Meeting Title: Data Pipeline and Security Sync Date: 2026-02-05 Meeting participants: Katherine Bayless, Uttam Kumaran
WEBVTT
1 00:03:38.260 ⇒ 00:03:39.450 Uttam Kumaran: Hello!
2 00:03:39.850 ⇒ 00:03:40.740 Katherine Bayless: Hello.
3 00:03:41.300 ⇒ 00:03:42.330 Katherine Bayless: How’s it going?
4 00:03:43.020 ⇒ 00:03:44.040 Uttam Kumaran: Good, I just…
5 00:03:44.210 ⇒ 00:03:49.619 Uttam Kumaran: We’re… we have another client that’s, like, a big retailer, and we’re dealing with a bunch of retail.
6 00:03:50.090 ⇒ 00:03:50.550 Katherine Bayless: Mmm.
7 00:03:50.550 ⇒ 00:03:52.190 Uttam Kumaran: data endpoints.
8 00:03:52.480 ⇒ 00:03:53.250 Katherine Bayless: Yeah.
9 00:03:53.460 ⇒ 00:03:58.449 Uttam Kumaran: and… It’s, like, so painful. We had to call their support, and they’re like.
10 00:03:59.190 ⇒ 00:04:08.929 Uttam Kumaran: the guy was like… he comes on, he’s like, sorry, I’m in back-to-back, and I’m like, this guy hates his job, it’s gonna be so painful to ask him any questions. He’s like, they’re, like, alone.
11 00:04:09.170 ⇒ 00:04:13.529 Uttam Kumaran: his title is, like, product manager for the APIs, but he’s taking, like, support calls, and I’m like.
12 00:04:13.530 ⇒ 00:04:15.619 Katherine Bayless: but…
13 00:04:15.620 ⇒ 00:04:22.670 Uttam Kumaran: we were trying to build this pipeline, and then turns out, deep in the docs, they’ve shipped ability to go direct to S3.
14 00:04:23.020 ⇒ 00:04:23.460 Katherine Bayless: Oh.
15 00:04:23.460 ⇒ 00:04:27.650 Uttam Kumaran: I mean, I’m like, guys, why did we not do this? Why are we writing this pipeline?
16 00:04:28.060 ⇒ 00:04:30.189 Uttam Kumaran: Set this integration up, so…
17 00:04:30.190 ⇒ 00:04:30.730 Katherine Bayless: Yeah.
18 00:04:30.730 ⇒ 00:04:37.120 Uttam Kumaran: I guess that’s… that’s his job, is to inform us that it’s deep in some weird part of the docks that we didn’t find before.
19 00:04:37.550 ⇒ 00:04:41.290 Katherine Bayless: That’s kind of funny, like, definitely an easier method.
20 00:04:41.290 ⇒ 00:04:44.880 Uttam Kumaran: Yeah, we’re writing, like, a very complicated Python and looping, we’re like.
21 00:04:45.150 ⇒ 00:04:45.850 Katherine Bayless: Yeah.
22 00:04:45.850 ⇒ 00:04:50.690 Uttam Kumaran: And I’m like, oh, why don’t we just do this? But… Yeah.
23 00:04:50.860 ⇒ 00:04:52.859 Katherine Bayless: Well, you know, better late than never?
24 00:04:52.860 ⇒ 00:04:55.780 Uttam Kumaran: Yeah, I guess, I guess.
25 00:04:55.780 ⇒ 00:05:04.820 Katherine Bayless: Yeah. Oh, actually, it’s a good reminder, I need to reach out to the guy from the little chatbot thing, because he had offered to just give us, yeah, like, a big dump of data into S3.
26 00:05:04.820 ⇒ 00:05:05.310 Uttam Kumaran: Cool.
27 00:05:05.310 ⇒ 00:05:08.490 Katherine Bayless: So, so yeah, I need to… Should make a.
28 00:05:08.490 ⇒ 00:05:11.750 Uttam Kumaran: Yeah, if you want to just shoot, like, you could just put me on that thread.
29 00:05:12.410 ⇒ 00:05:13.100 Katherine Bayless: Yeah.
30 00:05:13.100 ⇒ 00:05:14.199 Uttam Kumaran: That would be perfect.
31 00:05:14.540 ⇒ 00:05:15.290 Katherine Bayless: Okay.
32 00:05:16.190 ⇒ 00:05:20.050 Uttam Kumaran: I’m going to create a ticket.
33 00:05:20.510 ⇒ 00:05:21.420 Katherine Bayless: Hmm, nice.
34 00:05:21.870 ⇒ 00:05:22.640 Katherine Bayless: Yes, yes.
35 00:05:26.710 ⇒ 00:05:29.889 Katherine Bayless: Whew, okay, so many things.
36 00:05:32.060 ⇒ 00:05:44.439 Katherine Bayless: Yes, where to start? Okay. So, cursor demo debrief, I did manage to get a chance to kind of vent to the marketing guy, for that and some other things, so…
37 00:05:44.440 ⇒ 00:05:45.080 Uttam Kumaran: Okay.
38 00:05:45.080 ⇒ 00:05:47.340 Katherine Bayless: I got most of the rant off my chest.
39 00:05:47.340 ⇒ 00:05:47.960 Uttam Kumaran: Okay.
40 00:05:48.150 ⇒ 00:05:49.489 Katherine Bayless: Jay was,
41 00:05:49.750 ⇒ 00:06:03.529 Katherine Bayless: Yeah, it did not… it did not go well. He was like, he’s like, I feel bad that he’s stuck in the Dark Ages using Cursor, like, are you sure you trust him to help you with data pipelines? And then now you see he’s suddenly in the, you know, channel with the, like, all this.
42 00:06:03.530 ⇒ 00:06:05.999 Uttam Kumaran: What is the Dark Ages? What does that mean?
43 00:06:07.290 ⇒ 00:06:13.970 Katherine Bayless: Who knows? But he has the Claude, you know, code hooked up to the 1Password MCP, so it can pull his passwords down, that’s great.
44 00:06:15.240 ⇒ 00:06:19.150 Uttam Kumaran: Okay, I feel like we’re in, like, warp speed, like…
45 00:06:19.730 ⇒ 00:06:24.080 Uttam Kumaran: way crazier than what we’re seeing anywhere else. But.
46 00:06:24.570 ⇒ 00:06:25.230 Uttam Kumaran: Yeah.
47 00:06:25.230 ⇒ 00:06:26.380 Katherine Bayless: Yeah.
48 00:06:26.560 ⇒ 00:06:36.510 Katherine Bayless: Yeah, yeah, yeah. So, yeah, I think we’ve somehow, turned him into, like, must demonstrate technical competence mode. And so, yeah, that’s, now we’re getting.
49 00:06:36.510 ⇒ 00:06:40.960 Uttam Kumaran: Good. I feel like, yeah. He’s thinking me about code owners and stuff, I said.
50 00:06:41.570 ⇒ 00:06:45.130 Uttam Kumaran: Yeah, let’s do it. I’m… I can dance, like, I don’t mind.
51 00:06:45.780 ⇒ 00:06:54.309 Katherine Bayless: Yeah, I mean, yeah, yeah, exactly, exactly. I also, I mean, I… it’s ironic, because I… I feel like I…
52 00:06:54.530 ⇒ 00:07:03.630 Katherine Bayless: you know, I have to mind my own hypocrisy here, because I’m like, we do need to be much better about security and stuff like that, and so, like, you know, and just good development best practices.
53 00:07:03.630 ⇒ 00:07:04.020 Uttam Kumaran: Yes.
54 00:07:04.020 ⇒ 00:07:11.889 Katherine Bayless: Anyway, and so it’s like, the things he’s suggesting are like, yes, we should do them, but, like, he doesn’t do any of that, right?
55 00:07:11.890 ⇒ 00:07:18.300 Uttam Kumaran: Yeah, that’s a… but that’s… it’s also, like, yeah, it’s… there’s a phrase where it’s like, for you, but not for me, I forgot what that… what that is.
56 00:07:18.300 ⇒ 00:07:19.730 Katherine Bayless: say not as I do, yeah.
57 00:07:19.730 ⇒ 00:07:25.389 Uttam Kumaran: Yeah, and so, like, yeah, I’m like, he’s like, let’s add cooners. I’m like, yes, we’re gonna do all that, but, like.
58 00:07:25.560 ⇒ 00:07:29.940 Katherine Bayless: Right. It’s also more important that we get some models out for this thing, because no…
59 00:07:30.020 ⇒ 00:07:39.599 Uttam Kumaran: Membership does not care if I have linting on the repo. Like, I can’t tell them that, like, oh, we didn’t get the linting done, so I can’t ship this for you.
60 00:07:39.600 ⇒ 00:07:39.950 Katherine Bayless: Exactly.
61 00:07:39.950 ⇒ 00:07:49.680 Uttam Kumaran: So, we’re like, I’m like, let’s get this done, and then, yeah, as now, Kyle’s like, hey, I need to move, I’m like, okay, we should put PR reviews, it naturally happens this way.
