Meeting Title: Brainforge x CTA: Analytics! Date: 2025-09-16 Meeting participants: Uttam Kumaran, Katherine Bayless
WEBVTT
1 00:01:55.250 ⇒ 00:01:56.659 Uttam Kumaran: Hey, how are you?
2 00:01:56.660 ⇒ 00:01:57.880 Katherine Bayless: Good, how’s it going?
3 00:01:57.880 ⇒ 00:02:00.830 Uttam Kumaran: Good. Great to see ya. How’s the week going?
4 00:02:01.270 ⇒ 00:02:11.530 Katherine Bayless: Well, yeah, we, like I said in my Evo, we launched, registration last week, so it’s, it’s a little calmer this week than last, but definitely a little chaos.
5 00:02:11.530 ⇒ 00:02:18.389 Uttam Kumaran: How is, how has it been, like, in the machine now? I’ve just been a… as soon as I started talking to Christine, I’ve just been a big fan of, like.
6 00:02:18.520 ⇒ 00:02:35.540 Uttam Kumaran: CES for my whole life, so I was like, it’s awesome to just even hear what the inside is like. And during, I don’t know, every year I feel like technology is big, but this is a particularly, like, major time for technology. But yeah, like, how is it… how’s it been so far?
7 00:02:35.870 ⇒ 00:02:50.480 Katherine Bayless: I mean, so this is my first lap around the sun. I started in April, and similarly, you know, heard about CES my whole life, never gone. It’s definitely an interesting year. I mean, there’s certainly a spotlight on technology. The AI thing is.
8 00:02:50.480 ⇒ 00:02:50.850 Uttam Kumaran: Yes.
9 00:02:50.850 ⇒ 00:03:02.599 Katherine Bayless: Because obviously, AI is technology, but we are slow, we aren’t, like, the AI association, right? And so it’s like, we need to get our, you know, hands into that world a little bit, and… oh my god, your dog is adorable.
10 00:03:02.600 ⇒ 00:03:02.980 Uttam Kumaran: But no.
11 00:03:02.980 ⇒ 00:03:08.239 Katherine Bayless: And, you know, not for nothing, but the international attendance is a huge part of the show, and…
12 00:03:08.240 ⇒ 00:03:08.750 Uttam Kumaran: I see.
13 00:03:08.750 ⇒ 00:03:11.249 Katherine Bayless: International relations as a country right now are not the strongest.
14 00:03:11.250 ⇒ 00:03:23.540 Uttam Kumaran: I guess there’s… but I would say, you know, in the history of technology, like, since I’ve been tracking, like, the last 15 years or so, this is the biggest year for, like, robotics, you know, so for consumer electronics, I feel like…
15 00:03:23.540 ⇒ 00:03:24.220 Katherine Bayless: Yeah.
16 00:03:24.220 ⇒ 00:03:36.170 Uttam Kumaran: stuff that’s with robotics, like, there’s a lot of new, like, home vacuum, outdoor robotics, all of those have, like, some type of AI thing, so it’s not just, like.
17 00:03:36.290 ⇒ 00:03:40.249 Uttam Kumaran: Google Glasses or, like, Amazon Echos, there’s actually, like.
18 00:03:40.370 ⇒ 00:03:45.130 Uttam Kumaran: Home robotics that are pretty significant, so…
19 00:03:45.660 ⇒ 00:04:00.090 Katherine Bayless: Yeah, one of the teams I would love to, like, have time to get to know a little bit better are the ones that help work with our, like, medical technology companies, because to your point, right, like, we’ve definitely entered the stage where, like, this is real, it’s here, it’s in the field, like, yeah.
20 00:04:00.090 ⇒ 00:04:01.020 Uttam Kumaran: Yeah.
21 00:04:01.020 ⇒ 00:04:01.740 Katherine Bayless: Yeah.
22 00:04:01.890 ⇒ 00:04:11.419 Uttam Kumaran: Awesome. Well, yeah, I mean, I started talking to Christina, like, a few months ago, just got introduced by, a friend of mine here in town,
23 00:04:11.510 ⇒ 00:04:26.479 Uttam Kumaran: And so, a little bit about us. My background is in data engineering. I led… was a data engineer for a while, led several data teams, and just really opinionated about how to introduce great data practices and build up the data function, and
24 00:04:26.480 ⇒ 00:04:42.790 Uttam Kumaran: You know, I worked in New York for quite a while, then moved here to Austin a few years ago, and started Brainforge about 2 years ago, so we primarily started as a data analytics company, so we come in, we’re basically, typically come in, and we sort of are poke and…
25 00:04:42.850 ⇒ 00:05:00.070 Uttam Kumaran: fill in the gaps on where you don’t have data team staff, so that could be, like, head of data type activities, could also be anything across the entire data stack, so data engineering, analytics engineering, BI, and then we also do a lot of product analytics work, so, like, Amplitude, MixedPanel, stuff like that. So.
26 00:05:00.070 ⇒ 00:05:06.239 Uttam Kumaran: We don’t… we don’t tend to come in, like, purely just, like, kind of a dev shop. It’s usually all, like.
27 00:05:06.240 ⇒ 00:05:07.210 Uttam Kumaran: partner…
28 00:05:07.210 ⇒ 00:05:27.000 Uttam Kumaran: space where we’re, like, a partner to our clients, just because myself and my business partner, we’ve worked with executives and executive functions our whole career on establishing data… data teams, so that’s sort of what we’ve been doing for about 2 years. And in the last year or so, we also started doing a lot of AI-related services, because I was using AI a lot to build this business.
29 00:05:27.000 ⇒ 00:05:40.859 Uttam Kumaran: And I found it to be really challenging. After doing that for, you know, more than a year, I found that we knew a lot about how to do… how to actually implement AI in the business, just from doing it ourselves, so now we actually have
30 00:05:40.860 ⇒ 00:05:49.700 Uttam Kumaran: sort of services that we offer, and really where I hope this is going is that those start to blend in, right? Like, because we come in and we gather
31 00:05:49.700 ⇒ 00:06:02.450 Uttam Kumaran: all of the semantic context for the data world, that is really, really amazing and rich for AI-related use cases, and so we’re starting to kind of blend those two together, but we’re a consultancy, and so we’ve helped
32 00:06:02.450 ⇒ 00:06:14.390 Uttam Kumaran: A lot of folks across, B2B, SaaS, e-commerce, legal, health, sort of just stand up data functions and sort of plug in where needed.
33 00:06:14.440 ⇒ 00:06:24.180 Uttam Kumaran: So that’s kind of how Christina and I started talking. I know, you know, in your world, I know it’s a lot of customer 360, like, understanding marketing initiatives, the performance of those.
34 00:06:24.240 ⇒ 00:06:39.720 Uttam Kumaran: Also, nicely, it’s sort of squared on a timely basis, right? There is, like, a big activation, and so that’s also a really interesting data problem. But yeah, at that point when I was talking to Christina, she just mentioned that she was just kind of figuring it out.
35 00:06:39.720 ⇒ 00:06:52.799 Uttam Kumaran: And so, yeah, I’m glad that I… I feel like you’re… I’m sure you’re… you’re in the weeds now, kind of seeing the state of the world, hearing from her, and sort of bridging the gap between what’s on the ground, but yeah, just wanted to have a conversation and see if there’s any way
36 00:06:52.800 ⇒ 00:06:56.260 Uttam Kumaran: Based on what you’re seeing, we could be helpful.
37 00:06:56.350 ⇒ 00:06:58.460 Uttam Kumaran: Like, yeah, just, like, happy to chat.
38 00:06:58.780 ⇒ 00:07:13.460 Katherine Bayless: Yeah, yeah, totally. And I mean, you know, I’m very nerdy, so I’m always happy to meet other nerds, and I can do that, so if you’ve got help that I need, I’m more than happy to continue the conversation. Maybe in that spirit, kind of like, at a high level.
39 00:07:13.740 ⇒ 00:07:15.129 Katherine Bayless: I’d be curious.
40 00:07:16.160 ⇒ 00:07:19.630 Katherine Bayless: I think I, like, organized my thoughts a little bit. Like…
41 00:07:22.450 ⇒ 00:07:31.640 Katherine Bayless: CTA accomplishes a lot, considering the maturity of their tech stack. Yeah. Yeah, right? Okay. Yep.
42 00:07:31.830 ⇒ 00:07:35.629 Katherine Bayless: So, like, a lot of my work right now is kind of…
43 00:07:36.060 ⇒ 00:07:51.640 Katherine Bayless: dealing with the existing systems, and hoping slash planning for improvements, replacements, you know, that kind of thing. Necessarily, obviously, my colleague in IT, Jay, he and I, you know, we talk a lot.
44 00:07:52.170 ⇒ 00:08:04.580 Katherine Bayless: I would like to use Box instead of SharePoint, for example. Right? And so, you know, some of these decisions kind of fall on blurred lines between data and IT. Fortunately, I think as much as I drive him crazy, we do get along. And so…
45 00:08:04.580 ⇒ 00:08:05.820 Uttam Kumaran: That’s usually the case.
46 00:08:05.820 ⇒ 00:08:12.849 Katherine Bayless: Yeah, and so I’d be curious, like, from your perspective, like, the work that you’ve done, whether as Brainforge or before, like.
47 00:08:13.920 ⇒ 00:08:24.799 Katherine Bayless: What level of, you know, sort of experience and comfort do you have around, like, dealing with the realities of a lot of legacy systems and things that are kind of, you know.
48 00:08:24.960 ⇒ 00:08:40.000 Katherine Bayless: not gonna have the off-the-shelf connectors and, you know, the easy Zapier-type integrations like that, right? Like, we have 50-some different systems total. I’ve identified 8 core CRMs. One of them is Salesforce, but ironically.
