2026-01-16_brainforge_x_cta_weekly_098c3085

Meeting Title: Brainforge x CTA: Weekly! Date: 2026-01-16 Meeting participants: Ashwini Sharma, Kyle Wandel, Uttam Kumaran, Katherine Bayless

---

WEBVTT

1 00:00:10.470 ⇒ 00:00:11.550 Ashwini Sharma: Hey, Klaid.

2 00:00:11.890 ⇒ 00:00:12.990 Kyle Wandel: Hey, how’s it going?

3 00:00:13.400 ⇒ 00:00:14.559 Ashwini Sharma: Good, how are you?

4 00:00:15.020 ⇒ 00:00:17.610 Kyle Wandel: Alright, been better.

5 00:00:17.770 ⇒ 00:00:22.220 Kyle Wandel: I’m still recovering from a flu that we got from CES, so that’s always fun.

6 00:00:22.220 ⇒ 00:00:23.959 Ashwini Sharma: Oh, man, lots of people.

7 00:00:24.290 ⇒ 00:00:32.179 Kyle Wandel: Yeah, I know of a few people already. I think Catherine also got something as well, and I think she’s joining, but I’m not 100% sure.

8 00:00:32.180 ⇒ 00:00:33.180 Ashwini Sharma: Okay.

9 00:00:35.400 ⇒ 00:00:36.339 Kyle Wandel: How are you, Ben?

10 00:00:36.860 ⇒ 00:00:38.070 Ashwini Sharma: I’ve been good.

11 00:00:39.160 ⇒ 00:00:39.840 Kyle Wandel: Good.

12 00:01:09.090 ⇒ 00:01:11.559 Kyle Wandel: Alright, Kathleen says I should be here in a few minutes, so…

13 00:01:14.120 ⇒ 00:01:15.210 Ashwini Sharma: Cool, cool, yeah.

14 00:01:18.600 ⇒ 00:01:20.529 Kyle Wandel: Do you know if, Utom is joining as well?

15 00:01:20.780 ⇒ 00:01:21.860 Ashwini Sharma: He will join.

16 00:01:22.380 ⇒ 00:01:22.950 Kyle Wandel: Cool.

17 00:02:02.100 ⇒ 00:02:03.109 Uttam Kumaran: Hey, Kyle!

18 00:02:03.740 ⇒ 00:02:05.099 Kyle Wandel: Good morning, Tom, how you doing?

19 00:02:05.290 ⇒ 00:02:06.650 Uttam Kumaran: Hey, good, how are you?

20 00:02:07.140 ⇒ 00:02:19.590 Kyle Wandel: Oh, not too bad. It’s just telling, your colleague that, had the flu, so, still have a slight fever right now. Yeah, it’s all good. It’s been, like, 4 days of it, though, so it’s been quite… it’s more annoying at this point than anything.

21 00:02:21.270 ⇒ 00:02:25.020 Uttam Kumaran: Yeah, I know a lot of people are getting sick. Here in Austin, it’s like Cedar.

22 00:02:25.320 ⇒ 00:02:32.600 Uttam Kumaran: cedar season, so it’s, like, everyone… I don’t have… like, I don’t have allergies. Everybody here is, like, getting really, really sick.

23 00:02:33.940 ⇒ 00:02:44.209 Kyle Wandel: Yeah, that’s what… I mean, we have, like, 4 or 5 colleagues, at least I know of so far, that, definitely got at least something more than a cold. I know Catherine got also pretty sick, too.

24 00:02:49.500 ⇒ 00:02:55.559 Uttam Kumaran: Cool, Catherine mentioned she’s gonna be a few minutes late, but yeah, today we just sort of wanted to…

25 00:02:55.610 ⇒ 00:03:14.090 Uttam Kumaran: kind of, like, circle back on a few deliverables. I think, Ashwini has a couple things that he closed out for you, Kyle, so there’s… that’s one thing on my agenda, is to cover that. Second, I would love to discuss with Catherine, and you, sort of, like, the deliverables for this end of…

26 00:03:14.310 ⇒ 00:03:20.950 Uttam Kumaran: CES report that’s, like, I think she mentioned is something she wants to get out the door on Friday, next Friday.

27 00:03:21.020 ⇒ 00:03:38.389 Uttam Kumaran: So, would love to discuss, like, the path there. We need… we’ll need to just confirm some access to Power BI, confirm some data sources, and then sort of just, like, drive, from here to there. And then would like to just, like, we can take some time to just talk about,

28 00:03:38.610 ⇒ 00:03:52.350 Uttam Kumaran: like, how we want to get back into normal ways of… of… of working. Like, if we want to add, like, a working session midweek, and just make sure we have time for that as, like, we’re starting to develop more quicker.

29 00:03:53.690 ⇒ 00:04:01.760 Uttam Kumaran: And then also, like, kind of talk about, like, how do we want to loop in Kai and make sure that you guys are just, really aware of, sort of, the models that we’re getting out, so…

30 00:04:02.020 ⇒ 00:04:14.989 Kyle Wandel: Yep. No, that’d be great. I think that sounds really good. I, yeah, I think I want to know more about the, end of the report… end of CES post-show report as well, because I don’t think Catherine’s looped me in on that, as much, but,

31 00:04:15.060 ⇒ 00:04:23.599 Kyle Wandel: Yeah, that all sounds good to me. I think it’ll be interesting to try to find a new normal, the next couple months. I actually am going on… I will be going on…

32 00:04:24.000 ⇒ 00:04:34.320 Kyle Wandel: leave, because, we actually have a baby due in late April, so probably in, like, August or September… August timeframe I’ll go on leave, but, still plenty of time until… until then.