62 00:07:49.680 ⇒ 00:07:57.719 Katherine Bayless: Yeah. Anyway, so… That’s exactly right. Yeah, it’s like this organic evolution of, like, as we realize this thing will help, we put it in the place, right?
63 00:07:57.720 ⇒ 00:07:58.740 Uttam Kumaran: Yeah, yeah.
64 00:07:58.740 ⇒ 00:08:03.650 Katherine Bayless: Yeah, yeah. Yeah, so… it’s fine. I will admit, though, having seen.
65 00:08:03.650 ⇒ 00:08:10.320 Uttam Kumaran: Well, you also tell me, like, how we can win them over, because I like a challenge like that. That’s why I mentioned, if it’s helpful for…
66 00:08:10.640 ⇒ 00:08:19.340 Uttam Kumaran: Me to just meet directly with him, or, like, we just have some two weeks where we’re, like, getting his feedback, and it allows him to still feel in the loop, like.
67 00:08:19.460 ⇒ 00:08:21.409 Uttam Kumaran: I’m down to do that.
68 00:08:21.660 ⇒ 00:08:35.059 Katherine Bayless: Yeah, I… and I… I do think that, like, bringing him closer in is a good idea. I think it is where I start to be… like, where my nerves start to come into the picture are, like, I just…
69 00:08:36.890 ⇒ 00:08:50.370 Katherine Bayless: I mean, like that thread, right? Like, he can go in those, like, sort of directions where it’s like, okay, yes, but we need to, like, remember the goal here, and so it’s like, I don’t want him to just come in and then, like…
70 00:08:50.910 ⇒ 00:09:03.790 Katherine Bayless: derail or bog down, like, the speed that we’re moving with, like… but at the same time, acknowledging that he’s really smart and talented and very, like, you know, tied into this organization and its practices, and so, like.
71 00:09:03.790 ⇒ 00:09:04.110 Uttam Kumaran: Yeah.
72 00:09:04.110 ⇒ 00:09:11.350 Katherine Bayless: But I just feel like I could see where it goes from, like, you know, we’re delivering really fast.
73 00:09:11.350 ⇒ 00:09:12.209 Uttam Kumaran: Yeah, yeah, yeah.
74 00:09:12.210 ⇒ 00:09:14.570 Katherine Bayless: Then this becomes… Right.
75 00:09:14.570 ⇒ 00:09:15.180 Uttam Kumaran: Yeah.
76 00:09:15.740 ⇒ 00:09:32.239 Katherine Bayless: But yeah, but I think… I mean, in terms of the winning him over thing, honestly, I… I don’t know. Guys, sometimes I’m not sure how he likes me, because I think all I do is drive him crazy. But, like, the security thing is a big piece, and, like, I think if there’s a way to, like…
77 00:09:33.040 ⇒ 00:09:47.159 Katherine Bayless: help him feel like he’s in that driver’s seat, but, like, empower and encourage, right? Like, we really… and I keep pushing him, I’m like, we need… we don’t have end-to-end observability on our environment right now, and that’s just terrifying to me, right? Like…
78 00:09:47.160 ⇒ 00:10:04.300 Uttam Kumaran: If you think the security angle is that, then why don’t, like… I mean, there’s a host of ways that I can prove what we’re doing there, so maybe that’s where I most be like, hey, I want to just make sure you’re aware of how we’re doing governance, how we’re thinking about security here, things like that.
79 00:10:04.300 ⇒ 00:10:06.270 Katherine Bayless: that. I’m thinking more like…
80 00:10:06.270 ⇒ 00:10:10.080 Uttam Kumaran: Because then that’s… that’s, like, so harm… there’s, like, nothing you could do except layer on more.
81 00:10:10.400 ⇒ 00:10:17.240 Uttam Kumaran: security stuff, and there’s, like, no… does not, like, touch any of our, like, DPT stuff.
82 00:10:17.240 ⇒ 00:10:25.030 Katherine Bayless: Right. No, no, but it’s like, I’m thinking more, like, I genuinely would like to see him work on, like, what does secure… like, what…
83 00:10:25.030 ⇒ 00:10:25.650 Uttam Kumaran: Oh, yeah, yeah.
84 00:10:25.650 ⇒ 00:10:42.050 Katherine Bayless: like, security that need to change here, but, like, especially in that context of AI, like, you know, are you running Clawed code, like, raw against all your files, or are you putting it in a Dockerized, you know, containerized, sort of situation? Like, some of these things, like, they’re very sticky to him, and so it’s like, if you give him, like, little rabbit holes.
85 00:10:42.050 ⇒ 00:10:43.129 Uttam Kumaran: Yeah, yeah.
86 00:10:43.130 ⇒ 00:10:47.680 Katherine Bayless: Like, I think that autonomy and utility of the task, I think those.
87 00:10:47.680 ⇒ 00:10:55.160 Uttam Kumaran: So that’s why I tried to stress that, like, the Snowflake CLI, it’s all through your Okta, and then it’s all so, like.
88 00:10:55.410 ⇒ 00:10:56.780 Uttam Kumaran: Nothing is, like…
89 00:10:57.050 ⇒ 00:10:57.510 Katherine Bayless: Right.
90 00:10:57.510 ⇒ 00:11:05.189 Uttam Kumaran: and then it’s also, like, all within Snowflake, so… you know, and so, okay, that’s helpful that, like.
91 00:11:05.300 ⇒ 00:11:10.370 Uttam Kumaran: Maybe if we have questions, or we need second opinion on those things, I can toss that to him as a way to loop him in.
92 00:11:11.250 ⇒ 00:11:12.130 Katherine Bayless: Yeah.
93 00:11:12.340 ⇒ 00:11:26.829 Uttam Kumaran: Or you tell me, I just… or we just keep rolling with it. It’s more of, like, I don’t want him to end up… typically, in these situations, I just don’t want people to end up being the blocker, so if I can lay… even if they don’t end up liking us, as long as they’re, like, they’re fine.
94 00:11:27.480 ⇒ 00:11:34.660 Uttam Kumaran: then it’s like, okay, that’s a win for me, like, so that they don’t end up being… if and when we need a favor.
95 00:11:35.240 ⇒ 00:11:39.690 Uttam Kumaran: We have… we have ability to get that, versus it’s not just, like.
96 00:11:40.110 ⇒ 00:11:52.519 Uttam Kumaran: Yeah, I don’t know. It’s just company pol… I just learned… I’ve become a… I was talking to someone yesterday, you just, like, learn these politics habits of, like, how to get around this type of stuff, and so… happy to help with that, if you need help with that.
97 00:11:52.750 ⇒ 00:12:00.070 Katherine Bayless: Yeah, well, I think it’s, like, I’m also, you know, I mean, I’m mindful of the fact that, like, our two worlds, I mean, the lines are so blurry between.
98 00:12:00.070 ⇒ 00:12:00.520 Uttam Kumaran: Yes.
99 00:12:00.520 ⇒ 00:12:15.940 Katherine Bayless: data at this point in time anyway, and so it’s like, I want… because there is a lot of IT work that needs to happen, right? And so it’s like, I want to have him feel like maybe we’re the, you know, the exciting and energizing, you know, like, group, but also, like, not just getting into the weeds of our stuff, but like, you know…
100 00:12:15.940 ⇒ 00:12:16.640 Uttam Kumaran: Yes.
101 00:12:16.990 ⇒ 00:12:18.430 Katherine Bayless: fixing all the IT stuff, too.
102 00:12:18.430 ⇒ 00:12:19.260 Uttam Kumaran: Yes, yes.
103 00:12:19.260 ⇒ 00:12:27.380 Katherine Bayless: Observability, I think, is a piece that’s gonna, like, like, that piece, I think we need it, I think it makes sense for him to tackle, and I think, like…
104 00:12:27.410 ⇒ 00:12:40.619 Katherine Bayless: that would be a massive improvement in our cybersecurity posture, just generally, is, like, what is happening on our network at all times, right? I mean, we have bits and pieces, but not much in the way of, like, robust…
105 00:12:41.110 ⇒ 00:12:41.960 Katherine Bayless: Stuff.
106 00:12:42.180 ⇒ 00:12:43.240 Uttam Kumaran: Yeah, yeah.
107 00:12:43.500 ⇒ 00:12:48.409 Katherine Bayless: I know he’s trying to, like, move on the box stuff, too, I guess, earlier this week.
108 00:12:48.470 ⇒ 00:13:06.859 Katherine Bayless: like, finance or HR, somebody realized that, like, Glean can’t see the contents of something, but it can see the file names of, like, Microsoft Forms or something like that, like, submissions, and so it was returning things that, like, knew about people’s, like, promotion requests. Okay.
109 00:13:06.860 ⇒ 00:13:07.650 Uttam Kumaran: Yeah, yeah, yeah, yeah.
110 00:13:07.650 ⇒ 00:13:15.020 Katherine Bayless: Yeah, yeah, right? So he tried to, like, say, like, okay, well, we need to move to Box, right? Because they’re still in mapped network drives on those teams.