49 00:08:40.520 ⇒ 00:08:55.600 Katherine Bayless: while that system contains the data that goes with the most of our revenue, it doesn’t actually have a lot of data data in it. Like, it’s not a core source of information for us, per se, but there’s a lot of gravity towards Salesforce when I talk to these consulting companies, because they’re like, you have Salesforce, right?
50 00:08:55.600 ⇒ 00:08:56.489 Uttam Kumaran: Yeah, yeah, yeah.
51 00:08:56.490 ⇒ 00:09:07.199 Katherine Bayless: The reality is, the data that we need is in different systems that are, you know, sort of more association or event-specific. So just curious to get your level of experience there.
52 00:09:07.200 ⇒ 00:09:09.530 Uttam Kumaran: Yeah, I mean, I would say…
53 00:09:09.610 ⇒ 00:09:25.820 Uttam Kumaran: I started this consultancy because I actually… I’m not a consultant myself, I actually brought on a lot of consultants and was very unimpressed. I feel like, ultimately, what makes a great consultant is, one, is we’ve seen so many
54 00:09:25.820 ⇒ 00:09:30.379 Uttam Kumaran: Of these issues that we come to the table with how to fix them fast.
55 00:09:30.520 ⇒ 00:09:41.540 Uttam Kumaran: We’ve also done this so many times that we come in with some playbooks on, like, how do we architect a strategy around this? Third is, like, we’re very opinionated, so…
56 00:09:41.790 ⇒ 00:09:53.940 Uttam Kumaran: we don’t come in… we come in as a… as I mentioned, we come in as a partner, but that’s because, like, we can’t… if we were to just take direction from all of our consultants, we’re coming in because there is a problem, right? And so, for us, we’re extremely outcome-driven.
57 00:09:54.020 ⇒ 00:10:13.880 Uttam Kumaran: I’m very agnostic about vendors and tools. In fact, in our company, we decided early on not to take kickbacks from any vendors. One, I just… I’m kind of a nerd in that I know all the tools in the space. I’ve bought them and used them my whole career, and so I know all the good and bad, but also.
58 00:10:13.880 ⇒ 00:10:25.299 Uttam Kumaran: my clients, they want an outcome, and so whatever the shovel is I need to use to dig that hole, I will. Today, it may be Tableau or Looker, tomorrow it will be Omni or Sigma.
59 00:10:25.300 ⇒ 00:10:32.899 Uttam Kumaran: And that is fine. Like, I don’t need… I don’t need to, like, attach my card to them. And in fact.
60 00:10:32.900 ⇒ 00:10:47.349 Uttam Kumaran: our job is to procure the best solution for the client based on their budget, based on their timeline, and based on who’s using it. And so, we’ve walked into many different situations where we have control over what tools to bring in, and we don’t.
61 00:10:47.350 ⇒ 00:11:07.309 Uttam Kumaran: But we’ve come in a situation where there’s, like, Looker Studio, and we have to deal with that for, like, 6 months until I can convince a stakeholder that, hey, maybe you want to invest in a little bit of a more broader BI tool. Or, yes, there are disparate databases, there’s some stuff in Postgres, there’s some stuff, flat files in CSV, there’s some stuff that’s coming through, like.
62 00:11:07.310 ⇒ 00:11:11.410 Uttam Kumaran: SFTP, okay, we need to, like, grab all that. The biggest…
63 00:11:11.550 ⇒ 00:11:17.040 Uttam Kumaran: kind of paradigms for us is, one, we need to have some central warehouse, whether that is
64 00:11:17.360 ⇒ 00:11:31.260 Uttam Kumaran: a snowflake, whether that is a mother duck or a BigQuery, just someplace to land everything. We just apply our usual sort of modeling, where we have, like, raw, we have staging, intermediate, and we build the data marks.
65 00:11:31.300 ⇒ 00:11:44.799 Uttam Kumaran: So we’re used to walking into situations where there aren’t APIs, where there are, like, flat files coming in, there’s historical data that’s been there that no one’s touched, and, like, no one knows where it came from.
66 00:11:44.800 ⇒ 00:12:07.999 Uttam Kumaran: And we need to use… we need to continue to drag that along with us in Union In, so we’ve… we’ve dealt with that. We’ve also dealt with bigger migrations, like folks moving from Redshift onto something more modern, folks moving BI tools. But also, like, I don’t… I never… we never come into a client and are like, you have to migrate all this. We have to solve… there is, like, a pressing problem that, like, our job is to make sure they solve, and then
67 00:12:08.060 ⇒ 00:12:21.239 Uttam Kumaran: for me, selfishly, I want to make sure we get trust, so that when we do propose these changes, we can clearly articulate, and they… and they also know that it’s something that can be accomplished. Data migrations and these things are, like.
68 00:12:21.490 ⇒ 00:12:35.790 Uttam Kumaran: they’re just, like, always popping, and everybody’s so afraid of them. And so for us, I’m very, very careful on not saying that your tool is the reason you, like, have a problem. It’s most likely people, and it’s most likely just…
69 00:12:35.830 ⇒ 00:12:46.710 Uttam Kumaran: like, having a strategy, and someone to, like, own the fire. So when we walk into clients, I basically say, like, shove us into… into the…
70 00:12:46.820 ⇒ 00:12:55.410 Uttam Kumaran: the hottest burning building. My team probably doesn’t enjoy that, but I know that that’s why we’re there, and if we can’t solve that.
71 00:12:55.500 ⇒ 00:13:15.249 Uttam Kumaran: then, you know, then it’s not worth it, you know, for us to help. So that’s sort of, like, our philosophy on it. But again, we’ve worked in the modern data stack environment. We’ve used the plethora of tools across the whole stack, across ETL, warehouse, BI. We’ve used dbt for all of our modeling. I feel like it’s this kind of standard.
72 00:13:15.420 ⇒ 00:13:23.500 Katherine Bayless: Yeah, it’s such funny, it’s like, somehow in my entire career, I have yet to actually use dbt, but this is, like, this, I’m like, this time, this time we’re gonna do it.
73 00:13:23.500 ⇒ 00:13:28.320 Uttam Kumaran: It’s pretty nice, I mean, it’s just, like, it’s just advanced… it’s just advanced SQL, where you could…
74 00:13:28.680 ⇒ 00:13:42.929 Uttam Kumaran: reference… Yeah. So it’s… it’s pretty nice, and we use it for everything. And then we also… other stuff we do is we have everything version controlled. We also then eventually try to implement some type of observability and alerting strategy, like.
75 00:13:42.930 ⇒ 00:14:01.389 Uttam Kumaran: Yep. Numbers are beyond range, or stuff is null, we have, like, staging environments. So, like, as we go in and we solve, like, very short-term, like, I don’t have this report, or this is broken because a past consultant came in and set it up, we don’t know. We solve that. In parallel, we’re like, okay, what is, like, the architecture
76 00:14:01.390 ⇒ 00:14:11.950 Uttam Kumaran: And, like, the vendor strategy. And then a lot of that is cost, so can we go in and… because we’ve negotiated a lot of these deals before, I can go in and try to get the best deal, understand
77 00:14:12.500 ⇒ 00:14:24.459 Uttam Kumaran: have duplication and tools that you should consolidate. So that is something that I think, when we tend to work with CFOs and about budgets, we tend to actually find opportunities to
78 00:14:24.480 ⇒ 00:14:34.270 Uttam Kumaran: We find that, again, when you’re dealing with a software vendor, they’re gonna just upsell you as much as possible. They’re gonna say, cool, it looks like you’re growing on this trend, so we’re gonna sell you on a 5-year thing.
79 00:14:34.600 ⇒ 00:14:38.410 Katherine Bayless: I mean, if somebody tries to kill me Data Cloud one more time, I will punch them in the face.
80 00:14:38.410 ⇒ 00:14:56.200 Uttam Kumaran: Yeah, so, like, we come in and kind of ruin their party, because I actually know a lot of those folks have now become my friends, and they… I come in, I’m like, we don’t need any of this. We’re just doing basic SQL modeling, like, don’t sell us, like, an advanced machine learning situation, so…
81 00:14:56.620 ⇒ 00:14:57.300 Uttam Kumaran: Yeah.
82 00:14:57.670 ⇒ 00:15:12.689 Katherine Bayless: Yeah, yeah, no, yeah, yesterday, our marketing teams, we have Salesforce Marketing Cloud, and the contract’s up for renewal, and they’re deprecating ad audiences, which isn’t necessarily a big deal, but of course, they were immediately like, well, it’s available in Data Cloud, and I was like.
83 00:15:13.720 ⇒ 00:15:25.989 Uttam Kumaran: But the thing is, they’ll convince your team that you need it. Oh, they have, yeah. I mean, people in here definitely want it, and I’m like, you don’t understand, it would not solve… That’s the barbar, I know. But see, the thing, for me…
84 00:15:26.110 ⇒ 00:15:28.080 Uttam Kumaran: A lot of our team would be like.
85 00:15:28.310 ⇒ 00:15:32.890 Uttam Kumaran: you don’t know, but for me, I’m like, okay, we have to show some proof of concept that we can achieve that.
86 00:15:33.090 ⇒ 00:15:45.580 Uttam Kumaran: And then they’re… they’re just… they just want the output, right? Like, I think people are… generally, especially, like, people these days, I think, are so hooked onto the tools. Yeah. Like, I only know HubSpot, I only know Salesforce, I only do Snowflake.