33 00:04:35.190 ⇒ 00:04:35.710 Uttam Kumaran: Cool.

34 00:04:37.210 ⇒ 00:04:38.310 Uttam Kumaran: Hello, love!

35 00:04:38.640 ⇒ 00:04:40.479 Katherine Bayless: Well, hello, how’s everybody doing?

36 00:04:41.490 ⇒ 00:04:43.010 Uttam Kumaran: Good. How are you?

37 00:04:43.830 ⇒ 00:04:48.829 Katherine Bayless: Alright, okay. I guess I, I guess I’m losing my voice.

38 00:04:49.110 ⇒ 00:04:49.709 Kyle Wandel: Oh, shit.

39 00:04:50.710 ⇒ 00:04:54.980 Kyle Wandel: As you can tell, Kathy and I hung out a lot at CES, so…

40 00:04:55.160 ⇒ 00:05:03.260 Katherine Bayless: I know. Well, it’s funny, actually, I feel like we hung out for, like, 2 hours at the end of the day, right? Clearly, they were the 2 hours that mattered most.

41 00:05:03.590 ⇒ 00:05:05.520 Katherine Bayless: Sorry, go ahead, sorry.

42 00:05:05.850 ⇒ 00:05:09.050 Kyle Wandel: I think everybody just got… I think everybody just got sick, so… Yes.

43 00:05:09.050 ⇒ 00:05:19.179 Katherine Bayless: Yeah, that flu B line lit up on my test, like, quite brightly yesterday. But I feel okay today, so fingers crossed that maybe I’m through the worst of it.

44 00:05:20.120 ⇒ 00:05:20.690 Katherine Bayless: Fingers.

45 00:05:20.690 ⇒ 00:05:34.040 Kyle Wandel: I was actually… I was talking to, MRD yesterday, and I just hoped that, I didn’t get Travis Scott sick. So, yeah, I gave a, like, an hour and a half tour to Travis Scott at CES.

46 00:05:34.040 ⇒ 00:05:36.920 Uttam Kumaran: Wow, no way! That’s awesome!

47 00:05:37.610 ⇒ 00:05:42.689 Kyle Wandel: Yeah, that was really fun. He was really cool, he was dumb to earth. But it was really… I just hope I didn’t get him sick, because that was the last day.

48 00:05:42.690 ⇒ 00:05:46.379 Uttam Kumaran: What was… what was he asking about? Like, did he ask about any part of the…

49 00:05:46.900 ⇒ 00:05:49.190 Uttam Kumaran: Do you show, in particular? Yeah, yeah.

50 00:05:49.190 ⇒ 00:06:06.539 Kyle Wandel: He had never been before, so he really wasn’t, like, really too… he didn’t really care too much, he just liked to see, like, the big… the big things, so, like, the cars, automotives. He looked at a boat, he got inside of a tractor. We walked by humanoid robots, and I thought, I’ll never forget this. He goes, man, I don’t know about those. And I said, yeah, they’re not… they’re not great.

51 00:06:06.880 ⇒ 00:06:26.620 Kyle Wandel: And then he really loved this. We had a… we had a vendor set up, or exhibitor set up a piano robot, basically, and they played one of his songs. So, like, in 12 hours, the next day, I was like, I went to them, the day before, and was like, hey, can you guys… I’m gonna bring Travis Scott here, what do you guys think? And they did a nice little piano of goosebumps, and that was really cool.

52 00:06:27.140 ⇒ 00:06:27.750 Uttam Kumaran: Nice.

53 00:06:28.240 ⇒ 00:06:29.190 Kyle Wandel: Overall, good time.

54 00:06:29.700 ⇒ 00:06:31.000 Uttam Kumaran: That’s awesome.

55 00:06:31.390 ⇒ 00:06:51.210 Katherine Bayless: The best part was, I texted a friend, like, a screen grab of Kyle messaging me in Slack, because I was like, hey, can you help with X? And he’s like, yes, definitely, but might have to give a tour to Travis Scott. And so I texted it to my friends, like, bullshit! And I was like, no, seriously! This is my weird life. Have you ever managed a data engineer who also gives tours?

56 00:06:51.210 ⇒ 00:06:53.219 Katherine Bayless: I know, I was gonna say, like.

57 00:06:53.220 ⇒ 00:06:56.600 Uttam Kumaran: There’s not someone else that does the tours.

58 00:06:56.600 ⇒ 00:07:06.650 Kyle Wandel: So normally, MRD does all the tours, but this is, like, the first tour I’ve ever, like, hosted, so I’m hoping that they ask me to come back and do it again, because anything I can do to avoid Reg, I’m gonna do that.

59 00:07:07.030 ⇒ 00:07:09.849 Katherine Bayless: Yeah, we’re not doing reg. Yeah, it’s like the most…

60 00:07:09.850 ⇒ 00:07:10.610 Uttam Kumaran: Can I tell you guys a note?

61 00:07:10.810 ⇒ 00:07:12.429 Katherine Bayless: Other duties as assigned.

62 00:07:13.030 ⇒ 00:07:19.720 Uttam Kumaran: Can I tell you guys another funny, like, you’re just, like, no data people, and I’m in this data chat with a bunch of, like, growth…

63 00:07:19.870 ⇒ 00:07:27.329 Uttam Kumaran: data folks, and someone got a quote for what it would cost to advertise on the Sphere, and it’s like.

64 00:07:27.870 ⇒ 00:07:29.969 Uttam Kumaran: insane. It’s like…

65 00:07:29.970 ⇒ 00:07:30.360 Katherine Bayless: Yeah.

66 00:07:30.360 ⇒ 00:07:32.709 Uttam Kumaran: The most money ever!