111 00:13:15.020 ⇒ 00:13:15.730 Uttam Kumaran: Yeah.
112 00:13:15.730 ⇒ 00:13:19.950 Katherine Bayless: And he got kind of shut down, and I don’t really understand why.
113 00:13:20.130 ⇒ 00:13:27.439 Uttam Kumaran: But, like, anything I can do to help him advance those conversations, too. Okay. Like, no more network drives.
114 00:13:27.440 ⇒ 00:13:33.140 Katherine Bayless: No more, like, shadow IT everywhere, right? Eventually, no more spreadsheets, that one I’m trying to tackle.
115 00:13:33.140 ⇒ 00:13:33.470 Uttam Kumaran: Yeah.
116 00:13:33.510 ⇒ 00:13:34.959 Katherine Bayless: But it’s like, I think…
117 00:13:35.310 ⇒ 00:13:46.409 Katherine Bayless: things that help him feel like he’s actually getting to do the job that he’s been denied for so long. I mean, truthfully, in fairness to him, the organization, I think, for a long time was just like, IT is the, you know, the.
118 00:13:46.410 ⇒ 00:13:48.919 Uttam Kumaran: Yeah, yeah, yeah, yeah. Right. Yeah.
119 00:13:48.970 ⇒ 00:13:51.470 Katherine Bayless: It’s like, I’m trying to give him a glow-up, a, you know…
120 00:13:51.470 ⇒ 00:13:52.729 Uttam Kumaran: Yes, yes, yes.
121 00:13:52.730 ⇒ 00:13:54.170 Katherine Bayless: main character role.
122 00:13:54.170 ⇒ 00:13:56.130 Uttam Kumaran: Sir, go from IT to engineering.
123 00:13:56.130 ⇒ 00:13:57.010 Katherine Bayless: Yeah, right.
124 00:13:57.010 ⇒ 00:13:59.910 Uttam Kumaran: Yeah, great rebrand. Exactly, exactly.
125 00:13:59.910 ⇒ 00:14:01.520 Katherine Bayless: Exactly, exactly.
126 00:14:01.520 ⇒ 00:14:06.010 Uttam Kumaran: That’s why when I talk about us, I’m like, yeah, the… the… the,
127 00:14:06.110 ⇒ 00:14:11.060 Uttam Kumaran: like, industry of IT service companies. It just makes it seem way…
128 00:14:11.280 ⇒ 00:14:23.499 Uttam Kumaran: way easier to go up against, versus I’m like… but then what I talk about it is, I’m like, we’re an engineering firm, like, I just, like… you know, but it’s just small, small.
129 00:14:23.500 ⇒ 00:14:24.579 Katherine Bayless: Small.
130 00:14:24.580 ⇒ 00:14:26.500 Uttam Kumaran: All branding tips.
131 00:14:26.810 ⇒ 00:14:27.580 Katherine Bayless: Totally.
132 00:14:27.990 ⇒ 00:14:29.649 Katherine Bayless: Totally.
133 00:14:29.830 ⇒ 00:14:53.779 Katherine Bayless: But yeah, so anyway, that was… that was kind of that, but yeah. I think… so then, truthfully, Kyle and I were both kind of like, okay, but, like, what does make Cursor special? Because, like, he does… like, we had it, like, set up, like, with VS Code and, like, you know, similar kind of interface with the Claude Code extension and things like that, and so I was like, well, there’s got to be something that makes Cursor, like, better, and so we were like, well, let’s just play around with it and explore a bit, but I think…
134 00:14:54.040 ⇒ 00:15:00.869 Katherine Bayless: Kyle is potentially… I do think we need to get some of those pairing sessions set up, because he’s… Yeah.
135 00:15:01.460 ⇒ 00:15:08.320 Katherine Bayless: I think he’s got, like, something’s not quite configured right, because he’s like, the repo I see online and the repo that I pull, and he’s like.
136 00:15:08.320 ⇒ 00:15:16.650 Uttam Kumaran: Yeah, I was DMing him today, and I was like, just tell me when you’re having Git issues, so I’m starting to work closer with him on, like, that type of stuff.
137 00:15:16.650 ⇒ 00:15:34.050 Katherine Bayless: Okay. And I really, I mean, like, yeah, I mean, he kind of, like, sought that out, but then also, yeah, like, just kind of working on building his, like, engineering skills and confidence, and, like, I think he really would like to see, and me too, to be honest, like, what does this look like from, you know, raw data file to prod.
138 00:15:34.050 ⇒ 00:15:53.220 Uttam Kumaran: Yeah, and that’s why I went… I’m… I don’t care what your VS Code or… like, at his level, it’s not gonna matter. Like, that, okay, cursor indexes files a little bit better, you can… you can switch models, you have this… like, I was like, I don’t care, I’m not… don’t want to debate, I want… what I want us to do is use some type of agentic system.
139 00:15:53.230 ⇒ 00:16:02.279 Uttam Kumaran: To do our work. Whatever you feel like. I don’t want to talk… I don’t work for Cursor. That’s what we use. You know, so…
140 00:16:02.280 ⇒ 00:16:07.269 Katherine Bayless: he’s been really aggressively learning, like, and using Claude Code, and, like, I think.
141 00:16:07.270 ⇒ 00:16:07.770 Uttam Kumaran: Yeah.
142 00:16:07.770 ⇒ 00:16:17.060 Katherine Bayless: in some ways, I do also need him to, like, calm a little bit back down, right? Because, like, what you were showing in Curse was really helpful, where it’s like, you know, you’re not saying, like, build me a pipeline, right? You know, you’re like.
143 00:16:17.060 ⇒ 00:16:23.779 Uttam Kumaran: No, no, no, it’s very specific tasks, yeah, yeah, it’s very specific tasks that I… that I start to plan out first.
144 00:16:23.950 ⇒ 00:16:29.410 Uttam Kumaran: And then I’m like, okay, it will do a lot of the work that I was just gonna write by hand, you know?
145 00:16:29.410 ⇒ 00:16:30.750 Katherine Bayless: Yeah, yeah, exactly.
146 00:16:30.750 ⇒ 00:16:33.670 Uttam Kumaran: There is an understanding of, like, okay, this step, this step, this step.
147 00:16:33.840 ⇒ 00:16:34.770 Uttam Kumaran: Yeah.
148 00:16:34.770 ⇒ 00:16:41.380 Katherine Bayless: Right. And so, like, I think right now he’s kind of like, you know, he’s got the repo, and he’ll say to Cloud Code, like, I need a pipeline for the scanners, right?
149 00:16:41.380 ⇒ 00:16:41.890 Uttam Kumaran: Yeah, yeah.
150 00:16:41.890 ⇒ 00:16:43.780 Katherine Bayless: The whole thing, you know?
151 00:16:43.780 ⇒ 00:16:49.049 Uttam Kumaran: But that’s so much… that’s, like, so nonspecific. There’s no nuance in that ask, yeah.
152 00:16:49.050 ⇒ 00:16:49.770 Katherine Bayless: Right, and I think.
153 00:16:49.770 ⇒ 00:16:58.330 Uttam Kumaran: But on data side, again, that’s what I try to express, we’re not building, like, there’s not much you’re doing, like, end-to-end, where you can one-shot prompt it.
154 00:16:58.330 ⇒ 00:16:58.660 Katherine Bayless: Right.
155 00:16:58.660 ⇒ 00:17:06.329 Uttam Kumaran: Like, these are just, like, incremental things where the task in between each, like, check, you can start to speed up.
156 00:17:06.690 ⇒ 00:17:08.439 Uttam Kumaran: Yeah. Yeah.
157 00:17:08.440 ⇒ 00:17:17.079 Katherine Bayless: Yeah, and so it’s like, I think he’s, like, this is a good time to, like, work with him, right? Because it’s, like, he’s organically identifying this friction and, like, right?
158 00:17:17.089 ⇒ 00:17:17.549 Uttam Kumaran: Yeah.
159 00:17:17.550 ⇒ 00:17:21.880 Katherine Bayless: I feel the power, but I also am, like, struggling to, like, harness it, you know?
160 00:17:21.880 ⇒ 00:17:22.489 Uttam Kumaran: Yeah, yeah.
161 00:17:22.900 ⇒ 00:17:41.469 Katherine Bayless: And so, yeah, so I think, and also, you know, in fairness to him, I mean, he went from, you know, occasionally writing some R on a SQL server to, like, whatever this chaos is that I’ve created. And so, like, I think also the CICD stuff is, like, a black box, and he’s kind of like, how does it…
162 00:17:41.960 ⇒ 00:17:42.850 Katherine Bayless: Like, how do.
163 00:17:42.850 ⇒ 00:17:47.859 Uttam Kumaran: Well, that’s what I… I almost, like, don’t want people to… I just, like, focus on building the models.
164 00:17:48.150 ⇒ 00:17:49.089 Uttam Kumaran: And then…
165 00:17:49.540 ⇒ 00:18:05.720 Uttam Kumaran: just be okay with, like, this is just more data, like, data engineering stuff that, I guess, there’s a whole other world here. So I’m kind of just want him to be… but again, it depends on how curious people… some people are like, okay, some people are like, no, no, no, I need to know how it works, and I’m like, well, then here’s a.