87 00:15:45.590 ⇒ 00:15:49.750 Katherine Bayless: Right? There are a lot of primitives, like, across all these things, and.
88 00:15:49.750 ⇒ 00:16:02.119 Uttam Kumaran: Fundamentally, you’re gonna pay the price, and those tools are slow, and it’s a tax. Like, your team has to go somewhere and do something, and then you need a Salesforce consultant, and you need to call them to get the feature.
89 00:16:02.470 ⇒ 00:16:07.370 Uttam Kumaran: Like, it’s just, like, they tend to dominate our industry, so…
90 00:16:07.770 ⇒ 00:16:22.409 Katherine Bayless: Well, so I think… I think that’s helpful. Maybe, maybe as additional sort of clarifying, so… I think maybe we are two people, very, very similar. So yeah, I came in, I was like, where’s the biggest fire?
91 00:16:22.410 ⇒ 00:16:22.860 Uttam Kumaran: Yeah.
92 00:16:22.860 ⇒ 00:16:25.290 Katherine Bayless: It tends to… it gravitates to the edges, right? You’re either…
93 00:16:25.290 ⇒ 00:16:25.610 Uttam Kumaran: that.
94 00:16:25.610 ⇒ 00:16:30.770 Katherine Bayless: It’s either where it’s coming in or where it’s going out, and in this case, the going out was where I wound up.
95 00:16:30.770 ⇒ 00:16:31.180 Uttam Kumaran: Okay.
96 00:16:31.180 ⇒ 00:16:34.759 Katherine Bayless: So, like, marketing, the data team from marketing is…
97 00:16:35.090 ⇒ 00:16:46.019 Katherine Bayless: kinda sorta where my headcount came from. Okay. But, you know, so now I’m a proper data ops team tasked with, you know, turn the dumpster fire into something beautiful.
98 00:16:46.020 ⇒ 00:16:55.319 Katherine Bayless: So I’ve been doing a lot of work with the marketing team, and now with CES registration having opened, and we’re kind of in that season, I’m working really heavily with that team.
99 00:16:55.620 ⇒ 00:17:13.020 Katherine Bayless: But everything’s pretty blank slate, honestly. The marketing team had the closest thing to a central sort of data warehouse or CRM that we have. Like I said, we have kind of a multi-CRM environment, and so they had this old SQL server. I have… I think this week, I will finally just deprecate it. I’ve been doing…
100 00:17:14.339 ⇒ 00:17:16.039 Katherine Bayless: maybe I’ll need one more thing, right?
101 00:17:16.039 ⇒ 00:17:35.629 Katherine Bayless: So, shutting that down, that’s Azure-hosted. I don’t want to deal with that anymore. I’m an AWS girl, and so what I’ve started building out is a Postgres sort of, like, faux CRM, because truly the organization needs one. Like, there are so many business rules that are going into not just the analytics, but just using the data and moving it between systems.
102 00:17:35.639 ⇒ 00:17:55.599 Katherine Bayless: And so I’m using a Postgres instance for the moment. We’re gonna be onboarding Snowflake in a couple weeks, because we have to use it in order to connect to the data from one of our vendors. Okay. We could write the API calls, but the data share via Snowflake, it’ll give us the data four times. Very, very nice. Yeah, exactly. It’ll be a faster path.
103 00:17:55.600 ⇒ 00:18:03.279 Uttam Kumaran: Something’s getting a bit bloated. Like, we have some clients that it’s perfect. For a lot of our clients, though, we’ve started using Mother Duck because it’s very cheap.
104 00:18:03.280 ⇒ 00:18:04.010 Katherine Bayless: Yeah.
105 00:18:04.010 ⇒ 00:18:09.169 Uttam Kumaran: And they all… they have, like, 20 tables, so I’m like, perfect.
106 00:18:09.170 ⇒ 00:18:09.550 Katherine Bayless: Have you?
107 00:18:09.550 ⇒ 00:18:23.280 Uttam Kumaran: It’s probably in a different spot, but I think it’s nice that there’s some options now, like, you can… and I’ve used Snowflake my whole career, so I’m a big fan, and yeah, you can basically route anything, and the data sharing is getting.
108 00:18:23.510 ⇒ 00:18:39.630 Katherine Bayless: really, really nice, you know. Yeah, I’ve actually never used Snowflake before. I’ve always been a Redshift girl, and then with this one, I was like, actually, I kind of want proper Postgres, because I’m tired of dealing with antiquated Redshift, and some of the, you know, some of the things that it’s been, you know, quirky for a while.
109 00:18:39.630 ⇒ 00:18:48.799 Uttam Kumaran: Yeah, I mean, I use it… I’ve been using it for 10 years, and it’s the same. It’s, like, mostly the same. When I first started using it, when I started my career, I’m still, like.
110 00:18:48.860 ⇒ 00:18:52.519 Uttam Kumaran: Having to remember to do, like, disk keys, and, like, and…
111 00:18:52.540 ⇒ 00:19:09.479 Uttam Kumaran: I have… like, last week, I set up, like, an auto vacuum for a client, and it’s, like, the same old, but I got used to Snowflake, where you don’t have to do any of that, which got really, really nice. But they are running… they’re running a lot of stuff, and it’s, like, dirt cheap for them, so…
112 00:19:09.480 ⇒ 00:19:09.860 Katherine Bayless: Yeah.
113 00:19:09.860 ⇒ 00:19:11.370 Uttam Kumaran: It’s really nice, but…
114 00:19:11.890 ⇒ 00:19:16.169 Katherine Bayless: Yeah, so yeah, so Snowflake, so we’ll probably… the Postgres thing, I…
115 00:19:16.860 ⇒ 00:19:21.929 Katherine Bayless: I will probably wind up keeping around. It’ll eventually maybe get merged into Snowflake, but…
116 00:19:22.230 ⇒ 00:19:36.850 Katherine Bayless: The next big piece that we’re kind of anchored to at the moment is Power BI. We have a really extensive deployment of dashboards and reports. I pulled the plug on all the data that connects to them and nobody’s noticed yet, which I’m thinking gives us a lot of wiggle room to, like.
117 00:19:37.750 ⇒ 00:19:46.769 Katherine Bayless: The reality is, we’ll probably have to rebuild on Power BI for the term, just because it’s what’s already, like, people are familiar with it, we already have the licenses, etc, etc.
118 00:19:46.770 ⇒ 00:19:47.360 Uttam Kumaran: Oh, God.
119 00:19:47.360 ⇒ 00:19:50.739 Katherine Bayless: I have no desire to stay on Power BI after 2025.
120 00:19:50.740 ⇒ 00:20:00.229 Uttam Kumaran: I’m a Tableau person, but I also hate everything Salesforce has done to Tableau, so I’ve been looking at either Looker or some of the other… Did you consider Omni, if you haven’t looked at Omni?
121 00:20:00.230 ⇒ 00:20:09.509 Katherine Bayless: I hooked it on me as well. And also, AWS is, like, QuickSite, and then part of me is like, can we just skip that part? Like, I feel like it’s 2025, AI is here, like, can’t we just.
122 00:20:09.510 ⇒ 00:20:15.629 Uttam Kumaran: That’s why, yeah, I agree, like, the reason I’ve been liking Omni is it’s sort of a mix of Looker and Tableau.
123 00:20:15.630 ⇒ 00:20:16.260 Katherine Bayless: Nope.
124 00:20:16.260 ⇒ 00:20:18.320 Uttam Kumaran: Look at the worksheet tabs.
125 00:20:18.440 ⇒ 00:20:24.499 Uttam Kumaran: and you get, like, the data and the visualization, like, I’ve been doing a lot on Omni the last past few weeks.
126 00:20:24.500 ⇒ 00:20:24.900 Katherine Bayless: Yeah.
127 00:20:24.900 ⇒ 00:20:27.139 Uttam Kumaran: It comes out of the box with the AI piece.
128 00:20:29.350 ⇒ 00:20:51.470 Katherine Bayless: where I kind of wanted to go, honestly, is, like, and admittedly, I am a builder at heart, and so I have to remind myself that I don’t have to build all the things, but I’ve got a line going with the AWS ProServe team, just to kind of buy some speed here, but, like, I’ve built this out before for other clients, like, a bedrock to Slack, basically.
129 00:20:51.470 ⇒ 00:20:52.669 Uttam Kumaran: Yeah, yeah, yeah, exactly.
130 00:20:52.670 ⇒ 00:21:04.150 Katherine Bayless: Like, most of the things people are using those Power BI reports for is to look up a person, or look up a country, or look up a company, right? Like, they’re not actually looking at, like, you know, bar charts and pie charts and, you know.
131 00:21:04.150 ⇒ 00:21:06.920 Uttam Kumaran: filtering to somebody to get their ID, or to get… yeah.
132 00:21:06.920 ⇒ 00:21:31.139 Katherine Bayless: Right, like, they have questions that need answering, and so I’m like, okay, well, rather than have them go and have to, like, self-serve in a dashboard that’s built to answer 10,000 different questions, why can’t we just have them go right where they’re already working and ask a natural question against the data and get the response they need? I think not only is that a better experience for the user, but also better for security, because we have a lot of interns, and if I found out I could get Jensen Fuong’s cell phone number, I’d probably take that and just put my internship
133 00:21:31.560 ⇒ 00:21:32.910 Katherine Bayless: You know? Right? Right.
134 00:21:32.910 ⇒ 00:21:44.779 Uttam Kumaran: I mean, I… that’s a similar thing we’re pitching internally, is like, I want to have… I mean, again, it’s like, just have an MCP server for writing, issuing the query, and then… I mean, again, it’s sort of like…
135 00:21:45.260 ⇒ 00:21:49.850 Uttam Kumaran: It just depends on how… how many side quests you want to take on, because…
136 00:21:49.850 ⇒ 00:21:55.619 Katherine Bayless: Part of it is, like, there’s still maybe questions that you want to then go onto a dashboard to then get.