67 00:07:32.710 ⇒ 00:07:34.989 Katherine Bayless: Really? Actually, I mean…

68 00:07:35.440 ⇒ 00:07:42.829 Katherine Bayless: Can you share, like, a… an over-under? Yeah, let me… because I am curious, after the Lenovo buyout, like, what does that cost?

69 00:07:42.830 ⇒ 00:07:44.650 Uttam Kumaran: Yeah, let me look, I,

70 00:07:44.920 ⇒ 00:07:51.820 Uttam Kumaran: I wouldn’t even do justice, I think, if I, like… So basically, they have, like, two packages. There’s, like.

71 00:07:52.090 ⇒ 00:07:53.409 Uttam Kumaran: One week.

72 00:07:53.680 ⇒ 00:07:56.440 Uttam Kumaran: Takeover, and then there’s, like, a full day.

73 00:07:56.600 ⇒ 00:08:02.059 Uttam Kumaran: one week is, like, you sort of just, like, cycle in. Full day is, like.

74 00:08:03.380 ⇒ 00:08:05.250 Uttam Kumaran: Around, like, a million dollars.

75 00:08:05.560 ⇒ 00:08:13.340 Kyle Wandel: Yeah. That’s what I thought, yeah. I think… I think we heard about… heard about this, like, a few years ago. I think it’s a million dollars, for people to do it.

76 00:08:13.570 ⇒ 00:08:16.490 Katherine Bayless: Pretty crazy That is true. But that’s still non…

77 00:08:16.490 ⇒ 00:08:16.930 Kyle Wandel: It’s face.

78 00:08:16.930 ⇒ 00:08:20.520 Uttam Kumaran: as Super Bowl, though, which is even more wild, in my opinion.

79 00:08:20.780 ⇒ 00:08:40.949 Uttam Kumaran: It definitely is not as expensive as Super Bowl, and so we did… I was… I worked at Flow Code, which is a QR code company. We did a lot… we really pushed, like, the reason why QR is, like, on TV. This is, like, right during the start of COVID, like, in between COVID, and so we helped some brands get onto Super Bowl. We did a lot of work with CNN, Fox, NBC.

80 00:08:40.950 ⇒ 00:08:44.090 Uttam Kumaran: And yeah, like, a 60-second slot.

81 00:08:44.090 ⇒ 00:08:48.560 Uttam Kumaran: Super… it’s also, it’s funny, they do a Super Bowls, like, Really, it’s… it’s like…

82 00:08:48.750 ⇒ 00:08:55.409 Uttam Kumaran: All these various pockets, like, right after halftime show, right after each of the quarters.

83 00:08:57.040 ⇒ 00:09:02.250 Uttam Kumaran: And, yeah, it’s… it’s also extremely, extremely expensive. Like, per second, I think…

84 00:09:02.400 ⇒ 00:09:06.349 Uttam Kumaran: Maybe the most expensive ad slot,

85 00:09:06.690 ⇒ 00:09:10.280 Uttam Kumaran: Maybe that and, like, World… maybe World Cup Final, you know?

86 00:09:11.280 ⇒ 00:09:21.480 Katherine Bayless: Yeah, it’s a good point. I’ve always wondered about the comparison to, like, World Cup Finals, because, like, with the Super Bowl, you do get more commercials, right? Like, overall… well, I don’t know, I guess about overall, but at least you get the.

87 00:09:21.480 ⇒ 00:09:22.000 Uttam Kumaran: licks.

88 00:09:22.000 ⇒ 00:09:24.390 Katherine Bayless: First, whereas, like, World Cup, you get one break.

89 00:09:25.030 ⇒ 00:09:33.219 Uttam Kumaran: Yeah, like, I mean, for football, I think it’s… it’s 4 quarters, 48 minutes. I think there’s typically only, like, 10 minutes of, like.

90 00:09:33.410 ⇒ 00:09:41.060 Uttam Kumaran: like, I think… I think it’s, like, 30 minutes of actual, like, game time, and then it’s, like, typically it goes over 2 hours.

91 00:09:41.220 ⇒ 00:09:45.510 Uttam Kumaran: So they… Pile in the ads.

92 00:09:46.400 ⇒ 00:09:48.780 Katherine Bayless: No, I guess, more of an opportunity.

93 00:09:48.780 ⇒ 00:09:49.300 Kyle Wandel: Yep.

94 00:09:49.300 ⇒ 00:09:49.750 Uttam Kumaran: Yeah.

95 00:09:49.750 ⇒ 00:10:02.049 Kyle Wandel: Like, just looking, like, 1 point something billion people watched the World Cup last time, and only around 120 million U.S. viewers for the Super Bowl, so that’s a pretty big difference.

96 00:10:02.410 ⇒ 00:10:05.970 Uttam Kumaran: Yeah, but it’s also the propensity to purchase, I think, like.

97 00:10:06.120 ⇒ 00:10:06.690 Kyle Wandel: Yeah, yeah.

98 00:10:06.690 ⇒ 00:10:09.070 Uttam Kumaran: You know, for, like, the US brands, like.

99 00:10:09.200 ⇒ 00:10:12.519 Uttam Kumaran: It’s… but, like, for example, we used to do, like, primetime…

100 00:10:12.840 ⇒ 00:10:17.339 Uttam Kumaran: Fox would be, like, 60 grand for, like, a 30-second slot.

101 00:10:17.560 ⇒ 00:10:18.510 Uttam Kumaran: like…

102 00:10:18.820 ⇒ 00:10:27.410 Uttam Kumaran: And that’s happening, like, every day. It’s just an interesting TV world. And the reporting is horrible. You get no data back. Like, no decent data back.