166 00:18:05.720 ⇒ 00:18:06.310 Katherine Bayless: room.
167 00:18:06.310 ⇒ 00:18:19.479 Uttam Kumaran: here’s a… how does GitHub, how does Git work? If you want to watch that, then that’s how you learn this, because I can’t… I can’t… it’s really tough. There’s, like, a lot of stuff going on to learn how Git works.
168 00:18:19.760 ⇒ 00:18:39.660 Katherine Bayless: Right, right. But it’s like, that is his brain, right? Like, so, like, earlier today we were talking, he’s like, he’s very stuck on… when I look at it in my local files, it has this name, but when it makes it into Snowflake, the names are different. And I’m like, well, I think it’s, like, part of the CICD pipeline strips out some of the local naming conventions and puts in the ones that match Snowflake. Yeah. And that’s just, like…
169 00:18:40.040 ⇒ 00:18:40.850 Katherine Bayless: Right?
170 00:18:41.340 ⇒ 00:18:46.570 Katherine Bayless: But, like, okay, well, I made a dbt model, how do I get it to, like, Go.
171 00:18:46.570 ⇒ 00:18:47.580 Uttam Kumaran: No, yeah.
172 00:18:47.580 ⇒ 00:18:49.100 Katherine Bayless: to, right? Like, so just, like…
173 00:18:49.100 ⇒ 00:18:49.640 Uttam Kumaran: Yes.
174 00:18:49.640 ⇒ 00:18:50.360 Katherine Bayless: But, like, yeah.
175 00:18:50.360 ⇒ 00:18:50.880 Uttam Kumaran: Okay, okay.
176 00:18:50.880 ⇒ 00:18:56.189 Katherine Bayless: definitely has that kind of, like, I need to take it apart before I want to play with it sort of curiosity.
177 00:18:56.190 ⇒ 00:18:56.660 Uttam Kumaran: Okay.
178 00:18:56.660 ⇒ 00:18:57.230 Katherine Bayless: Yeah.
179 00:18:57.630 ⇒ 00:19:02.280 Uttam Kumaran: Okay, cool. That’s helpful. So yeah, I’ll book a session with him either tomorrow or Monday, and then we’ll.
180 00:19:02.280 ⇒ 00:19:04.870 Katherine Bayless: Maybe he’s off tomorrow, but yeah, it’ll be Monday.
181 00:19:04.870 ⇒ 00:19:07.310 Uttam Kumaran: So then we’ll just work together on stuff, yeah.
182 00:19:07.310 ⇒ 00:19:14.119 Katherine Bayless: Yeah, yeah, I think, like I said, yeah, like I said, like, a couple hours of pairing here and there, where he can, like, you know, kind of do homework, come with questions, yeah.
183 00:19:14.120 ⇒ 00:19:14.510 Uttam Kumaran: Yeah.
184 00:19:14.510 ⇒ 00:19:15.980 Katherine Bayless: forget it fast.
185 00:19:15.980 ⇒ 00:19:18.140 Uttam Kumaran: Yeah. How is Kai doing?
186 00:19:18.330 ⇒ 00:19:34.939 Katherine Bayless: She’s good, yeah. I think we’re sort of, I mean, her world also, like, so many things all at once, right? And so, like, really, focusing on documentation and trying to, like… she’s been working a lot on, like, the Asana piece, and then sort of, like, okay, well, how do we start
187 00:19:34.940 ⇒ 00:19:39.370 Katherine Bayless: Capturing things, where do we need to put them, and, like, we actually…
188 00:19:39.370 ⇒ 00:19:52.010 Katherine Bayless: I don’t know if it’ll show up on Zoom or it’ll just be blurry, but we had the meeting with the, membership team, a different meeting with the membership team on, like, Tuesday, I think it was, because we’re going to start doing journeys for them in Marketing Cloud.
189 00:19:52.010 ⇒ 00:20:06.720 Katherine Bayless: And so I was working with her on, like, okay, how are we gonna model this in Marketing Cloud, and, like, what are these fields, and how do we need to think through, like, the translation layer from membership to marketing, and, like, some of those things, and so, like, I think… I think she’s good. I think…
190 00:20:06.750 ⇒ 00:20:09.839 Katherine Bayless: what I need to do is give her more, like…
191 00:20:10.120 ⇒ 00:20:19.079 Uttam Kumaran: specific locations that things go, just to help her not have to, like, everything, you know, has to go through that whole routing algorithm in her mind, right? It’s like, okay… Yeah.
192 00:20:19.080 ⇒ 00:20:24.059 Katherine Bayless: field definition, and it’s something that we’ve modeled, go park in Snowflake. If it’s, you know… Okay.
193 00:20:24.420 ⇒ 00:20:33.729 Katherine Bayless: Okay. But yeah. And then we’re… recruitment is open for the other data engineer role. Apparently, there might be somebody in market research who also wants to come join the team?
194 00:20:34.240 ⇒ 00:20:34.610 Uttam Kumaran: Okay.
195 00:20:34.610 ⇒ 00:20:48.459 Katherine Bayless: We’ll see, it might be a poach situation. Cool. But yeah. And then I got the AWS ProServe contract is signed. So they said it would probably take, like, 6 weeks to really provision those resources, but that’s fine by me, because I’m not an.
196 00:20:48.460 ⇒ 00:20:48.960 Uttam Kumaran: Yeah.
197 00:20:48.990 ⇒ 00:21:02.010 Katherine Bayless: But yeah, so that’ll come through. And then the other two roles that may or may not report to me are still kind of TBD. I’m hoping maybe this afternoon, Christine will have an update on some of that. Okay. But yeah, so…
198 00:21:02.660 ⇒ 00:21:07.169 Uttam Kumaran: Okay, great. Yeah, I think on our site, we… we… I think we pushed out a lot of models this week.
199 00:21:07.170 ⇒ 00:21:07.730 Katherine Bayless: Yeah, you did.
200 00:21:07.730 ⇒ 00:21:19.180 Uttam Kumaran: We fixed a bunch of stuff in dbt. Yeah, I hope the session with Awash was good yesterday. He’s really, really, really good, yeah. Me and him sort of do most of the snowflake work at the company, so…
201 00:21:19.180 ⇒ 00:21:19.680 Katherine Bayless: No.
202 00:21:20.890 ⇒ 00:21:27.589 Uttam Kumaran: And so he’ll be… yeah, he’ll be helping me as we figure out… so tomorrow I’m meeting with Snowflake to learn about
203 00:21:27.840 ⇒ 00:21:31.129 Uttam Kumaran: All the new Cortex features and things like that.
204 00:21:31.280 ⇒ 00:21:38.209 Katherine Bayless: Yeah, I actually demoed some of it to the membership team earlier, because why not, right? And I was like, oh shit, this is good.
205 00:21:38.210 ⇒ 00:21:39.630 Uttam Kumaran: It was working? Okay, cool, great.
206 00:21:39.900 ⇒ 00:21:47.170 Katherine Bayless: Like, I asked it, like, I just, like, why not, right? And so I was like, what was the most popular session at CES? And it got it right, and then I said.
207 00:21:47.170 ⇒ 00:21:48.260 Uttam Kumaran: Wow.
208 00:21:48.260 ⇒ 00:21:50.049 Katherine Bayless: like, top 5… There’s no…
209 00:21:50.050 ⇒ 00:21:52.290 Uttam Kumaran: context in there, so it must have just looked at the…
210 00:21:52.710 ⇒ 00:21:57.240 Uttam Kumaran: Column headers, and kind of, like, profiled some rows, and figured it out.
211 00:21:57.580 ⇒ 00:21:58.230 Katherine Bayless: Yeah.
212 00:21:58.620 ⇒ 00:22:00.050 Uttam Kumaran: Damn. Okay, cool.
213 00:22:00.260 ⇒ 00:22:03.769 Katherine Bayless: Yeah, like, I… like, I think this is gonna be game-changing.
214 00:22:03.770 ⇒ 00:22:04.480 Uttam Kumaran: Dope.
215 00:22:04.480 ⇒ 00:22:09.339 Katherine Bayless: Yeah. Right. Yeah. So, which is a good segue, so it’s a membership team meeting this morning.
216 00:22:09.490 ⇒ 00:22:11.899 Katherine Bayless: So I…
217 00:22:12.510 ⇒ 00:22:23.789 Katherine Bayless: I did a couple things just kind of, like, level set, right? I was like, first of all, please know Snowflake, historically, is not a tool people would have been in that aren’t engineers, right? Like, this was a data warehouse as a service, right? Like, this is…
218 00:22:23.790 ⇒ 00:22:24.340 Uttam Kumaran: Yes.
219 00:22:24.340 ⇒ 00:22:40.420 Katherine Bayless: slightly non-conventional approach, but it is totally where the platform’s trying to go, and I think it is the right place for us to go, you know, for future, proofing our analytics stuff, right? And so… and fortunately, they’re really, like, excited, gung-ho, kind of, like, they’re great beta testers for this. Yeah.