137 00:21:56.230 ⇒ 00:22:12.219 Uttam Kumaran: Omni came out of the box with MCP, and we’re gonna try it, because our semantic understanding will kind of, for one of our clients, will stay in there, which is, like, what are the tables available? You can also put in descriptions, and all the metadata will stay there, and then I could call that MCP in Slack.
138 00:22:12.530 ⇒ 00:22:12.930 Katherine Bayless: Yeah.
139 00:22:12.930 ⇒ 00:22:15.770 Uttam Kumaran: And then, they could still use that for…
140 00:22:15.910 ⇒ 00:22:23.860 Uttam Kumaran: like, Viz, and so all that gets retained there, versus, like, if you were to just hit a table, you’d have to maintain
141 00:22:23.880 ⇒ 00:22:35.700 Uttam Kumaran: comments and descriptions, like, I think the metadata piece is the tough part. The other thing that we were, you know, I gave a talk about this a few weeks ago, is, like, the context that’s in, like, meetings and in Slack is also…
142 00:22:35.700 ⇒ 00:22:49.610 Uttam Kumaran: in wherever you guys are doing documentation is also certainly something to consider bringing in. I think that’s also mostly a data problem, but it’s like, can you categorize all the meetings or the notes associated with a topic so that when you query the
143 00:22:49.690 ⇒ 00:22:55.570 Uttam Kumaran: customers table, all, like… it can also, like, do a quick rag or retrieval over a bunch of documents, right?
144 00:22:56.010 ⇒ 00:23:04.300 Katherine Bayless: Yeah, and that’s where I’m hoping to bring Box in next year. Jenny, rightly, was like, I’m not launching an enterprise file tool, three months before CES, and I was like.
145 00:23:04.300 ⇒ 00:23:04.790 Uttam Kumaran: Yeah.
146 00:23:04.790 ⇒ 00:23:17.999 Katherine Bayless: Yeah, yeah, right. So yes, I think we’ll bring on Box next year, which is a tool that I’m just, yeah, I’m a fangirl for, and it’s exactly that. Like, if everybody’s parking stuff there already, and then we have the intelligent metadata and tagging happening, all of that richness comes.
147 00:23:18.000 ⇒ 00:23:32.329 Uttam Kumaran: Yeah, I guess tell me why… why Box is… I don’t know, I guess I haven’t… that’s the part of the data world where, like, for structured file storage, I haven’t done a lot. Mostly, we just… everything gets tossed in S3, and then we pipe stuff in, but, like, curious what… what do you like about Box?
148 00:23:32.490 ⇒ 00:23:46.510 Katherine Bayless: I like Box because they’ve really leaned into not just, like… so, like, some of the, I feel like, the AI platforms, like Glead and things like that, trying to be, like, organizational knowledge, like, they’re… they’re focused on the connectors, like, how much can we dominate.
149 00:23:46.510 ⇒ 00:23:47.440 Uttam Kumaran: Yeah, yeah, yeah, yeah.
150 00:23:47.440 ⇒ 00:23:48.220 Katherine Bayless: Right?
151 00:23:48.330 ⇒ 00:23:50.749 Katherine Bayless: Which is great, except that
152 00:23:51.350 ⇒ 00:23:56.850 Katherine Bayless: you know, what, like, 1% of your content is actually useful? Like, there’s a lot of noise relative to Signal.
153 00:23:56.920 ⇒ 00:24:15.050 Katherine Bayless: And Box is trying to solve for the other piece of it. Like, how do we get meaning out of this data? Not just, like, how can we get you to throw it in as easy as possible, but, like, what sort of deep-type things can be happening in the background so that you’ve got easy pulling out of context, even from, you know, page 256, a 20-year-old PDF file.
154 00:24:15.050 ⇒ 00:24:23.749 Uttam Kumaran: And so, like, that’s what intrigues me about Box. And also, admittedly, I’ve worked with them before, and I think it’s a nice platform. Their APIs are good, they’re easy to work with, like.
155 00:24:23.750 ⇒ 00:24:27.309 Katherine Bayless: Because we’re on Google Drive for all of our stuff sucks. Search sucks.
156 00:24:27.310 ⇒ 00:24:33.960 Uttam Kumaran: But we’re gonna just build… I’m just gonna build something on top of… on top of that, like, it’s… but it’s horrible, like.
157 00:24:33.960 ⇒ 00:24:56.030 Katherine Bayless: You should check out Box. They actually, they announced some stuff, their conference last week or the week before that’s pretty cool. I forget the… what is it? Box Extract or something like that? They don’t great at names, to be honest. But yeah, like, I think their… their work trying to figure out, like, the intelligent metadata and surface context from deep in your data lake, I think is really compelling.
158 00:24:56.290 ⇒ 00:25:07.450 Uttam Kumaran: I think you also like a lot of the AI features within Snowflake. They’ve… you can actually, within queries now, call LLMs, have them structure jobs, so again, if…
159 00:25:07.660 ⇒ 00:25:26.459 Uttam Kumaran: like, you may take a bunch of customer info and say, like, produce this email or something like that, and actually do that nicely within Snowflake. I still think that not many people are using it, because I just think people are solving, like, their usual day-to-day problems, but they’ve released a lot of AI stuff that I feel like is really cool in Snowflake, yeah.
160 00:25:26.460 ⇒ 00:25:35.240 Katherine Bayless: Yeah, a friend of mine is, works for a company that does, like, lead gen as a service kind of thing out of Baltimore, and they have been using a lot of that stuff really heavily in Snowflake. Oh, great.
161 00:25:35.240 ⇒ 00:25:53.349 Uttam Kumaran: like, as soon as I get it in here, I totally want to, like, sit down with them and be like, okay, show me exactly what you’re using. Yeah. Because I do think there’s a lot of stuff that’s in there that I haven’t, even… Calling an LLM at query runtime is interesting. You can just, like, combine a bunch of things. And for me, its biggest thing is, like, I want to relate… we have some, like, crazy regex or…
162 00:25:53.810 ⇒ 00:26:02.779 Uttam Kumaran: case when statements that are like, I just wish I could pass it to an LLM to make a determination. I’m like, perfect, we’re gonna replace all of those with this.
163 00:26:02.780 ⇒ 00:26:03.200 Katherine Bayless: Right?
164 00:26:03.200 ⇒ 00:26:03.630 Uttam Kumaran: Yeah.
165 00:26:03.630 ⇒ 00:26:14.590 Katherine Bayless: Actually, that’s a good segue. So, I think, in addition to the challenge we face with tiny tools that… I mean, like, frankly, a lot of the APIs I’m going to be writing are SOAP.
166 00:26:14.960 ⇒ 00:26:15.630 Uttam Kumaran: Yeah.
167 00:26:15.630 ⇒ 00:26:24.449 Katherine Bayless: like, a lot of our vendors, if they have an API, they don’t have a REST API. Yeah. And if they even have a REST API, they have these, like, weird, like, you can only query it 100 times a day, and I’m like…
168 00:26:24.450 ⇒ 00:26:33.159 Uttam Kumaran: Same thing, and so what we’ve actually… so we have a client that’s pulling stuff from, like, Uber Ads, you know, API, so we actually built… have you heard of BrowserBase before?
169 00:26:33.160 ⇒ 00:26:34.000 Katherine Bayless: Yeah, yeah, yeah, yeah.
170 00:26:34.000 ⇒ 00:26:37.920 Uttam Kumaran: We build a browser-based scraper that logs in.
171 00:26:38.070 ⇒ 00:26:45.280 Uttam Kumaran: And then, actually, we scrape values from the DOM. There’s no alternative. They don’t have… they don’t have any…
172 00:26:45.380 ⇒ 00:26:52.450 Uttam Kumaran: way of giving us the data. I guess maybe if we were spending, like, a bajillion dollars, the client was spending, maybe they would hook us up, but…
173 00:26:52.450 ⇒ 00:26:53.160 Katherine Bayless: Right.
174 00:26:53.160 ⇒ 00:27:07.159 Uttam Kumaran: there’s no… so, but again, like, yeah, we… we see that type of stuff all the time, or yeah, where there’s, like, some really, like, undocumented API, and then you’re like, none of these parameters work, and they’re like, oh, try this, try this.
175 00:27:07.160 ⇒ 00:27:08.469 Katherine Bayless: Yeah, the Oh, try this, isn’t.
176 00:27:08.470 ⇒ 00:27:09.220 Uttam Kumaran: Yeah, yeah.
177 00:27:09.220 ⇒ 00:27:27.670 Katherine Bayless: Well, so what I was gonna say, though, is, like, yes, that, but then also, because we have all these different systems, I mean, our… the identity resolution piece, right? Like, that’s a huge thing on my mind. Like, I think I’ll be able to limp through this CES season on, like, this is how we’ve made those calls in the past.
178 00:27:27.950 ⇒ 00:27:31.930 Katherine Bayless: But, like, going forward, we’ve got to have better data, and I think…
179 00:27:32.050 ⇒ 00:27:42.140 Katherine Bayless: it’s two pieces, right? Like, so one is the data coming in from our systems that we can only get from our systems could be cleaner, could be better harmonized. Those are problems to be solved.
180 00:27:42.140 ⇒ 00:27:59.339 Katherine Bayless: But then I’m also… I’m cognizant of the fact that, like, we track a lot of information about companies, which are… it’s probably public data, and so instead of relying on our people to have to key the exact URL, or to figure out every URL that goes with Google, because as a member company, their subsidiary companies also get membership, right?