103 00:10:27.780 ⇒ 00:10:30.510 Uttam Kumaran: Yeah.

104 00:10:32.040 ⇒ 00:10:33.119 Uttam Kumaran: That’s true.

105 00:10:33.120 ⇒ 00:10:47.139 Katherine Bayless: funny about scale numbers. I was at the delegation leader’s dinner, and comically, it was in a steakhouse, but I’m vegetarian, and I ended up seated at the table with the delegation from India, so it was the veggie table at the steakhouse. But they asked.

106 00:10:47.140 ⇒ 00:10:47.790 Uttam Kumaran: It’s coming.

107 00:10:48.070 ⇒ 00:10:59.829 Katherine Bayless: like, CES costs, and I was like, well, so it’s about $100 million of our revenue, and so you figure we’re a non-profit, probably costs most of that, right? And they’re just like, how much is that in billions? And I was like.

108 00:11:00.360 ⇒ 00:11:08.939 Katherine Bayless: I… I don’t… I don’t really know how to answer that question. I was like, I give you the conversion, but, like, I don’t… have I underwhelmed? I’m very confused, but they’re like, I don’t.

109 00:11:08.940 ⇒ 00:11:10.820 Uttam Kumaran: 0.1?

110 00:11:10.820 ⇒ 00:11:15.399 Katherine Bayless: Yeah, yeah, they were like, how many billions? No billions, it is zero billions.

111 00:11:17.530 ⇒ 00:11:19.089 Katherine Bayless: So, anyway.

112 00:11:20.180 ⇒ 00:11:21.030 Katherine Bayless: Congratulations.

113 00:11:21.560 ⇒ 00:11:38.949 Uttam Kumaran: Yeah, maybe I wanted to… so I kind of, like, wanted to separate this conversation into a couple different pieces. One, I wanted to talk about immediate, like, these next two weeks to hit next Friday’s, like, trying to get out the post-show report. So I do want to first talk about Ashwini, the work you’ve done so far, and just, like.

114 00:11:38.970 ⇒ 00:11:43.279 Uttam Kumaran: reorient the team on, like, where we’re at. I also do want to talk about

115 00:11:43.410 ⇒ 00:11:55.779 Uttam Kumaran: the gap between, like, what we do need to produce for next Friday. So those are the first two items. The next item is, like, just confirming we have Power BI edit access, and we can sort of develop that.

116 00:11:55.950 ⇒ 00:11:58.539 Uttam Kumaran: That, roughly, will kind of, like.

117 00:11:58.820 ⇒ 00:12:10.140 Uttam Kumaran: circle out everything for the next two weeks, and then I can… we can talk about a couple things beyond that. So maybe, Ashwini, you want to go first? We can talk about just, like, to date, the work we’ve done, and then let’s talk about

118 00:12:10.310 ⇒ 00:12:15.540 Uttam Kumaran: The merits, the merits data we got, and then, like, yeah, kind of, like, what else is remaining.

119 00:12:16.770 ⇒ 00:12:19.519 Ashwini Sharma: Are you guys able to see my screen?

120 00:12:19.770 ⇒ 00:12:20.690 Uttam Kumaran: Yes.

121 00:12:20.780 ⇒ 00:12:24.070 Katherine Bayless: Yeah, so I was working on this before.

122 00:12:24.280 ⇒ 00:12:28.440 Ashwini Sharma: last year, before you went for the CES thing, and…

123 00:12:28.650 ⇒ 00:12:37.250 Ashwini Sharma: So using this, using the data that you have shared in S3, I think I’m able to derive this portion clearly.

124 00:12:37.370 ⇒ 00:12:42.220 Ashwini Sharma: I’m not able to derive these things, and not even this part.

125 00:12:43.220 ⇒ 00:12:43.760 Katherine Bayless: Okay.

126 00:12:44.000 ⇒ 00:12:48.570 Ashwini Sharma: Yeah, and some of this part is derivable, but not everything.

127 00:12:49.080 ⇒ 00:12:56.679 Ashwini Sharma: Go down. This was, I think, this could be derived, EV bill…

128 00:12:57.210 ⇒ 00:13:02.780 Ashwini Sharma: And, again, it’s not clear how am I going to derive these numbers.

129 00:13:03.960 ⇒ 00:13:04.710 Katherine Bayless: Yeah.

130 00:13:05.460 ⇒ 00:13:18.410 Ashwini Sharma: And, I mean, the more important thing about the extract that you have shared, I think we’ll have to think of a way to formalize that pipeline, rather than, you know, getting an extract and then creating a table out of CSVs, because

131 00:13:18.540 ⇒ 00:13:21.679 Ashwini Sharma: Those extracts are not, standard.

132 00:13:22.590 ⇒ 00:13:28.379 Ashwini Sharma: Like, the attendance extracts, especially, like, if you see, across the different years.

133 00:13:29.300 ⇒ 00:13:31.809 Ashwini Sharma: It’s completely different. It does not match.

134 00:13:31.930 ⇒ 00:13:32.860 Katherine Bayless: So…

135 00:13:33.880 ⇒ 00:13:37.880 Ashwini Sharma: Yeah, so a formal pipeline would really help.

136 00:13:39.030 ⇒ 00:13:48.399 Katherine Bayless: I think, yeah, so a couple… a couple thoughts on that, which actually, this was on my mind to bring up, so excellent segue. So…

137 00:13:48.870 ⇒ 00:14:06.510 Katherine Bayless: specifically on the historical registration data sets and that issue with the schemas not matching, they do tend to match a little bit more for, like, 23, 24, 25, 26 is the most different, but there’s variations between all the years, for sure.