220 00:22:40.420 ⇒ 00:22:49.249 Katherine Bayless: And so I was showing them, like, the Streamlit app, and then I was showing them the Cortex for the agent, and then I was even like, look, if you guys are feeling brave, you can’t break anything in Snowflake.
221 00:22:49.250 ⇒ 00:22:49.910 Uttam Kumaran: Yeah, yeah, yeah.
222 00:22:49.910 ⇒ 00:23:02.309 Katherine Bayless: You really can’t. And so I was showing them, like, you can ask the little thing to write you the sequel to talk to the table, and I eventually will ask you to stop exporting data, but right now, realistically, I understand that you’ll probably need to sometimes.
223 00:23:02.410 ⇒ 00:23:13.760 Katherine Bayless: I do think, definitely there’s some tweaks we’ll need to the Prod Marts, like, one that was kind of funny was, like, Research Downloads has, like, no joinable key in it. I was like.
224 00:23:13.760 ⇒ 00:23:14.410 Uttam Kumaran: Yeah, okay.
225 00:23:14.410 ⇒ 00:23:23.299 Katherine Bayless: But, like, little things here and there, and I think those are great opportunities for Kyle to learn how to make those edits, right?
226 00:23:23.300 ⇒ 00:23:23.990 Uttam Kumaran: Yeah.
227 00:23:23.990 ⇒ 00:23:31.350 Katherine Bayless: Yeah. And then, I also think, so the role scoping, so I gave them prod.
228 00:23:31.350 ⇒ 00:23:31.970 Uttam Kumaran: Okay.
229 00:23:32.210 ⇒ 00:23:37.329 Katherine Bayless: And so they were able to see everything fine, but they also see, like, all the stuff below.
230 00:23:37.330 ⇒ 00:23:37.909 Uttam Kumaran: Yeah, yeah.
231 00:23:37.910 ⇒ 00:23:43.320 Katherine Bayless: Right? And so I was like, it’s not… it’s not that I’m, like, worried, but it’s more just, like, that’s a lot of distraction.
232 00:23:43.320 ⇒ 00:23:47.599 Uttam Kumaran: Yeah, so let me… let me just try to fix that today with Awash.
233 00:23:47.600 ⇒ 00:23:47.940 Katherine Bayless: No, bro.
234 00:23:47.940 ⇒ 00:23:55.719 Uttam Kumaran: And then… and then I also want to create, like, a little bit of, like, so we… a dash that we can start to see, like, user…
235 00:23:55.890 ⇒ 00:23:57.600 Uttam Kumaran: Logins, and, like…
236 00:23:57.600 ⇒ 00:23:58.220 Katherine Bayless: Yeah.
237 00:23:58.500 ⇒ 00:24:02.890 Uttam Kumaran: some type of query sessions, less about our stuff, so I’ll filter out
238 00:24:03.520 ⇒ 00:24:07.770 Uttam Kumaran: I’ll filter out us, and then so we can just see, like, who’s externally coming in.
239 00:24:08.210 ⇒ 00:24:16.179 Katherine Bayless: Yeah, and Kai immediately, when I was, we debriefed, she’s like, so, when people talk to the agent, do we get to know what they ask? And I was like.
240 00:24:16.180 ⇒ 00:24:16.770 Uttam Kumaran: Oh, yeah.
241 00:24:16.770 ⇒ 00:24:17.849 Katherine Bayless: I hope so.
242 00:24:17.850 ⇒ 00:24:21.900 Uttam Kumaran: Boy, yes, we have to, because then we’re… we… that’s what informs…
243 00:24:21.900 ⇒ 00:24:26.570 Katherine Bayless: If things are breaking, and then they’ll… some people don’t want to be like, this broke, and I’m like, let’s go look through your…
244 00:24:26.920 ⇒ 00:24:29.249 Uttam Kumaran: query and find out, so totally. So, okay, great.
245 00:24:29.250 ⇒ 00:24:33.440 Katherine Bayless: Yeah, yeah, yeah. So yeah, so I think that would be awesome, a little bit of observability.
246 00:24:33.440 ⇒ 00:24:34.750 Uttam Kumaran: Okay, perfect.
247 00:24:35.320 ⇒ 00:24:48.510 Katherine Bayless: But yeah, overall, I mean, honestly, I was blown away at how much you guys got through on those marts, and, like, the little onesie-twosie changes, no big deals. Yeah. And then I think, it sounds like, Ashwini’s working on the identity stitching.
248 00:24:48.510 ⇒ 00:24:52.939 Uttam Kumaran: Yes. And so, like, the membership team, they are super excited. Okay.
249 00:24:52.940 ⇒ 00:25:07.499 Katherine Bayless: Great, cool. I feel really good going into… we’re gonna do another training with them tomorrow, because I was like, I’m gonna overwhelm you today, and then you’re gonna think about it, and then tomorrow, we can do all the questions, and they’re like, okay, wait, show me that again, and how does this work, and where’s that button, and all that stuff.
250 00:25:07.500 ⇒ 00:25:08.290 Uttam Kumaran: Okay.
251 00:25:08.290 ⇒ 00:25:13.019 Katherine Bayless: So yeah, I, feeling pretty good. I was so nervous before the meeting.
252 00:25:13.020 ⇒ 00:25:19.709 Uttam Kumaran: Well, it’s so… yeah, pulling up… well, that’s how I felt during the cursor thing, too, is I’m like, this could be really overwhelming, but…
253 00:25:19.890 ⇒ 00:25:25.080 Uttam Kumaran: And also, I just want people to… they’ll hear something, and then maybe 3 months from now, they’re like.
254 00:25:25.250 ⇒ 00:25:26.100 Uttam Kumaran: Wait…
255 00:25:26.120 ⇒ 00:25:29.849 Katherine Bayless: That’s what you’re trying to say, and I’m like, okay, that’s all I need to happen.
256 00:25:29.850 ⇒ 00:25:34.540 Uttam Kumaran: We’re gonna catch people at whatever stage of adoption of AI they are, I’m just gonna show them.
257 00:25:34.800 ⇒ 00:25:37.880 Uttam Kumaran: A bunch of things, and let’s see what sticks.
258 00:25:37.880 ⇒ 00:25:41.940 Katherine Bayless: Yeah. And then, yeah, I think more working sessions between the teams is gonna be good.
259 00:25:42.530 ⇒ 00:25:43.290 Katherine Bayless: Yeah.
260 00:25:43.290 ⇒ 00:26:02.900 Katherine Bayless: Yeah, we actually… so we’re meeting, I think it’s, like, every Thursday at 10 is the recurring one, so, like, I’m happy to, like, add you guys to the invites if you want, because I think they are going to be very much like that kind of working session of, like, I need this data point, and where is that, and why is this record not coming through? Oh, well, it’s because the data’s kind of funky, and so it’s not joining, right? All that stuff.
261 00:26:02.900 ⇒ 00:26:15.490 Uttam Kumaran: Yeah, please include me, and then we just, like, if we could just pair on random stuff, or just, like, kind of keep it as office hours, that’s perfect, yeah. Yeah, okay. And similarly, I think for most of our meetings, I think we’re not doing so heavy on…
262 00:26:15.610 ⇒ 00:26:32.920 Uttam Kumaran: like, I mean, again, like, if you want, we could do decks and everything, but I think most of it is just, like… Yeah, so… so most of it is, like, I just wanna… I’m trying to make sure that, even now, I’m just making sure that everything ends up in Asana, and then I think we kind of use that…
263 00:26:33.120 ⇒ 00:26:37.140 Uttam Kumaran: as, like, I’m gonna try to use that more to present, like, on progress.
264 00:26:37.140 ⇒ 00:26:41.859 Katherine Bayless: And then as much time as we can use to just, like, be in Snowflake together as a crew, talking.
265 00:26:42.180 ⇒ 00:26:43.769 Uttam Kumaran: We just do that, you know?
266 00:26:44.120 ⇒ 00:26:56.670 Katherine Bayless: Yeah, and actually, another, you know, testament to, like, Kyle’s instincts are totally there, like, the organic friction identification. He asked me this morning, he’s like, so can I, like, stop doing Postgres stuff? And I’m like, please, yes, ever.
267 00:26:56.670 ⇒ 00:26:58.960 Uttam Kumaran: Yes. Everything’s there, yes.
268 00:26:58.960 ⇒ 00:27:06.090 Katherine Bayless: Well, and I think he was kind of, like, trying to figure out, like, okay, well, do I need to limp these things along over here? And I’m like, no, no, we’re pushing, pushing, pushing.
269 00:27:06.090 ⇒ 00:27:11.329 Uttam Kumaran: That’s what I told… I said, like, even if you have a query that doesn’t run or something, just, like, put that in the channel, and then we’ll…
270 00:27:11.460 ⇒ 00:27:21.910 Uttam Kumaran: We’ll work with you on adapting it to the new thing, and yeah. But we’ve… everyone… and then, ideally, you find something that we haven’t piped in, or something’s missing, and that’s, like, what we need.