181 00:27:59.580 ⇒ 00:28:12.789 Katherine Bayless: it’s just a lot of manual work, and right now it’s mostly just being farmed out to interns, but I’m like, there’s gotta be better ways. Like, I know we could purchase data Dun & Bradstreet, we could maybe crawl SEC filings, right? A lot of these are publicly traded companies.
182 00:28:12.800 ⇒ 00:28:19.779 Katherine Bayless: But I think solving for how we get good, reliable, trustworthy data on the companies.
183 00:28:19.780 ⇒ 00:28:20.350 Uttam Kumaran: Yeah.
184 00:28:20.350 ⇒ 00:28:28.820 Katherine Bayless: I think looking to external sources for that makes a lot of sense, and then incorporating that back in so that we’re matching the right company to the right record in all of these different systems.
185 00:28:28.840 ⇒ 00:28:40.790 Katherine Bayless: And then on the people side, I think it’s more of the sort of, you know, traditional identity resolution, but ideally with more modern tools, just to figure out, you know, the Catherine over here is the Catherine over here, is the Catherine over there, kind of thing.
186 00:28:41.080 ⇒ 00:28:47.009 Uttam Kumaran: Yeah, I mean, on the company enrichment side, 100%, I mean, for most of my career, I’ve been working on…
187 00:28:47.130 ⇒ 00:28:56.529 Uttam Kumaran: again, sort of, like, enrichment and, you know, waterfall level enrichment, things like D&B, but also, more recently, this company called People Data Labs, which.
188 00:28:56.530 ⇒ 00:28:57.080 Katherine Bayless: Yeah.
189 00:28:57.080 ⇒ 00:29:06.879 Uttam Kumaran: really, really amazing from. There’s… I mean, we’ve used Apollo, Clearbit, ZoomInfo, all that stuff. So, I think, certainly, if you’re having people look up, like.
190 00:29:07.110 ⇒ 00:29:20.469 Uttam Kumaran: funding announcements, revenue data, people, like, who are the titles? And People Data Labs is really good stuff, like, they have signals based on, like, different growth in different parts of their organizations. So, I think, certainly.
191 00:29:20.840 ⇒ 00:29:29.090 Uttam Kumaran: you should just get it from somebody, it’s gonna be worth it. And then… and then on the… yeah, on the broader, like.
192 00:29:29.280 ⇒ 00:29:46.619 Uttam Kumaran: Identity resolution piece, 100%. I mean, we have clients, too, that the same way. But, you know, for some folks, we’ve just gone ahead and set up, like, a thing like Segment, or we’ve set up… set them up with CDP. But again, it’s like a build versus buy, and sometimes we just came in and that’s the situation. So it’s…
193 00:29:46.620 ⇒ 00:29:53.439 Uttam Kumaran: it’s sort of… I think, at your stage, you could totally build some, like, light version of that, you know, that just does exactly what you need, versus…
194 00:29:53.640 ⇒ 00:29:56.080 Uttam Kumaran: They’re gonna charge an insane amount, you know?
195 00:29:56.080 ⇒ 00:30:02.169 Katherine Bayless: Right. That’s exactly where my head’s been at, is like, I… a segment comes up in my brain often, but the reality is I’m gonna have to…
196 00:30:02.170 ⇒ 00:30:05.739 Uttam Kumaran: Pricing is horrible. Pricing is so bad, and so, yeah.
197 00:30:05.790 ⇒ 00:30:23.160 Uttam Kumaran: And then, I’m also curious about how you guys are… what are you doing, like, offline? Like, how are the offline signals from the events actually coming back? Like, is that something that you’ve… you’ve thought about? My only reason is I… I used to work at this company called Flowcode. They do, like, circular QR codes.
198 00:30:23.160 ⇒ 00:30:25.499 Katherine Bayless: We see them, like, on TV, or…
199 00:30:25.830 ⇒ 00:30:41.839 Uttam Kumaran: your restaurants and stuff, but I, like, built a lot of the data platform there, and we did a lot of… our whole pitch was, like, offline to online, which is, like, getting offline signals. A lot of our clients were, like, event vendors and people that wanted to track, like, foot traffic and things like that.
200 00:30:41.840 ⇒ 00:30:42.690 Katherine Bayless: Oh,
201 00:30:42.690 ⇒ 00:30:52.609 Uttam Kumaran: It could be foot traffic, but it could also be getting people from an offline experience into, like, an online lead funnel, or a form, or some conversion event. I don’t know if you guys are doing anything.
202 00:30:52.610 ⇒ 00:30:58.149 Katherine Bayless: Just curious, like, wise from the floor itself. Yes, I have time, but just in case you need.
203 00:30:58.150 ⇒ 00:30:59.569 Uttam Kumaran: Oh, I do have time, yeah, yeah.
204 00:30:59.570 ⇒ 00:31:16.200 Katherine Bayless: Okay, yeah, I can go for maybe another 15 minutes or so. Yeah, so right now, we do have a couple things. So, we do have, you know, badge scanners, two versions, one that we use to check people into sessions and events, and then another that we give to the exhibitors for, like, lead gen. We…
205 00:31:16.330 ⇒ 00:31:32.159 Katherine Bayless: we have that data somewhere. And then we also have, like, the mobile app, and we have these maps through a company called Pointer, which is kind of like the beacon technology, turn-by-turn directions in Vegas. So we have a lot of data there.
206 00:31:32.210 ⇒ 00:31:45.880 Katherine Bayless: I haven’t had a chance to really work with it from the prior years. I would also say that one of our biggest challenges is that because we are not very technical internally, we lean very heavily on vendors, which means we don’t tend to, like.
207 00:31:46.110 ⇒ 00:31:48.829 Katherine Bayless: Bring that data back in for use outside of the…
208 00:31:48.830 ⇒ 00:31:49.290 Uttam Kumaran: Oh, I see that.
209 00:31:49.290 ⇒ 00:31:54.500 Katherine Bayless: So, like, we’ll get a report from the vendor on how many people went, you know, here, there, and everywhere, but, like, we don’t…
210 00:31:54.700 ⇒ 00:31:57.110 Uttam Kumaran: There’s no, like, well, where’s the API key, please?
211 00:31:57.110 ⇒ 00:32:21.290 Katherine Bayless: Right, exactly, and so, like, I’ll be the first person to change that, but I’m kind of of the mindset that, like, probably there’s more potential going forward than I am interested in the historical data to a certain extent, but the fact that our attendees are already used to, you know, mobile app, a beacon system, badging into rooms, like, I think the ability to collect that information is really strong. I’m just not sure if the historical
212 00:32:21.290 ⇒ 00:32:23.439 Katherine Bayless: Stuff is really any good that we have.
213 00:32:23.650 ⇒ 00:32:33.589 Uttam Kumaran: Great. And then what are, like, what are the growth KPIs, like, for the business right now? Like, what are you coming into, like, when you gotta try to affect XYZ?
214 00:32:33.910 ⇒ 00:32:40.470 Katherine Bayless: So, the big thing, I mean, you know, like any good business, right, the thing that makes us the most money, we would like to grow that thing.
215 00:32:40.470 ⇒ 00:32:41.960 Uttam Kumaran: More money, less cost, yes.
216 00:32:41.960 ⇒ 00:32:56.769 Katherine Bayless: Yeah, exactly. But what we’re really focusing on at the sort of, like, you know, five-year plan kind of level is we want to diversify the revenue so that the split between the show and everything else is, you know, we’re not going to get to 50-50, but if we can get, you know, closer to kind of, like.
217 00:32:56.770 ⇒ 00:32:57.690 Uttam Kumaran: Interesting, okay.
218 00:32:57.690 ⇒ 00:33:00.470 Katherine Bayless: 70-30, 80-20 kind of mark, right?
219 00:33:00.490 ⇒ 00:33:17.840 Katherine Bayless: So we’d like to grow the show, but also grow the revenue in the other areas as well. And so that’s at the top level, kind of where we’re looking. But, I mean, reality, truthfully, on the ground is, like, I mean, in terms of, to your point, right, constrained costs, like, I think the tech debt is costing us.
220 00:33:18.090 ⇒ 00:33:18.720 Uttam Kumaran: Yes.
221 00:33:18.720 ⇒ 00:33:24.300 Katherine Bayless: unimaginable sum of money. And some of that’s in my world, some of that’s in Jay’s world with IT, but…
222 00:33:24.300 ⇒ 00:33:34.979 Uttam Kumaran: Like, I think, yeah, I think the tech debt is really debt, debt. That’s why when you said 2025, I was like, good luck, because once November hits, people are out.
223 00:33:34.980 ⇒ 00:33:36.509 Katherine Bayless: Great.
224 00:33:36.810 ⇒ 00:33:49.440 Katherine Bayless: I know, it’s funny, like, when I was talking to the ProServe guys, because I’ve done projects with them before, and so, like, laughing, and they’re like, girl, by the time we get a contract through your team and our team, it’s gonna be mid-December, and I’m like, well, you better work fast!
225 00:33:49.440 ⇒ 00:34:02.530 Uttam Kumaran: Yeah, yeah, that’s great, that’s great. No, I mean, we have the same part, but, you know, it’s sometimes tough for the business folks to understand how much tech debt there really can be, especially an organization that probably bought many tools.