138 00:14:06.510 ⇒ 00:14:20.429 Katherine Bayless: But, it is also… those flat files are all we have for it, so I think probably, like, a one-time effort to say, okay, these are gonna be the canonical flat files for those prior year registrations.

139 00:14:20.430 ⇒ 00:14:43.410 Katherine Bayless: we will store them here, we will harmonize and model them this way, and then going forward, we’ll use that sort of harmonized and modeled version of the data. Because, yeah, there’s just no ability for us to draw that from the source anymore at this point. And so, if we start with 23, 24, 25, 26, and harmonize and model those, I think that’s more than enough for the moment.

140 00:14:43.410 ⇒ 00:14:52.970 Katherine Bayless: And then we can eventually kind of work at adding backwards the additional previous years that we have. Kyle, I saw you come off mute, so if you want to add anything to that.

141 00:14:53.230 ⇒ 00:15:15.979 Kyle Wandel: Yeah, yeah, I do have a… at least a start, slash, I think it’s a pretty good overview of all the master col… like, I call it the master column sheet, so that goes back from 2019 to 25, I believe. I’m just doing my best to map all of the… basically, regs data, that’s mainly what it is, is regs data. Try to map as much regist data as I possibly can and make it uniform.

142 00:15:16.010 ⇒ 00:15:21.709 Kyle Wandel: So I do have something like that, and I can share that, but it’s also in the, data warehouse as well.

143 00:15:24.150 ⇒ 00:15:27.060 Katherine Bayless: And then, sort of similarly, or…

144 00:15:27.310 ⇒ 00:15:35.480 Katherine Bayless: Relatedly, of… okay, wait, hang on, I can’t type in… or write and say the same thing. There we go.

145 00:15:35.640 ⇒ 00:15:54.340 Katherine Bayless: I guess… so I know that, like, for the purposes of kind of getting you going on the member engagement report, I just kind of, like, tossed everything in an S3 bucket, and, you know, we went from there, but I think knowing that we’ll have the dependency on some of these flat files for the archive.

146 00:15:54.340 ⇒ 00:16:05.800 Katherine Bayless: Probably is a good idea to have a… at least, like, a quick whirlpool around, like, okay, what S3 bucket do we want to be the source of flat files for Snowflake data?

147 00:16:05.900 ⇒ 00:16:19.809 Katherine Bayless: where am I mistaken, and we should put them into a database, and then connect to Snowflake? You know, I’m kind of agnostic, but, like, where do we want to store these old flat files? And then… yeah, actually, I guess really that’s the whole question is, where exactly should we store them?

148 00:16:20.100 ⇒ 00:16:21.210 Katherine Bayless: Officially.

149 00:16:23.190 ⇒ 00:16:25.119 Uttam Kumaran: I would vote for S3.

150 00:16:25.120 ⇒ 00:16:26.210 Katherine Bayless: Yeah.

151 00:16:26.210 ⇒ 00:16:27.329 Uttam Kumaran: time, because…

152 00:16:27.770 ⇒ 00:16:28.270 Katherine Bayless: Yeah.

153 00:16:28.270 ⇒ 00:16:43.709 Uttam Kumaran: You know, Snowflake is, like, it’s sort of virtual, and, like, we may want to reload things or move things around. We don’t want, like, we can… Snowflake has really great retention policy, so it’s, like, even if it gets dropped, it’s not, like, the worst thing, but we may want to use those files

154 00:16:43.760 ⇒ 00:16:49.800 Uttam Kumaran: like, in different ways, so I would suggest, like, having S3 be, like.

155 00:16:50.310 ⇒ 00:16:55.800 Uttam Kumaran: the primary data store, even for, like, archive, or… Like, historical data.

156 00:16:55.970 ⇒ 00:16:57.940 Uttam Kumaran: And…

157 00:16:57.940 ⇒ 00:16:58.560 Katherine Bayless: That would…

158 00:16:58.560 ⇒ 00:17:08.210 Uttam Kumaran: you know, even when we do, like, if we kind of collaborate on the AWS stuff, we’ll set, like, retention policies, so again, you just want to make sure that there’s, like, not a risk of that getting lost.

159 00:17:08.900 ⇒ 00:17:16.500 Uttam Kumaran: And that way, it’s like, even if you… even if you come across, Catherine, like, data where you’re like, I don’t know if this is gonna be useful, just, like, place a shove it.

160 00:17:16.770 ⇒ 00:17:23.430 Uttam Kumaran: So that we can start turning off some systems, or, like, going away from them, and then it’s at least somewhere, you know?

161 00:17:24.349 ⇒ 00:17:41.669 Katherine Bayless: Yeah, that… okay, that was kind of my instinct. So I have an S3 bucket that I had set up that is, version controlled and has the retention policy. Like, I did all the homework on the front end kind of thing to make it our data lake inside a folder in there that was archived.

162 00:17:41.739 ⇒ 00:17:49.809 Katherine Bayless: And so I can start just kind of using that as where we will store this stuff, if that makes the most sense, yeah.

163 00:17:51.729 ⇒ 00:17:53.169 Katherine Bayless: Okay, cool, cool.

164 00:18:00.010 ⇒ 00:18:08.779 Ashwini Sharma: And, yeah, what I would need, from you or Kyle would be some of the definitions that can help me deriving these numbers, right?

165 00:18:09.290 ⇒ 00:18:17.539 Ashwini Sharma: Because, like, just by looking at the extract, it’s quite difficult for me to figure out what rules should I apply on those data to

166 00:18:17.760 ⇒ 00:18:22.070 Ashwini Sharma: To extract something like this, like, $29,174.