271 00:27:22.090 ⇒ 00:27:24.029 Uttam Kumaran: We need more of, so… yeah.
272 00:27:24.030 ⇒ 00:27:30.300 Katherine Bayless: Yeah, so I think I’ve been… I know I started this in the fall, finally circled back around.
273 00:27:30.530 ⇒ 00:27:49.929 Katherine Bayless: did one last round of QA checks on the data that I migrated from the old, old SQL server. So that’s all done, shored up. I need to go in and, like, actually delete the resources, which is very scary, in that old Azure account, and then my next focus is, like, decommissioning that Postgres, database. There is still the lingering Power BI question.
274 00:27:49.930 ⇒ 00:27:50.320 Uttam Kumaran: Yeah.
275 00:27:50.320 ⇒ 00:27:51.640 Katherine Bayless: I think…
276 00:27:51.830 ⇒ 00:27:55.259 Uttam Kumaran: I think you wait. Yeah, we just… let’s just try to… let’s see…
277 00:27:55.670 ⇒ 00:28:00.639 Uttam Kumaran: from what I’m seeing, I feel like you can… you may kick it even further.
278 00:28:00.820 ⇒ 00:28:06.860 Uttam Kumaran: And… I don’t know, like, I think you’re gonna… you’re, like, one of the first people to be able to test, like, whether…
279 00:28:07.530 ⇒ 00:28:12.100 Uttam Kumaran: can get away with most of the reporting through natural language, and actually, I think…
280 00:28:12.200 ⇒ 00:28:13.860 Uttam Kumaran: What’s gonna be the test is…
281 00:28:13.960 ⇒ 00:28:17.210 Uttam Kumaran: If we can enable people to just generate the graphs on the fly.
282 00:28:18.450 ⇒ 00:28:20.220 Uttam Kumaran: So, I feel like…
283 00:28:20.530 ⇒ 00:28:26.930 Uttam Kumaran: It’s worth trying, because it’s really cool if you’re able to do that, you know, and it’s gonna save the business a lot of money, and…
284 00:28:27.000 ⇒ 00:28:41.369 Uttam Kumaran: Any BI tool we bring, we have to do a ton of configuration. Right. So if most of our time is spent instead just, like, making sure the semantic layer and description and stuff all are good, and building, like, testing the natural language within Snowflake.
285 00:28:41.630 ⇒ 00:28:48.159 Katherine Bayless: Yep. Well, yeah, I did ask Kyle, I was like, start thinking about, like, what are our benchmark questions? Like.
286 00:28:48.160 ⇒ 00:28:48.560 Uttam Kumaran: Yes.
287 00:28:48.560 ⇒ 00:28:51.560 Katherine Bayless: things that, like, if AI can’t get this right, then don’t go for.
288 00:28:51.560 ⇒ 00:28:58.270 Uttam Kumaran: Yeah, so the way we… the way we will do this, and we have… we have a template, is we sort of build what’s called a golden data set.
289 00:28:58.270 ⇒ 00:28:58.590 Katherine Bayless: Yeah.
290 00:28:58.590 ⇒ 00:29:09.220 Uttam Kumaran: And so, we can build… we can have that, which is, like, what are the golden data set, easy, medium, hard questions? And we will actually have the right answers.
291 00:29:09.220 ⇒ 00:29:11.420 Katherine Bayless: And, like, the method to figure out the answer.
292 00:29:11.420 ⇒ 00:29:20.789 Uttam Kumaran: And we can run evals and things on the inputs, right? And so, that’s how we do all of our agent development work, is we build out that golden data set.
293 00:29:21.070 ⇒ 00:29:27.069 Uttam Kumaran: And then we can sort of… we have at least a roadmap of, like, what we’re trying to get to hard question-wise.
294 00:29:27.310 ⇒ 00:29:35.699 Uttam Kumaran: Versus, like, off the cuff, you have to be like, oh, I don’t know what our question dataset is, like, we could… we could think through that. So maybe that’s something we could do as a group, too.
295 00:29:36.000 ⇒ 00:29:43.909 Katherine Bayless: Yeah, yeah, I think it’s good, like, you know, Q2 kind of work, honestly, right? Like, I think I very much agree with what you said at the beginning, right? It’s like, right now, we just build, build, build, build, build, build.
296 00:29:43.910 ⇒ 00:29:52.330 Uttam Kumaran: Yes, yeah. Yeah, if we get all the marts there, then it’s purely just building the semantics and, yeah, showing people how to use it, so…
297 00:29:52.670 ⇒ 00:29:57.680 Uttam Kumaran: That’s what I’m kind of, like, trying to push the team, now that everything’s landed, to just keep building all the core marts.
298 00:29:57.680 ⇒ 00:30:00.340 Katherine Bayless: Yeah. Yeah.
299 00:30:00.340 ⇒ 00:30:00.700 Uttam Kumaran: Yeah.
300 00:30:00.700 ⇒ 00:30:08.730 Katherine Bayless: I think, too, like, the thing that really works to our advantage with trying to move aggressively into this, you know, new world is, like.
301 00:30:08.840 ⇒ 00:30:25.829 Katherine Bayless: And even the other day, I was talking to Anna on the membership team, and watching her pull up something that we have, not only in Power BI and also in Snowflake, but I’m pretty sure in Remembers, but, like, she was pulling it up in the system for the vendor that we use for Reg. They have this, like, funny little analytics platform, and I was like.
302 00:30:25.930 ⇒ 00:30:37.459 Katherine Bayless: you went to decision point for that. Like, fascinating. I think at first she thought I was, like, criticized. I was like, no, no, I’m genuinely, like, fascinated. But, like, that is the level of, like, willingness that we have at our.
303 00:30:37.460 ⇒ 00:30:38.110 Uttam Kumaran: Yeah.
304 00:30:38.110 ⇒ 00:30:40.320 Katherine Bayless: Right? Like, these people, they want the data, and they’ll go.
305 00:30:40.320 ⇒ 00:30:41.080 Uttam Kumaran: Yeah, yeah, yeah.
306 00:30:41.080 ⇒ 00:30:49.769 Katherine Bayless: is, it doesn’t matter how illogical or unpleasant, and so I’m like, if I can give you a logical, unpleasant experience, I don’t think they’re gonna miss Power BI one bit.
307 00:30:49.960 ⇒ 00:31:04.439 Uttam Kumaran: Yeah, so maybe a good win when we do any of this, like, security stuff is to show Jay, like, how we’re building sort of this governed national language environment where everybody’s coming in through Okta into Snowflake, they have roles where it’s provisioned, so
308 00:31:04.660 ⇒ 00:31:09.360 Uttam Kumaran: And it’s using their roles that query all the data that only they can see, you know, and like…
309 00:31:09.540 ⇒ 00:31:10.350 Uttam Kumaran: Okay.
310 00:31:10.680 ⇒ 00:31:12.750 Katherine Bayless: Yeah, yeah, yeah.
311 00:31:12.870 ⇒ 00:31:14.459 Katherine Bayless: Yeah, I mean, I think…
312 00:31:14.840 ⇒ 00:31:30.230 Katherine Bayless: Yeah. I do think he sees what and why and how I have been building this approach, right? And I think it’s, like, it’s different from what he’s used to and how the organization used to function, but it’s like, I’m also, like, I’m getting results, right? Like, I’m…
313 00:31:30.230 ⇒ 00:31:32.820 Uttam Kumaran: Yeah, yeah, yeah, that’s the thing, and it’s moving fast.
314 00:31:32.820 ⇒ 00:31:38.880 Katherine Bayless: Right. Yeah. Yeah, and so, like, I think, you know, he’s a little cranky sometimes, and…
315 00:31:39.030 ⇒ 00:31:43.479 Katherine Bayless: Definitely does the, like, oh, you speak tech, me speak more tech, kind of thing.
316 00:31:43.480 ⇒ 00:31:55.160 Uttam Kumaran: Yeah. I’ve never heard someone be like, oh, you’re still using cursor? I’m like, dude, I’m not, like, this isn’t, like, a hobby, this isn’t, like, I don’t… I’m not, like, on MoltBook, like, for work.
317 00:31:55.160 ⇒ 00:32:13.459 Uttam Kumaran: Like, you know what I mean? Like, I’m not trying to one-up you on the AI side. I really… I don’t do… I don’t really do much AI outside of work, because it’s all I do. Yeah. So I’m just such a… I just chat GPT… I’m like a ChatGPT normie on the weekends. Like, I don’t think about this much.
318 00:32:13.460 ⇒ 00:32:13.950 Katherine Bayless: Oh.
319 00:32:13.950 ⇒ 00:32:15.510 Uttam Kumaran: Try not to.