226 00:34:02.530 ⇒ 00:34:12.969 Uttam Kumaran: and is now stitching them together. You probably have people who are sort of born into this, like, messed up environment, so all their habits are really bad. There’s no, like, develop… there’s no, like, development or developer…
227 00:34:12.969 ⇒ 00:34:16.450 Uttam Kumaran: Process or ecosystem to, like, promote good habits, like…
228 00:34:16.639 ⇒ 00:34:21.880 Uttam Kumaran: I… we have the same problem where I go into clients, and they want us to just, like, push new stuff. I’m like, this is…
229 00:34:21.880 ⇒ 00:34:37.489 Uttam Kumaran: you can’t expect us to work under these conditions, basically. I’m like, I will… we will continue to make mistakes just like your past people did, and they’re gonna… they left for… for a reason, because… and maybe they didn’t have the tools or the cachet to do this, but, like.
230 00:34:37.489 ⇒ 00:34:44.039 Uttam Kumaran: we need version control, I need alerting, I need CICD, I need these, like, I can’t… I just… yeah, it’s like a
231 00:34:44.360 ⇒ 00:34:47.359 Uttam Kumaran: Like, I’m, like, gonna go on strike, yeah, so…
232 00:34:47.360 ⇒ 00:34:55.849 Katherine Bayless: Like, day one, I walked in, and I’m like, where is your database? They’re like, they called it Toad, because they used Toad, the GUI.
233 00:34:55.870 ⇒ 00:34:56.359 Uttam Kumaran: Yeah, yeah.
234 00:34:56.360 ⇒ 00:35:03.430 Katherine Bayless: Yeah, yeah, so they’re like, yeah, it’s in code. And I’m like, oh. And then I’m like, where’s your code? And it’s on a mapped network drive.
235 00:35:04.430 ⇒ 00:35:06.930 Uttam Kumaran: Great, yeah, yeah.
236 00:35:06.930 ⇒ 00:35:16.570 Katherine Bayless: called GitHub, we’re putting a code in there, no. So it’s like, it’s funny, these things, they… to your point, if it’s not a technical team, they’re just not familiar with it, so I do have to be gentle, right, and say.
237 00:35:16.570 ⇒ 00:35:34.410 Uttam Kumaran: Totally, but also, it’s gonna affect retention and the happiness of everybody involved. Many people, when we walk into situations, like, we tend… we were open, because I said, like, if you have a, like, sometimes we go into clients where they have a stray analyst or a stray data engineer, I’m like, you join our stuff, I’ll show you what it’s like to have, like, we have stand-ups, we do planning, we have, like.
238 00:35:34.410 ⇒ 00:35:35.350 Katherine Bayless: Yep, same, exactly.
239 00:35:35.350 ⇒ 00:35:49.389 Uttam Kumaran: We have people who are, like, we have triage, we then have someone who’s on-call, like, you’ll get a sense of what it’s like to be part of a real data team, and then it’s also, like, playing defense, like, how do we actually defend against stuff that’s not important versus important?
240 00:35:49.390 ⇒ 00:36:01.389 Uttam Kumaran: And then those people tend to stay, and then they get better, right? And that’s like a… that’s a positive ROI for the business, versus that person one day may just make a mistake, get yelled at by somebody, and they’re like, F it, I’m outta here, you know?
241 00:36:01.390 ⇒ 00:36:05.559 Katherine Bayless: what has happened to at least 3 people since I started.
242 00:36:05.810 ⇒ 00:36:16.570 Uttam Kumaran: Yeah, and I… I mean, look, it’s… this is just what happens in data. I don’t know, it’s the same… it’s the same story, so I’m… I’m really glad to hear that you’re… you’re, like, in there shaking things up, so…
243 00:36:16.730 ⇒ 00:36:23.179 Katherine Bayless: I’m trying. I’m like, I try to shake gently, but I’m also… I’m not a very gentle person.
244 00:36:24.760 ⇒ 00:36:28.790 Katherine Bayless: I was talking to a friend of mine over the weekend, he’s like, you’re very impatient, you know that, right? And I was like.
245 00:36:28.790 ⇒ 00:36:39.480 Uttam Kumaran: That’s a good… that’s… that’s good, like, I don’t know, I’m also very… I guess I’m patient if I’m, like… I guess, I don’t know, I’m not really patient, honestly, at all, either, really. I don’t know, I just…
246 00:36:39.480 ⇒ 00:36:42.350 Katherine Bayless: Someone is like, there is a solution for this, like, I can’t walk.
247 00:36:42.350 ⇒ 00:36:50.160 Uttam Kumaran: In my life, I’ve now seen these so many times, so some things… the AI stuff is what’s actually really interesting to me these days, because
248 00:36:50.260 ⇒ 00:36:59.729 Uttam Kumaran: being able to build a semantic knowledge. When we come to a company, we’re the ones that tend to document everything, we build this great knowledge repository, but again, it just…
249 00:36:59.770 ⇒ 00:37:18.960 Uttam Kumaran: documentation does what it usually does, just sits. Now that we actually have a reason for it to get activated. For me, now every client we go into, one of the things is, like, how do we leave them with, like, an MCP on top of their data, or… not only will we ship a dashboard, we ship, like, a chat co-pilot alongside of it. Just, like, in case you want it, it’s there.
250 00:37:19.000 ⇒ 00:37:23.410 Uttam Kumaran: Like, and we try to, like, promote those things, but it’s so brand new that
251 00:37:23.500 ⇒ 00:37:35.289 Uttam Kumaran: We’re also just, like, figuring out, like, what works. Basic questions are okay, and then Omni’s thing is actually really good, you know, and… but some of these guys, I think, some of the vendors are not so good for, like, Texas Equal.
252 00:37:35.540 ⇒ 00:37:36.040 Katherine Bayless: Right.
253 00:37:36.040 ⇒ 00:37:39.309 Uttam Kumaran: Things like that. And also, you can build a lot of it on your own.
254 00:37:39.520 ⇒ 00:37:40.450 Katherine Bayless: Using old.
255 00:37:40.450 ⇒ 00:37:42.189 Uttam Kumaran: Of course, pretty, pretty easily.
256 00:37:42.190 ⇒ 00:37:57.279 Katherine Bayless: Yeah, and I actually, I do think, like, maybe, you know, later, kind of next year-y sort of thing, like, I think some small language models would actually make a lot of sense for a couple of the use cases we have internally. Like, we have a… believe it or not, we have a library. Like, literal books on shelves.
257 00:37:57.280 ⇒ 00:37:58.210 Uttam Kumaran: Oh, great!
258 00:37:58.210 ⇒ 00:38:07.140 Katherine Bayless: And what’s interesting is I think people pick on them for being like, I already have a library, but I’m like, are you kidding? They have data on, like, the industry going back to the early 1900s.
259 00:38:07.140 ⇒ 00:38:08.919 Uttam Kumaran: Yeah, we need to… Is that the moose?
260 00:38:08.920 ⇒ 00:38:12.629 Katherine Bayless: Urgent cold mine? No. Is it interesting? Yes.
261 00:38:12.630 ⇒ 00:38:14.409 Uttam Kumaran: Yeah, we need to OCR all of this.
262 00:38:14.410 ⇒ 00:38:34.069 Katherine Bayless: Yeah, exactly, and so they’ve, I mean, for years they’ve been working on it, but I’m like, I would love, and I think I’ve finally taken the reins on that, I’m like, I would just love to actually kind of try and push that further over the finish line, and then figure out, like, yeah, the small language models, or whatever else we can do with it, because I think there is value in there, even if it’s, like, you know, it’s not going to put out a fire today.
263 00:38:34.070 ⇒ 00:38:35.250 Uttam Kumaran: Yeah, yeah.
264 00:38:35.500 ⇒ 00:38:57.649 Katherine Bayless: But, okay, well, let me ask you this question, just, like, from a more logistics perspective, because you talked about the team a little bit. So it’s… on my side, it’s me, hi. And then, I have a new data engineer who I stole from the market research team, and so he’s got kind of a data science background, mostly, like, R and SQL Server, but really smart. I don’t think he’ll have any trouble upskilling into, like, a true data engineer.
265 00:38:57.650 ⇒ 00:39:06.030 Katherine Bayless: Then I’m doing a little recruitment for another data engineer and a business intelligence analyst, just to kind of scale my ability to go sit with teams and understand processes, but…
266 00:39:06.030 ⇒ 00:39:09.929 Katherine Bayless: what sort of, like, model do you guys have? Like, if I wanted to, like.
267 00:39:10.120 ⇒ 00:39:20.049 Katherine Bayless: bring somebody in to, like, kind of join the team and work with us, or are you more in the, like, consulting and advising and some doing, or is it outsourced development? Like, what’s your.
268 00:39:20.050 ⇒ 00:39:43.079 Uttam Kumaran: Yeah, so… so all of our… so we have about 15 people, we have folks here in the States, kind of global, but it’s all, like, Brain Forge teams, so it’s… we don’t do, sort of, any subcontracting at all. It was all Brain Forge team members. We have folks kind of across the whole stack, so we have great, sort of, data engineers, folks that do analytics engineering, so traditional data modeling.
269 00:39:43.080 ⇒ 00:39:44.460 Uttam Kumaran: BI across
270 00:39:44.460 ⇒ 00:39:52.200 Uttam Kumaran: you know, name your tool, we’ve, like, we’ve played with it. And then also, like, product analytics, and then more, like, consulting-style BI.
271 00:39:52.200 ⇒ 00:40:04.509 Uttam Kumaran: probably the way I separate is there’s, like, dashboarding, and then there’s, like, tell me what the answer is type of analysis, and we found it hard to… that’s usually not one person, so I tend to, like, separate that out.