167 00:18:23.070 ⇒ 00:18:38.940 Katherine Bayless: Yeah, I’ll defer to Kyle, honestly, on those, because I’m… I’m not sure if they’re coming out of the files versus, like, they might just be calculations based on, like, what we know members save, or something along those lines, but…

168 00:18:38.940 ⇒ 00:18:43.350 Kyle Wandel: I think it’s… I think it’s based on the amount of money they saved

169 00:18:44.380 ⇒ 00:19:01.369 Kyle Wandel: doing XYZ. So, for registration, that company may have gotten free registration for X number of people, and that’s that 29,000. To be quite frank, I don’t know how they calculate that. This would be, like, an Adrian, this would be a question for our CS operations.

170 00:19:02.300 ⇒ 00:19:18.300 Kyle Wandel: we would need to go and kind of calculate that and figure out the logic behind it, but yeah, you’re right. I think that’s what they were doing on the back end, is that flat… they’d create these flat files based on logic, like, and then they would just upload those flat files to the database, and then they would use those flat files as the Power BI.

171 00:19:18.520 ⇒ 00:19:19.570 Kyle Wandel: dataset.

172 00:19:20.310 ⇒ 00:19:26.929 Katherine Bayless: Fascinating. Okay, yeah, so, yeah, then if we can figure out the logic for it, we can create that in code. Okay.

173 00:19:37.380 ⇒ 00:19:44.290 Ashwini Sharma: And sometimes the numbers have not matched. Some, like, the winners, right, where is that?

174 00:19:44.780 ⇒ 00:19:46.200 Katherine Bayless: Yeah, I think we’re gonna…

175 00:19:46.200 ⇒ 00:19:46.700 Ashwini Sharma: Yeah.

176 00:19:46.870 ⇒ 00:19:59.390 Katherine Bayless: A lot of numbers that don’t match, yeah, we’ll probably have to come up with a decent, like, you know, gut instinct margin of error that’s acceptable, like, if it’s off by, you know, a handful, sure. If it’s off by.

177 00:19:59.390 ⇒ 00:20:03.380 Uttam Kumaran: Well, at least, like, we’ll be able to back into everything, right, from our data.

178 00:20:03.540 ⇒ 00:20:06.529 Uttam Kumaran: So, if there is something that is, like.

179 00:20:06.780 ⇒ 00:20:12.269 Uttam Kumaran: Let’s try… let’s try to explain, at least on our side, we can go all the way back to the… to the rows that contribute.

180 00:20:12.760 ⇒ 00:20:22.310 Ashwini Sharma: How about this one? Like, this seems easy, but it’s not really. How do you determine the exhibit status for a particular year?

181 00:20:23.270 ⇒ 00:20:48.030 Katherine Bayless: This actually… so we also, similar to the reg data, we have exported, flat files that are the canonical archive for this data. I might have just not given them to you, so that might be on me, but those will be easy to pull once I give you the data. We might have to do the same, like, one-time harmonization, but I actually am not sure we will, because I think the exhibit data has stayed in the same structure for a few years now.

182 00:20:48.290 ⇒ 00:20:51.249 Katherine Bayless: But yeah, I’ll just make sure I actually gave you those files.

183 00:20:54.410 ⇒ 00:20:57.689 Ashwini Sharma: Alright. Yeah, I was looking at this one, right, and then…

184 00:20:57.930 ⇒ 00:21:01.929 Ashwini Sharma: for 2023, the max was exhibitors, so maybe, like…

185 00:21:02.070 ⇒ 00:21:07.720 Ashwini Sharma: You know, you map this thing over here, but then for 2025, it’s blank, so clearly that’s not the case.

186 00:21:09.540 ⇒ 00:21:13.730 Katherine Bayless: Yeah, yeah, so I guess, actually, that’s a good distinction, too. So there’s…

187 00:21:14.410 ⇒ 00:21:23.859 Katherine Bayless: There’s two data sets that’ll both contain exhibitor information with different lenses. So, in the registration data.

188 00:21:23.860 ⇒ 00:21:30.500 Katherine Bayless: we’ll see all of the people registered for the conference, and if they had an exhibitor ID listed.

189 00:21:30.500 ⇒ 00:21:50.009 Katherine Bayless: Along with them, and, like, so then their registration type was exhibit, or exhibitor. Then, separately, there’s the actual, like, data set of exhibitors, which is at the company level, and is whether or not the company was an exhibitor. I happened to have noticed this morning that we have, like.

190 00:21:50.150 ⇒ 00:22:11.019 Katherine Bayless: I think it was, like, 200 or so exhibitors that look like they have no registrants, and so I’m… I’m sure there’s a logical explanation there, I just… flu brain wasn’t coming up with it. But yeah, so, like, the registration data will give you people tied to exhibitors, and then the exhibitor data will give you whether or not the company itself was an exhibitor.

191 00:22:12.730 ⇒ 00:22:29.739 Ashwini Sharma: So, maybe, I think, like, you know, between Kyle, you, and if you could add some more folks from your team, like, one session on this report, if the delivery… if the goal is to deliver this report from Snowflake tables, I would like to have one session with you.

192 00:22:29.990 ⇒ 00:22:35.750 Ashwini Sharma: Where we can define exactly, you know, what data source is required to extract…

193 00:22:36.190 ⇒ 00:22:38.750 Ashwini Sharma: You know, this individual section, right?

194 00:22:39.060 ⇒ 00:22:45.420 Ashwini Sharma: And what’s the logic, right? So once we have that, I can start working on the marts.

195 00:22:46.140 ⇒ 00:22:54.640 Ashwini Sharma: And then I’ll expose that, using Power BI. And that reminds me, we don’t have access to Power BI, so maybe that’s something we can…

196 00:22:54.640 ⇒ 00:22:56.650 Uttam Kumaran: Okay, yeah, great, great, great.