320 00:32:15.510 ⇒ 00:32:22.720 Katherine Bayless: He thinks about it 24-7. Actually, no, that makes me realize the thing that I think he could really use
321 00:32:22.830 ⇒ 00:32:38.709 Katherine Bayless: help with is, like, and not just, like, our environment needs cybersecurity, please, but, like, his workflow. Like, he goes nuts with the way that it doesn’t listen, and he’s like, it prints my tokens. I’m like, why does it have access to them anyway, buddy? But…
322 00:32:38.710 ⇒ 00:32:39.340 Uttam Kumaran: Yeah.
323 00:32:39.340 ⇒ 00:32:53.979 Katherine Bayless: Right, and so, like, I think, like, when you were talking about rules, that was the concept that I hadn’t really come across, and, like, I know he does a little bit with, like, the QuadMD files and stuff, like, those things that help AI follow the rules are something he seems to struggle with a lot, and I think.
324 00:32:53.980 ⇒ 00:32:58.050 Uttam Kumaran: Yeah, because he was like, it doesn’t listen. I’m like, it does, because I…
325 00:32:58.070 ⇒ 00:32:58.560 Katherine Bayless: Huh?
326 00:32:58.560 ⇒ 00:33:07.799 Uttam Kumaran: I use Curse for 5 hours a day, it’s like, yes, this is not like a debate… it’s not really a debate. I don’t know. Oh my god, well, this is how it works.
327 00:33:08.020 ⇒ 00:33:14.540 Uttam Kumaran: Yeah. But I can. So I can talk a little bit about, like, our AI, like… workspace, or…
328 00:33:14.920 ⇒ 00:33:15.350 Katherine Bayless: Yeah.
329 00:33:15.510 ⇒ 00:33:17.719 Uttam Kumaran: Steering, like, how we do that a little bit.
330 00:33:17.720 ⇒ 00:33:32.919 Katherine Bayless: Right. Like, this morning, he posted in the Slack channel that we have for Claude Code, this, like, thing where he’s like, you know, Claude Code bypassed his commit checkers, and so he’s, like, screaming at it. And he says, you know, you committed a secret to the repo, and I’m like, okay, again, why does it have access?
331 00:33:32.920 ⇒ 00:33:42.759 Katherine Bayless: So then he went and figured out how to get it to change the little verbs that it shows while it’s working, to say, like, getting you fired and destroying your trust.
332 00:33:43.560 ⇒ 00:33:54.049 Uttam Kumaran: Yeah, like, but again, there’s common things, like, you still need to use gitignores, you can still do pre-commit checks for secrets, you can also layer in secret checks on the repo.
333 00:33:54.480 ⇒ 00:33:56.850 Uttam Kumaran: That’s just all things we have to do. Yeah.
334 00:33:56.850 ⇒ 00:34:01.250 Katherine Bayless: has done, and they don’t work. And I’m like, I just… I know that these things can be done, because…
335 00:34:01.250 ⇒ 00:34:02.190 Uttam Kumaran: Yeah, yeah.
336 00:34:02.190 ⇒ 00:34:06.370 Katherine Bayless: Yeah, yeah, and like, I… my AI listens to me, right? Yeah.
337 00:34:08.179 ⇒ 00:34:11.059 Katherine Bayless: That is totally a space where, like, if you can help, I think.
338 00:34:11.060 ⇒ 00:34:16.560 Uttam Kumaran: Okay. Or I could just show how we’re, like, we’re thinking about that very seriously as well.
339 00:34:16.560 ⇒ 00:34:17.460 Katherine Bayless: Yeah, yeah.
340 00:34:17.469 ⇒ 00:34:24.259 Uttam Kumaran: Level of seriousness. Yeah, yes, yes, yes, there you go. Yeah, I think that’s the space. Okay.
341 00:34:24.620 ⇒ 00:34:29.029 Uttam Kumaran: So I have, like, Gleans, so I still, I think, I wanna… I’ll follow up on Glean Access.
342 00:34:29.030 ⇒ 00:34:30.689 Katherine Bayless: Oh, you should have it.
343 00:34:30.690 ⇒ 00:34:36.170 Uttam Kumaran: Okay, then let me check. I was gonna try to add it… oh, maybe I just haven’t checked in Okta in a while.
344 00:34:36.170 ⇒ 00:34:41.029 Katherine Bayless: Yeah, it should be in Okta, yeah, because Ian said he set you up with it. If you don’t see it, let me know, but it should be in there.
345 00:34:41.030 ⇒ 00:34:53.300 Uttam Kumaran: So let me do that. We’ll… we’ll plan on just, like, yeah, using the back half of our, like, syncs for working sessions, but I think we can do next Thursday, and then I’ll try to grab time with Kyle, like, sometime on Monday. Okay.
346 00:34:53.659 ⇒ 00:35:00.910 Uttam Kumaran: I’ll create a ticket for, like, the chat sessions from the chat vendor, and then…
347 00:35:01.040 ⇒ 00:35:07.620 Uttam Kumaran: We also have, like, trying to set up something around, like, what people are using the natural language for, like, a little bit of a…
348 00:35:07.620 ⇒ 00:35:08.470 Katherine Bayless: Yeah. Observability.
349 00:35:09.060 ⇒ 00:35:10.240 Katherine Bayless: Yeah.
350 00:35:10.250 ⇒ 00:35:12.710 Uttam Kumaran: And then I have the one, sort of, mart.
351 00:35:12.910 ⇒ 00:35:23.449 Uttam Kumaran: thing, but anything like that, I… if you just throw it in the channel, I’m trying to use the Asana thing to just, like, quickly create tickets. Yeah. So that’s… that’ll be super easy. So, any sort of…
352 00:35:24.110 ⇒ 00:35:27.789 Uttam Kumaran: Yeah, anything on those marks, like, any feedback you get, we can make those changes.
353 00:35:27.980 ⇒ 00:35:40.749 Katherine Bayless: Yeah, and so, yeah, I mean, yeah, I will formally put in the comments about the research downloads, don’t worry, but we also told the membership team to do the same, so that they’ll be able to populate stuff to the Asana board if they see, like, quirks or questions or stuff like that.
354 00:35:41.020 ⇒ 00:35:52.440 Uttam Kumaran: Yeah, one thing that we’ve done also for another client is, like, I don’t know… I haven’t played around much with Snowflake chat in a bit, but we can add, like, feedback, so you could do thumbs up or thumbs down, or…
355 00:35:52.440 ⇒ 00:35:54.940 Katherine Bayless: Yeah. So that we can start to, like, measure…
356 00:35:55.320 ⇒ 00:35:57.640 Uttam Kumaran: some type of MPS, basically.
357 00:35:57.830 ⇒ 00:35:59.280 Katherine Bayless: Yeah. Or, like.
358 00:35:59.280 ⇒ 00:36:09.580 Uttam Kumaran: hey, this is wrong, okay, instead of it trying again, it just says, please tell me what’s wrong, so we naturally, like, QA it a bit. So we’ve… we’ve done that for… we do that for another client.
359 00:36:09.690 ⇒ 00:36:11.030 Uttam Kumaran: Yeah. Yeah.
360 00:36:11.340 ⇒ 00:36:18.200 Katherine Bayless: I want to get into, like, all of that. I mean, that’s kind of what I was talking about the other day, where it’s like, like, that, like, we need scale to work to our advantage, right?
361 00:36:18.200 ⇒ 00:36:18.720 Uttam Kumaran: Yes.
362 00:36:18.720 ⇒ 00:36:25.089 Katherine Bayless: interactions, the faster we can deliver, versus the, you know, old-school analytics of, like, now we’re bogged down in reports.
363 00:36:25.420 ⇒ 00:36:25.970 Uttam Kumaran: Yes.
364 00:36:25.970 ⇒ 00:36:37.229 Katherine Bayless: The one question I had about the thumbs up, thumbs down thing, because I know, like, in Claude, at least on the, like, you know, browser version, I disabled it, because, like, if you give it the thumbs up or thumbs down, it sends the conversation to Anthropic as feedback.
365 00:36:37.250 ⇒ 00:36:40.400 Uttam Kumaran: No, no, no, this isn’t like a… I’m not giving the model feedback.
366 00:36:40.400 ⇒ 00:36:41.150 Katherine Bayless: Right, right.
367 00:36:41.150 ⇒ 00:36:44.499 Uttam Kumaran: Giving us, like, this answer was not right.
368 00:36:44.500 ⇒ 00:36:44.950 Katherine Bayless: Yeah.
369 00:36:44.950 ⇒ 00:36:54.780 Uttam Kumaran: This answer, like, either didn’t… it timed out, or it was wrong, so that we look at it, we’re like, okay, yeah, because there’s two columns with the same name, there’s no description, we should fix that, yeah.
370 00:36:54.780 ⇒ 00:37:09.679 Katherine Bayless: Yeah, yeah, exactly, exactly, yeah, like, local, yeah. And then the same, yeah, like, I was trying to think of, like, if there are ways that, like, I could, like, embed the Asana form link, like, in Snowflake, in the, like, Streamlit apps, right? Then they can at least open it right there and put the stuff.