272 00:40:04.510 ⇒ 00:40:15.720 Uttam Kumaran: And then, yeah, we usually come in and morph. So, some clients, we come in, and we are their entire data team. Maybe they have one or two stragglers, or they’ve… they’re, like, an e-commerce company, and there’s no…
273 00:40:15.720 ⇒ 00:40:20.009 Uttam Kumaran: they just have an analyst, and that person’s, like, left to dry. So we come in and we basically
274 00:40:20.010 ⇒ 00:40:39.149 Uttam Kumaran: sort of build a function. There’s also other situations where we come in, we’re just part of the team. From our side, we have project management resources. Of course, we have folks at my level, like me, that are more solution architecture, and then a lot of doing. So we’re definitely not shy to the execution. I think compared to most
275 00:40:39.150 ⇒ 00:40:40.120 Uttam Kumaran: sort of…
276 00:40:40.220 ⇒ 00:40:55.149 Uttam Kumaran: like, data companies, we also want to be involved in, like, helping with strategy, and ideally taking some of those, like, that decision-making, or… for example, it’s like, go do a roadshow for vendors and find out what’s the best solution here.
277 00:40:55.150 ⇒ 00:41:05.380 Uttam Kumaran: analyze… like, taking on some of those higher-level head-of-data type decision making is what we love to do, and I feel like that’s sort of why we compete well, because I think there’s a lot of people that offer just, like.
278 00:41:05.490 ⇒ 00:41:13.350 Uttam Kumaran: if you need an analytics engineer, we’ll be there, we can write any SQL, that’s fine, but I think we also want to be involved in, sort of, like, more of the strategic side.
279 00:41:13.520 ⇒ 00:41:26.630 Uttam Kumaran: But we would totally plug in to your team if you just… if you guys are already running sprints and stuff, or if you want us to support you to establish that, that’s sort of what we do. So we… we run typically, like, just one week
280 00:41:26.790 ⇒ 00:41:31.679 Uttam Kumaran: We just do, like, one-week sprint cycles, and we try to get everybody on the same page.
281 00:41:31.900 ⇒ 00:41:43.969 Uttam Kumaran: And then, yeah, that’s usually how we… how we work. For a lot of companies where it’s not as clear as yours, like, I think you have a really good sense of where things are at, we usually do, like, a one-month audit, where we walk in the door and we do what you just did, which is, like.
282 00:41:44.440 ⇒ 00:41:57.639 Uttam Kumaran: tell me everything, and then we sort of propose, like, what the schema is of, like, what we want to do. In this situation, it seems like maybe it’s just a couple people that plug into your team, and help you, like, at a higher level with stuff.
283 00:41:57.850 ⇒ 00:42:01.790 Uttam Kumaran: Or even, like, again, executing at a lower level, so…
284 00:42:01.930 ⇒ 00:42:17.719 Katherine Bayless: Yeah, I mean, from what you’ve described, like, I hear a couple needs that would probably fit our use case. Like, definitely, I mean, two people against the world, right? Like, we could use all the help. But I think specifically, like, if there are people who are really deep in, like.
285 00:42:20.100 ⇒ 00:42:24.910 Katherine Bayless: Pipeline engineering, I just don’t… I mean, that’s a me skill set, but I’m supposed to be sitting.
286 00:42:24.910 ⇒ 00:42:25.360 Uttam Kumaran: Yes.
287 00:42:25.360 ⇒ 00:42:43.880 Katherine Bayless: strategizing and stuff, right? And so, like, a pipeline engineer that can work alongside me or, you know, take my random thoughts and go? Awesome. Somebody that could be, like, an analytics and BI engineer that could work really closely with the data engineer from the market research team as he’s kind of upskilling so that I’m not his only mentorship resource.
288 00:42:43.880 ⇒ 00:42:51.989 Katherine Bayless: That would be awesome, and I mean, really, the need for BI, I think, is gonna drown me in, like, 2 weeks. I think people are being nice right now, but, like, it’s gonna… it’s gonna.
289 00:42:51.990 ⇒ 00:42:56.370 Uttam Kumaran: No, and that’s also… there’s gonna be so many errors that, like, yeah, it’s… that’s gonna be…
290 00:42:56.610 ⇒ 00:43:00.180 Uttam Kumaran: And then, I think the strategy piece, like, I.
291 00:43:00.250 ⇒ 00:43:11.759 Katherine Bayless: I actually think would benefit me as well, honestly. Like, I mean, I definitely… I have a vision, I have a strategy, I have a, you know, a sense of where I want to get to, but I’m lonely, right?
292 00:43:11.760 ⇒ 00:43:13.970 Uttam Kumaran: It’s also true, like, yeah.
293 00:43:13.970 ⇒ 00:43:16.550 Katherine Bayless: And, like, I also… I don’t have…
294 00:43:16.550 ⇒ 00:43:37.110 Katherine Bayless: the right answers, and I know that there are pieces where I’m like, I know the strategy needs X, but as far as, like, what exactly X could be, there’s 10 different, you know, a million different options, and so, like, having somebody to help be a foil on the strategy side would be really beneficial, because otherwise, CTA’s gonna get, like, the Catherine flavor everything, which is, you know, I think a good flavor.
295 00:43:37.110 ⇒ 00:43:45.499 Uttam Kumaran: No, it’s very humble to hear that. I mean, I feel the same way, even in our business. I feel like me and my business partner, Robert, we’re that with each other. He comes from more of product analytics.
296 00:43:45.500 ⇒ 00:43:46.030 Katherine Bayless: Hmm, awesome.
297 00:43:46.030 ⇒ 00:43:52.629 Uttam Kumaran: I’m more data modeling, data engineering, and, like, I… we both worked with executives, so we sort of, like.
298 00:43:52.910 ⇒ 00:44:01.550 Uttam Kumaran: compete, because I think a lot more about, like, establishing systems and architecture. He’s more like, if that’s Excel and we can get a CSV over today, like, ship it.
299 00:44:01.550 ⇒ 00:44:01.920 Katherine Bayless: Okay.
300 00:44:01.920 ⇒ 00:44:07.539 Uttam Kumaran: We both help clients because we can do the end-to-end, and so certainly, I think…
301 00:44:07.750 ⇒ 00:44:23.240 Uttam Kumaran: I think it’s nice, because for a lot of our clients, we don’t even get to do a lot of the data procurement strategy or architecture, because they’re just like, whatever you decide, right? And given that you have a legacy, and you’re… you’re also opinionated, I think there’s… it could be really nice. And then certainly on the…
302 00:44:23.240 ⇒ 00:44:30.100 Uttam Kumaran: on the pipeline engineering, whether it’s, like, again, if you guys already have an established ETL tool, if you want to have an established
303 00:44:30.100 ⇒ 00:44:34.080 Uttam Kumaran: framework for bringing in stuff where there is a REST API,
304 00:44:34.280 ⇒ 00:44:48.950 Uttam Kumaran: and, like, we just need to connect to that where there’s not a REST API. There’s some great frameworks that got released, and setting up an orchestration tool like Dagster to orchestrate all that, like, that’s… that’s kind of, like, our bread and butter. That’s… that’s what we do, so…
305 00:44:48.950 ⇒ 00:44:51.820 Katherine Bayless: Yeah, I used glue really heavily at my last
306 00:44:52.400 ⇒ 00:44:58.940 Katherine Bayless: that was my primary sort of pipelining tool, and it’s funny, I kind of… I just don’t miss it.
307 00:44:58.940 ⇒ 00:44:59.500 Uttam Kumaran: Yeah.
308 00:44:59.500 ⇒ 00:45:10.610 Katherine Bayless: that bad, right? Like, it’s super powerful, and I feel like I’m very good at it. I mean, it took me… the learning curve was steep, but, like, I can… I can work with glue. I just wouldn’t mind if I didn’t have to.
309 00:45:10.610 ⇒ 00:45:25.859 Uttam Kumaran: Yeah, yeah, so I think, like, glue is an option, and Daxter is great, because any type of Pythonic Python you can write there. And then also, you know, within Snowflake, you can do a lot, so you may even find these days that Snowflake has tasks.
310 00:45:26.280 ⇒ 00:45:26.720 Katherine Bayless: Yeah.
311 00:45:26.720 ⇒ 00:45:30.039 Uttam Kumaran: And they have, sort of, orchestration built in.
312 00:45:30.290 ⇒ 00:45:40.759 Uttam Kumaran: like, one of the things that we haven’t tried… we haven’t been able to try with any client yet, but I’m pitching that you just… if most of your stuff is writing Python, you can actually orchestrate a lot within Snowflake.
313 00:45:40.760 ⇒ 00:45:41.650 Katherine Bayless: Yeah?
314 00:45:41.650 ⇒ 00:45:49.589 Uttam Kumaran: and call external APIs directly and kind of skip even having an external orchestration tool. So that’s something to look into.
315 00:45:49.590 ⇒ 00:45:51.450 Katherine Bayless: Okay. That’s actually…
316 00:45:51.450 ⇒ 00:45:54.330 Uttam Kumaran: Yeah, they have tasks, they have functions, they have alerts.
317 00:45:55.150 ⇒ 00:45:58.219 Uttam Kumaran: And, like, they built out kind of almost like a mini airflow type…
318 00:45:58.500 ⇒ 00:46:12.320 Katherine Bayless: Okay, that’s what I was thinking, it sounds like, yeah, interesting. Airflow and Airtable were ones that I had been, like, other folks had recommended, and I was like, I don’t know, they feel like they might be almost too much of a light touch. Because I am cognizant of the fact that, like, whatever I build.
319 00:46:12.920 ⇒ 00:46:15.849 Katherine Bayless: Most folks probably won’t know how to use, and so, like, the more users.