197 00:22:57.290 ⇒ 00:22:57.989 Katherine Bayless: You can get that.

198 00:23:00.940 ⇒ 00:23:08.369 Uttam Kumaran: Yeah, and then ideally, like, when we get access to Power BI, Catherine, it’d be good for us to just go through and, like, do the Snowflake connection live.

199 00:23:08.950 ⇒ 00:23:11.230 Katherine Bayless: To just see, like, what happens.

200 00:23:11.830 ⇒ 00:23:20.080 Katherine Bayless: Yeah, that was actually Kyle’s thought, too, was like, maybe we can find some documentation, or maybe we can just all get on a call and see if it works, yeah. Yeah.

201 00:23:20.780 ⇒ 00:23:26.360 Katherine Bayless: Yeah, if you scroll back up a little bit, though, so the committee stuff…

202 00:23:27.150 ⇒ 00:23:28.939 Ashwini Sharma: Siri, which one?

203 00:23:29.640 ⇒ 00:23:32.089 Katherine Bayless: One… there we go, committee participation.

204 00:23:32.090 ⇒ 00:23:32.680 Ashwini Sharma: Yeah.

205 00:23:32.680 ⇒ 00:23:49.220 Katherine Bayless: So this data… this should actually be in the remembers, data share. To be totally honest, I’m not 100% sure exactly how it’s modeled. I do find that AMS systems tend to model committees

206 00:23:49.870 ⇒ 00:24:01.679 Katherine Bayless: interestingly. Not terribly difficultly, but usually there’s some, like, quirks to the data. But this all should be in there somewhere. One sort of trick is that

207 00:24:02.760 ⇒ 00:24:06.980 Katherine Bayless: I think we will consider the data in Remembers to be authoritative.

208 00:24:06.980 ⇒ 00:24:16.890 Katherine Bayless: But I think, anecdotally, it is known to not be authoritative. There is another system where a lot of these committees get managed that’s called Causeway.

209 00:24:16.890 ⇒ 00:24:29.339 Katherine Bayless: there is a desire from that team to have us integrate it, and I think there’s actually some difficulty with the integration between Causeway and remembers, which is why the committee data might not be quite correct, but for the moment.

210 00:24:29.460 ⇒ 00:24:36.330 Katherine Bayless: we’ll consider the remembers data authoritative, we just eventually probably need to fix the connection out to Causeway.

211 00:24:38.310 ⇒ 00:24:41.409 Ashwini Sharma: Is this still part of the CRM, this committee thing?

212 00:24:43.210 ⇒ 00:24:49.240 Katherine Bayless: I think it probably would be, I don’t think it would be in one of the other…

213 00:24:49.240 ⇒ 00:24:51.150 Ashwini Sharma: committees are there in CRM only.

214 00:24:51.150 ⇒ 00:24:52.950 Katherine Bayless: Okay, okay, okay, yeah.

215 00:24:58.600 ⇒ 00:24:59.230 Uttam Kumaran: Okay.

216 00:24:59.230 ⇒ 00:25:02.720 Ashwini Sharma: Alright, third.

217 00:25:16.580 ⇒ 00:25:20.510 Uttam Kumaran: Do we want to talk to Ashwini, about the scanner data next?

218 00:25:21.340 ⇒ 00:25:21.890 Ashwini Sharma: Oh, yeah.

219 00:25:21.890 ⇒ 00:25:28.150 Uttam Kumaran: And you also mentioned that there was some stuff missing from the initial… data dump, right? So…

220 00:25:28.430 ⇒ 00:25:28.950 Ashwini Sharma: Oh, there too.

221 00:25:28.950 ⇒ 00:25:29.730 Uttam Kumaran: those topics.

222 00:25:29.730 ⇒ 00:25:32.729 Ashwini Sharma: Yeah, hold on a second, so…

223 00:25:35.250 ⇒ 00:25:40.789 Ashwini Sharma: This is an Excel… I’ll convert it to CHV and then put it, unless…

224 00:25:40.910 ⇒ 00:25:44.939 Ashwini Sharma: This has multiple tabs, I’m not sure what exactly is over there.

225 00:25:45.850 ⇒ 00:25:55.229 Katherine Bayless: If it has multiple tabs, the second tab is just, like, an auto-generated, like, these are the parameters of this report, and so you could delete it. There’s only one meaningful tab in there.

226 00:25:55.560 ⇒ 00:25:56.179 Katherine Bayless: But I think it’s.

227 00:25:56.180 ⇒ 00:25:57.610 Ashwini Sharma: Yeah.

228 00:25:58.550 ⇒ 00:26:06.429 Ashwini Sharma: So, basically, it was this, right? This is where you put most of the data. Oh, no, not this one, sorry.

229 00:26:06.930 ⇒ 00:26:21.129 Ashwini Sharma: It was member engagement reports. So, basically, up till EB bill, we have the CSV file, but after that, for example, event attendance, there’s no data, matchmaker, nothing.

230 00:26:21.450 ⇒ 00:26:27.179 Ashwini Sharma: Media opportunities. Yeah, so basically these data sets are missing, that’s what…

231 00:26:27.580 ⇒ 00:26:29.310 Ashwini Sharma: I was talking to Utam about.

232 00:26:30.230 ⇒ 00:26:41.269 Katherine Bayless: Gotcha. Yeah, so I think what I can do for the immediate interim is I can export the old data and just drop it in here. I do think…

233 00:26:41.380 ⇒ 00:26:59.649 Katherine Bayless: probably Kyle and I need some… need to do a little homework on figuring out where some of these came from in the first place. Like, the event attendance, I think, came out of Zoom and Cvent. The matchmaker stuff, though, I’m not sure. And media opportunities, I think, Kyle, you and I were like, that’s probably someone’s, like, manual spreadsheet running around somewhere.

