped data in. It’s currently on my plate to build some of the subscription models, so the… that integration with Polytomic was fine, and that data’s coming in just fine. We’ve been focusing on sort of building transactions, then customers first.
343 00:47:52.570 ⇒ 00:47:56.570 Uttam Kumaran: But it’s… it’s on the, like, short-term roadmap.
344 00:47:56.740 ⇒ 00:47:57.990 Uttam Kumaran: For all the subscriptions.
345 00:47:59.280 ⇒ 00:48:00.340 Zack Gibbs: Okay, gotcha.
346 00:48:00.620 ⇒ 00:48:01.210 Uttam Kumaran: Yeah.
347 00:48:01.530 ⇒ 00:48:08.139 Uttam Kumaran: So the biggest thing also is, like, you know, I don’t think the analyst team right now is too familiar with
348 00:48:08.460 ⇒ 00:48:15.639 Uttam Kumaran: like, running selects in Redshift and just, like, looking at the data, because I was like, hey, can you guys go check out this table? They’re like, is it in Looker? I’m like.
349 00:48:15.850 ⇒ 00:48:22.200 Uttam Kumaran: I don’t want to bring it into Looker until you just, like, run some sums and tell me what you think.
350 00:48:22.360 ⇒ 00:48:42.360 Uttam Kumaran: And so I’m trying to also think about that process, and maybe that’s something I can work with PK on, which is like, look, I’m not gonna… we’re not gonna bring it into Looker until it’s QA’d upstream. Otherwise, if there are fuels you don’t need, or fields you do need, just do that all now, and then… because Looker, we’re going to generate views, explorers, and all this stuff that we’ll have to update.
351 00:48:42.370 ⇒ 00:48:48.679 Uttam Kumaran: So, maybe that’s another win for us to think about, is, like, how do these guys come further upstream and help QA?
352 00:48:50.280 ⇒ 00:48:56.789 Uttam Kumaran: You know, because they’re like, oh, wait until it gets in Looker. I’m like, I don’t want to ship it until you tell me whether these columns are…
353 00:48:56.910 ⇒ 00:49:02.579 Uttam Kumaran: are right or not, you know? And I think maybe they’re just a little bit nervous about, like, what does that mean?
354 00:49:02.780 ⇒ 00:49:03.490 Uttam Kumaran: So…
355 00:49:03.880 ⇒ 00:49:09.470 Zack Gibbs: Yeah. From a tooling standpoint, I know that we’re using Retroflake, should we have considered, like, a Snowflake instead?
356 00:49:10.950 ⇒ 00:49:20.530 Uttam Kumaran: It would just be easier for people to run queries in, and, like, the user governance, like, all that pain we went through would not have been the case.
357 00:49:23.040 ⇒ 00:49:36.139 Uttam Kumaran: So, like, but again, like, the way we’re gonna get around it is if you’re comfortable… like, I can get everybody on Table Plus, and they can query directly. Because there’s actually not many, like, developers, right? Like, there’s only a couple of us and Emily.
358 00:49:36.220 ⇒ 00:49:45.569 Uttam Kumaran: I’m not that worried about people getting into Redshift and doing much. Like, if you had data science folks, and if you had other data analysts that were very, like, in the warehouse.
359 00:49:45.660 ⇒ 00:49:48.429 Uttam Kumaran: running queries and creating models themselves, it could be…
360 00:49:48.830 ⇒ 00:49:53.559 Uttam Kumaran: It could be better to do that, but right now, most of the people are accessing the data through Looker.
361 00:49:53.910 ⇒ 00:49:55.020 Uttam Kumaran: That’s… okay.
362 00:49:55.770 ⇒ 00:49:56.370 Zack Gibbs: Gotcha.
363 00:49:56.660 ⇒ 00:49:57.240 Uttam Kumaran: Yeah.
364 00:49:58.450 ⇒ 00:49:59.150 Zack Gibbs: Okay.
365 00:50:00.500 ⇒ 00:50:06.189 Uttam Kumaran: Okay, cool, so we have some action items on our side. Yeah, if there’s anything else to cover, let me know.
366 00:50:06.570 ⇒ 00:50:20.999 Zack Gibbs: Yeah. I think… I think one of the things, for next time, or, you know, our, kind of our next test races, I know that we’re… we’re contracted through the end of November. Yes. So it would be good to see, get a sense of, like.
367 00:50:21.550 ⇒ 00:50:34.500 Zack Gibbs: where are we going to end? And where are the risks? Where are the risk areas, that you guys see, you know, if we were to truly roll off at the end of November? So…
368 00:50:35.360 ⇒ 00:50:39.410 Zack Gibbs: I would like to get a better sense of that. I don’t know if we did that
369 00:50:39.670 ⇒ 00:50:41.180 Zack Gibbs: If we did that in…
370 00:50:42.120 ⇒ 00:50:48.749 Zack Gibbs: you know, middle of October, the timing’s probably fine. We could also do it sooner, if you guys wanted to do it sooner.
371 00:50:49.330 ⇒ 00:51:01.949 Uttam Kumaran: Yeah, let’s work… let’s work on that, team, and then we can send that async, and then if we want… if we’re… want to have a conversation sooner, we can do that. We’ll just plan… I’ll put a meeting… we’ll just put a recurring meeting on sometime for mid-next… mid-next month.
372 00:51:02.220 ⇒ 00:51:09.249 Uttam Kumaran: That way, at minimum, it happens, but I think we can try to get something together before that and send for comments.
373 00:51:09.250 ⇒ 00:51:10.690 Zack Gibbs: Okay, yeah, sounds good.
374 00:51:11.070 ⇒ 00:51:11.660 Uttam Kumaran: Okay.
375 00:51:12.240 ⇒ 00:51:13.610 Uttam Kumaran: Okay, perfect.
376 00:51:13.910 ⇒ 00:51:15.730 Uttam Kumaran: Alright, thanks everyone, appreciate it.
377 00:51:15.730 ⇒ 00:51:18.180 Zack Gibbs: Alright, appreciate it. Thanks. See ya.
378 00:51:18.180 ⇒ 00:51:18.760 Uttam Kumaran: Bye.