320 00:46:15.850 ⇒ 00:46:23.339 Uttam Kumaran: That’s also the case, is, like, you want to… that’s why glue would be, like, a lot.
321 00:46:23.340 ⇒ 00:46:29.190 Katherine Bayless: I struggled to onboard team members at my last place onto Glue, so yeah, the learning curve, yeah, formidable.
322 00:46:29.190 ⇒ 00:46:45.500 Uttam Kumaran: part of it is also, like, again, we… one of the things that I realized also is, like, when consultants come in and they, like, own and they gatekeep stuff is also, like, exact opposite of how we operate. I actually am, like, very happy to even help you hire those people, or train up the people, because
323 00:46:45.590 ⇒ 00:46:57.700 Uttam Kumaran: we don’t win unless you win. And so, even for those folks that are around you, we have clients where, as we’ve absorbed people into our sort of… into our sort of crew, they’ve elevated. Like, we’ve taught them how to use Cursor to develop.
324 00:46:58.180 ⇒ 00:47:15.439 Uttam Kumaran: set up local environments to test changes, understand the importance of, like, PR reviews, and, like, it’s great, because, like, they need to work well. It’s not like us versus them. Like, we all need to actually do well, and they come on and see what it’s like to work with people, and I think
325 00:47:15.440 ⇒ 00:47:20.549 Uttam Kumaran: At the stage you’re at, just establishing, like, the whole data foundation, it could be really helpful.
326 00:47:20.550 ⇒ 00:47:24.330 Katherine Bayless: Yeah, yeah. I know I’ve taken up a little totally.
327 00:47:24.330 ⇒ 00:47:26.120 Uttam Kumaran: Oh, this is great, I appreciate it.
328 00:47:26.120 ⇒ 00:47:28.320 Katherine Bayless: I, I will admit, I…
329 00:47:28.630 ⇒ 00:47:31.139 Katherine Bayless: I didn’t expect this to go this well.
330 00:47:31.140 ⇒ 00:47:31.900 Uttam Kumaran: Oh my god!
331 00:47:31.900 ⇒ 00:47:40.540 Katherine Bayless: I’m surprised, because I think my initial, read was, like, Christina kind of pitched it as, like, oh, they’re good at connecting to things, and I was like, oh, great, okay, I, yeah, sure, connectors.
332 00:47:40.540 ⇒ 00:47:59.259 Uttam Kumaran: Yeah, I’ve talked to the 100th person about it. No, I don’t know, I think we… I describe us as, like, we solve problems, but we do that with data and AI, if we can. If we can’t, I can recommend you a tool, or another friend, or, like, another agency that could do that, but, like, we come into situations because there are problems, and so for us, it’s…
333 00:47:59.260 ⇒ 00:48:01.590 Uttam Kumaran: To rec… to give an overview of, like.
334 00:48:01.590 ⇒ 00:48:20.469 Uttam Kumaran: what options could be and why, but it’s also to do that fast, and to do things that, like, can work on your pace. And ideally, to my team, I say we have to be the best communicating person, better than their most communicative employee. That’s our, like, goal. It’s also because consultants, we’re always on the chopping block, and so…
335 00:48:20.790 ⇒ 00:48:29.280 Uttam Kumaran: what I tell our team is that, like, we have to over-exceed, and we have to move fast, otherwise it’s not worth it, you know, for folks, and so…
336 00:48:29.280 ⇒ 00:48:41.979 Uttam Kumaran: Yeah, I’m happy to be helpful. So let me know, like, what do you… what do you think is… I can summarize some of this and send it over, and then, like, we can maybe email back and forth on, like, some scope, or what do you think is best?
337 00:48:42.120 ⇒ 00:49:01.679 Katherine Bayless: Ironically, I am the worst communicator. Email is where my brain just goes to die. I hate it. But I try, I try. I guess I would ask, in total fairness to you, I know sometimes I’m just, like, word vomit, so, like, if you want a more structured conversation, I totally am happy to schedule, like, a follow-up where we could maybe.
338 00:49:01.680 ⇒ 00:49:02.120 Uttam Kumaran: Sure.
339 00:49:02.120 ⇒ 00:49:25.620 Katherine Bayless: logically through this stuff, rather than leave you trying to digest the notes from this, you know, sort of all-over-the-place conversation, but I would definitely love to talk more and figure out, especially since you have a smaller team, like, realistically, what kind of timelines might be possible, just to even get started and that sort of thing. But, yeah, happy to do, like, a deep dive on, like, specific scopes of work that we could tackle together.
340 00:49:25.810 ⇒ 00:49:38.240 Uttam Kumaran: Okay, so what I’ll do is I’ll take this conversation, and I’ll just map out a couple of scopes. Like, we use Notion, so I’ll just write everything in one place, and then I’ll just… I’ll share that with you, and then we can hop back on, maybe sometime next week?
341 00:49:38.430 ⇒ 00:49:39.589 Katherine Bayless: Yeah, yeah, yeah, okay.
342 00:49:39.590 ⇒ 00:49:43.980 Uttam Kumaran: Yeah, same time next week, I could just try to put this on for then, and I’ll try to get.
343 00:49:44.190 ⇒ 00:50:00.690 Katherine Bayless: I think actually next Tuesday, I have, I think, yeah, I’m out next Tuesday, but, I can do… I mean, if you like Mondays, I can do Mondays at 2 or after, otherwise my Wednesday is totally open before 2PM.
344 00:50:00.970 ⇒ 00:50:07.390 Uttam Kumaran: Monday’s… Monday’s fine. That’s fine with you. I’d rather do as much stuff as early as possible.
345 00:50:07.390 ⇒ 00:50:08.760 Katherine Bayless: Yeah, yeah, totally, totally.
346 00:50:08.760 ⇒ 00:50:16.130 Uttam Kumaran: Yeah, cool, so Monday at 2 Eastern, I’ll just put something on, and then, yeah, I’ll just basically map out, like, we mentioned a couple things, I’ll just sort of
347 00:50:16.370 ⇒ 00:50:29.489 Uttam Kumaran: put all my thoughts there, and definitely have, like, open questions on things, and then we can just, like, run through that. I mean, I could definitely share you more context about the types of companies we’ve worked with, and, like, who’s on our team as well, and things like that, so…
348 00:50:29.490 ⇒ 00:50:34.020 Katherine Bayless: Yeah, that’d be awesome, and if you have anybody, I could talk to for, like, referrals, that kind of thing, just to.
349 00:50:34.020 ⇒ 00:50:34.800 Uttam Kumaran: Totally.
350 00:50:34.800 ⇒ 00:50:42.979 Katherine Bayless: Yeah, I could reach out to them and try and get those calls scheduled so that we’re not waiting until, you know, the next… let’s not do waterfall, let’s be agile here.
351 00:50:42.980 ⇒ 00:50:43.480 Uttam Kumaran: Okay, fair.
352 00:50:43.480 ⇒ 00:50:44.330 Katherine Bayless: Great, alright.
353 00:50:44.330 ⇒ 00:50:55.559 Uttam Kumaran: Yeah. Okay, well, this is great. Thanks. Very refreshing conversation. I’m glad. Usually, these are just more, like, trying to explain, like, why data is important, so I’m glad that you can talk.
354 00:50:55.560 ⇒ 00:50:58.030 Katherine Bayless: Instant, like, kindred spirit, honestly.
355 00:50:58.030 ⇒ 00:51:04.630 Uttam Kumaran: I mean, it’s also, it’s a lot of scarring. It’s a lot of scarring this year. So that’s mostly what it is.
356 00:51:04.630 ⇒ 00:51:08.709 Katherine Bayless: But it’s like, it’s the worst kind, because it’s like, I get the battle scars, but I just keep charging into battle.
357 00:51:08.710 ⇒ 00:51:12.160 Uttam Kumaran: I know, I’m like, here we go, like, but I don’t know, I just…
358 00:51:12.160 ⇒ 00:51:12.790 Katherine Bayless: I won’t.
359 00:51:12.790 ⇒ 00:51:21.319 Uttam Kumaran: here’s, like, such low-hanging fruit things that we just unlock. That’s a real joy, versus… I’ve been in companies where there’s, like, only, like, luxury problems.
360 00:51:22.140 ⇒ 00:51:35.240 Uttam Kumaran: And, like, I like doing the things where it’s like, oh, like, queries are slow, let’s just, like, cut all… cut that in by, like, 90%, and there’s just, like, some parameter I needed to change once. And, like, those are great things, you know?
361 00:51:35.240 ⇒ 00:51:45.410 Katherine Bayless: Right, exactly. Like, it’s funny, for a data person, my happy place is the engineering, not the analysis, and so I’m like, I just like doing the part where you guys can go use this information, whoever.
362 00:51:45.410 ⇒ 00:51:46.969 Uttam Kumaran: I’m the enabler, yeah.
363 00:51:46.970 ⇒ 00:51:54.959 Katherine Bayless: Exactly, exactly, exactly. Yeah, so it’s like, CTA has a lot… it’s a long road ahead, but it’s definitely the kind of road I enjoy.
364 00:51:55.200 ⇒ 00:52:00.349 Uttam Kumaran: Okay, perfect. Well then, yeah, look forward to talking next Monday, and thank you. Have a great week.
365 00:52:00.540 ⇒ 00:52:04.899 Katherine Bayless: Yeah, yeah, thank you so much, and I’m definitely looking forward to Monday. If you need anything in the meantime, just let me know.
366 00:52:04.900 ⇒ 00:52:06.669 Uttam Kumaran: Okay, perfect. Alright, thank you.
367 00:52:06.670 ⇒ 00:52:07.449 Katherine Bayless: Thank you, bye.