371 00:37:09.680 ⇒ 00:37:11.389 Uttam Kumaran: Well, like, I think we should… you should…
372 00:37:11.660 ⇒ 00:37:15.250 Uttam Kumaran: Well, we should go one step further, and it should just call…
373 00:37:15.450 ⇒ 00:37:17.950 Uttam Kumaran: an Asana endpoint, and, like, create the ticket.
374 00:37:18.680 ⇒ 00:37:19.190 Uttam Kumaran: You know.
375 00:37:19.190 ⇒ 00:37:19.880 Katherine Bayless: Yeah, yeah.
376 00:37:19.880 ⇒ 00:37:30.720 Uttam Kumaran: They don’t even think about… because they don’t even need to think about Asana, they’re just like, this is a problem, and maybe in the… in motion, in order to create the ticket, it asks a few follow-up questions, like, cool, we’ve logged this with the data team.
377 00:37:30.720 ⇒ 00:37:36.410 Katherine Bayless: Yeah, yeah, exactly. Like, yeah, that’s what I wanted to get to. I was like, first step is maybe just link the Asana form, but that’s what I want.
378 00:37:37.080 ⇒ 00:37:37.750 Katherine Bayless: Yeah.
379 00:37:37.750 ⇒ 00:37:40.720 Uttam Kumaran: Yeah, like, just have this feed itself. Yeah.
380 00:37:41.240 ⇒ 00:37:42.120 Uttam Kumaran: Okay.
381 00:37:44.440 ⇒ 00:37:45.060 Uttam Kumaran: Okay.
382 00:37:46.530 ⇒ 00:37:47.390 Uttam Kumaran: Okay.
383 00:37:49.520 ⇒ 00:37:50.050 Katherine Bayless: Yeah.
384 00:37:50.560 ⇒ 00:38:02.869 Katherine Bayless: Oh, other tiny, thing. I did ask the membership team if we use any enrichment tools currently, and it turns out the sales team uses Sales Intel, and so now membership is getting a license, and so… Oh, great.
385 00:38:02.870 ⇒ 00:38:03.490 Uttam Kumaran: Great.
386 00:38:03.490 ⇒ 00:38:19.630 Katherine Bayless: that contract is somewhere… somewhere in that pipeline. And so, yeah, so the sales intel subscription, once they have it come through, they said they’d give me a chance to, like, kind of take a look and play with it. It doesn’t mean that we couldn’t choose to use something different for, like…
387 00:38:19.630 ⇒ 00:38:25.289 Uttam Kumaran: If we already paid for something, we should just try it, at least to get them something, yeah.
388 00:38:25.290 ⇒ 00:38:44.489 Katherine Bayless: That was my thought, yeah. And then the guy in the foundation team, he said that now, and I’m totally blanking on the name of the, like, that they changed to, but, like, the GuideStar, charity navigator thing, like, we have access to their API, and I guess they changed it so that now you get two keys for the price of one, and he’s like, do you want one? I was like, yes.
389 00:38:44.560 ⇒ 00:38:56.340 Katherine Bayless: Because we do have a small slice of membership that, like, associations and nonprofits are eligible for based on certain criteria, and so I’m like, alright, well, I can hook into that for those data points.
390 00:38:56.340 ⇒ 00:39:07.509 Uttam Kumaran: Yeah, I guess, like, when we do go look at the enrichment, I guess that’s one thing I should… I did ask Omni this, but maybe I’ll just try to remember to ask everybody for what the discounts are for nonprofits.
391 00:39:07.510 ⇒ 00:39:08.750 Katherine Bayless: comments, basically.
392 00:39:09.030 ⇒ 00:39:19.659 Katherine Bayless: Yeah, we don’t tend to qualify for a lot of them because we’re a C6 and not a C3. Some places will still do it, but, like, a lot of folks, you have to be a C3 to really.
393 00:39:19.660 ⇒ 00:39:20.130 Uttam Kumaran: Okay.
394 00:39:20.130 ⇒ 00:39:21.250 Katherine Bayless: Yeah, yeah.
395 00:39:21.250 ⇒ 00:39:21.800 Uttam Kumaran: Okay.
396 00:39:22.180 ⇒ 00:39:22.750 Katherine Bayless: Okay.
397 00:39:23.960 ⇒ 00:39:25.050 Katherine Bayless: It’s worth asking, because.
398 00:39:25.050 ⇒ 00:39:25.710 Uttam Kumaran: Worth asking.
399 00:39:25.710 ⇒ 00:39:30.500 Katherine Bayless: that even though we’re a C6, they can do the, like, non-profit treatment for us.
400 00:39:30.500 ⇒ 00:39:35.249 Uttam Kumaran: I mean, some vendors, I feel like they may… it’s like they’re just putting a coupon code, like, on the thing.
401 00:39:35.250 ⇒ 00:39:36.270 Katherine Bayless: Yeah, basically.
402 00:39:36.530 ⇒ 00:39:44.279 Uttam Kumaran: I don’t know, I don’t know, see anything. I was just like, I want to see savings. Like, that’s what we want to see.
403 00:39:44.620 ⇒ 00:39:56.579 Katherine Bayless: Yeah, yeah, and I think there’s a lot of room for some of this platform consolidation-y kind of stuff. If the CES tech stack stuff does come my way, like, that opens up a lot of doors. Oh! Price tag for the Shopify scope.
404 00:39:57.220 ⇒ 00:40:02.680 Uttam Kumaran: Yes, so I am getting that. I have a draft in my inbox.
405 00:40:02.890 ⇒ 00:40:05.990 Uttam Kumaran: So I can send that over.
406 00:40:06.200 ⇒ 00:40:19.949 Katherine Bayless: Yeah, no worries. I just… I was, like, the scope looks good, Michael liked it, Casey had liked it, so they said, you know, go for it to do at least all of the research, and then we’ll regroup for the implementation pieces, but, like, I just didn’t have a price tag, so I couldn’t see it.
407 00:40:19.980 ⇒ 00:40:21.650 Uttam Kumaran: Okay, yeah, let me…
408 00:40:22.190 ⇒ 00:40:26.910 Uttam Kumaran: Yeah, I sent it on Tuesday, I got it back on Tuesday, I didn’t look at it yesterday, so…
409 00:40:26.910 ⇒ 00:40:27.729 Katherine Bayless: No worries.
410 00:40:27.730 ⇒ 00:40:28.440 Uttam Kumaran: Totally. Okay.
411 00:40:28.490 ⇒ 00:40:30.780 Katherine Bayless: I know we’re all moving, like, yeah.
412 00:40:31.260 ⇒ 00:40:33.360 Katherine Bayless: No, this is great, yeah.
413 00:40:33.360 ⇒ 00:40:40.209 Uttam Kumaran: We have some folks that have done this exact type of stuff on Shopify, so I’m pumped, like, I think we’ll be able to nail that.
414 00:40:40.380 ⇒ 00:40:40.910 Katherine Bayless: Nice.
415 00:40:41.230 ⇒ 00:40:49.849 Katherine Bayless: Yeah. Yeah. That’s a big win. I mean, it’s like a tiny piece of our overall organizational tech and financials, but, like, everybody hates it.
416 00:40:50.960 ⇒ 00:40:51.740 Uttam Kumaran: Yeah.
417 00:40:52.380 ⇒ 00:40:58.060 Uttam Kumaran: Yeah. No, I think it’s a great, like, it’s just sort of sitting there for someone to…
418 00:40:58.180 ⇒ 00:41:01.950 Katherine Bayless: Yeah. To take, you know? Yeah, exactly, exactly.
419 00:41:02.350 ⇒ 00:41:03.230 Katherine Bayless: So, yeah.
420 00:41:03.770 ⇒ 00:41:05.110 Katherine Bayless: But yes, okay.
421 00:41:05.420 ⇒ 00:41:08.190 Katherine Bayless: I think that is everything that was on my brain.
422 00:41:08.520 ⇒ 00:41:15.530 Uttam Kumaran: Okay, perfect. Then we’ll kind of try to wrap something into a presentation for tomorrow, like, in terms of where we landed this…
423 00:41:15.680 ⇒ 00:41:29.369 Uttam Kumaran: this week on March, and then… yeah, tomorrow I’ll also present a little bit on roles, and, like, how we kind of have… now that Kyle is familiar, I’ll talk a little bit of how we’re setting up dbt and the roles, and then we have that documentation in there.
424 00:41:29.530 ⇒ 00:41:33.019 Uttam Kumaran: And then we can use whatever time for a working session, so…
425 00:41:33.500 ⇒ 00:41:35.020 Katherine Bayless: Okay, cool, sounds good.
426 00:41:35.380 ⇒ 00:41:36.900 Uttam Kumaran: Perfect. Thank you.
427 00:41:36.970 ⇒ 00:41:38.819 Katherine Bayless: I’m gonna go get lunch.
428 00:41:38.820 ⇒ 00:41:39.610 Uttam Kumaran: Enjoy.
429 00:41:39.940 ⇒ 00:41:41.040 Katherine Bayless: Alright, see you later.
430 00:41:41.210 ⇒ 00:41:42.010 Uttam Kumaran: Bye.