234 00:27:00.100 ⇒ 00:27:04.289 Kyle Wandel: Yeah, my guess is CAT would be the one for that, but I’m not 100% sure.

235 00:27:06.920 ⇒ 00:27:16.889 Kyle Wandel: Yeah, a lot of these are in DB… are in the Postgres data warehouse that we have access to, and I can maybe try to look through and kind of…

236 00:27:17.450 ⇒ 00:27:19.650 Kyle Wandel: Map which ones are which, basically.

237 00:27:21.130 ⇒ 00:27:43.770 Katherine Bayless: Yeah, it’s probably a good segue activity to the other kind of open question in our Q1 work around, like, from that old data warehouse, what things do we need to, like… I mean, we’ll keep all of it, because I’m a data hoarder, but, like, what things do we need to actually, like, model, because they are the only archival data source, and so…

238 00:27:43.770 ⇒ 00:27:46.430 Katherine Bayless: Like, yeah, the event attendance stuff.

239 00:27:46.430 ⇒ 00:27:51.629 Katherine Bayless: we can probably get it from Zoom and Cvent going forward, but in terms of the historical

240 00:27:51.630 ⇒ 00:28:10.360 Katherine Bayless: probably the old data warehouse is the most authoritative there, and then the media opportunity is kind of the same. So it might help us with whittling down that list of 400 tables, like, okay, well, these are the ones that we actually need to continue considering source of truth, versus these are the ones that we’ll just hang on to in case it comes up.

241 00:28:11.040 ⇒ 00:28:11.650 Uttam Kumaran: Okay.

242 00:28:18.910 ⇒ 00:28:20.879 Katherine Bayless: Yeah, no, that was just the peak of it.

243 00:28:22.170 ⇒ 00:28:28.640 Katherine Bayless: So in the exhibit space one, did I put the files in there? Because those would be those historical exhibitor data sets.

244 00:28:28.640 ⇒ 00:28:29.350 Ashwini Sharma: Yeah, dude.

245 00:28:29.350 ⇒ 00:28:39.549 Katherine Bayless: Yeah, okay. So those are the ones where you would be able to derive the, exhibit status, for that piece of the Power BI report.

246 00:28:39.730 ⇒ 00:28:42.850 Ashwini Sharma: Oh, okay. Okay, that’s cool then.

247 00:28:43.280 ⇒ 00:28:48.210 Ashwini Sharma: I think this is the dataset that I was talking about, which is highly messed up.

248 00:28:48.210 ⇒ 00:28:50.550 Katherine Bayless: Exactly.

249 00:28:54.970 ⇒ 00:28:56.990 Uttam Kumaran: So I think we’ll just piece it together.

250 00:28:59.370 ⇒ 00:29:03.570 Kyle Wandel: Honestly, that’s one of our best data sets. Most uniform.

251 00:29:05.110 ⇒ 00:29:08.879 Katherine Bayless: Yeah, exactly. It’s a good intro.

252 00:29:13.370 ⇒ 00:29:18.439 Katherine Bayless: And then I guess what I’ll do is, I’ll move these files

253 00:29:19.080 ⇒ 00:29:26.729 Katherine Bayless: into… excuse me, into the, like, data lake bucket, so that we’re starting to use… okay, yeah, yeah, cool.

254 00:29:26.990 ⇒ 00:29:45.020 Kyle Wandel: And, Catherine, I can help you with that as well, not… maybe not the moving itself, but understanding which ones to move, because I did start moving over some of them from the old MigrationDB to the Postgres, so I have at least a decent understanding of at least some of the historical stuff, and so I can send that to you.

255 00:29:45.610 ⇒ 00:29:49.500 Katherine Bayless: Okay, yeah, totally. That would be awesome if we can kind of divide and cover.

256 00:29:50.180 ⇒ 00:29:56.690 Ashwini Sharma: So, this scanner data, does it derive anything in the… in the… this report? Member engagement report?

257 00:29:58.270 ⇒ 00:30:02.360 Katherine Bayless: Good question. I don’t think so, Kyle.

258 00:30:02.730 ⇒ 00:30:07.599 Kyle Wandel: That’s a good question. I don’t… I’m looking at it now on,

259 00:30:08.610 ⇒ 00:30:15.990 Kyle Wandel: This one. It would be under other engagements, so, like, the member lounge, I think.

260 00:30:17.670 ⇒ 00:30:19.160 Kyle Wandel: But I’m not 100% sure.

261 00:30:24.280 ⇒ 00:30:28.880 Kyle Wandel: Yeah, if it’s anywhere, it’d be under… it’d be under other engagements, but then… even then, I’m not sure.

262 00:30:29.560 ⇒ 00:30:30.260 Katherine Bayless: Yeah.

263 00:30:30.930 ⇒ 00:30:32.599 Ashwini Sharma: Not letting me reference this one.

264 00:30:41.060 ⇒ 00:30:57.900 Katherine Bayless: It probably should go into the member engagement report, honestly. I’m like, yeah, it’d be cool to add a, like, a call-out in that report that’s like, you know, you had, you know, 100 people attend CES, and then, you know, you also had those people attended, you know, these 5 sessions while they were there.

265 00:31:01.100 ⇒ 00:31:02.440 Kyle Wandel: Yeah, I agree.

266 00:31:04.000 ⇒ 00:31:04.790 Kyle Wandel: Hmm.

267 00:31:04.790 ⇒ 00:31:08.050 Ashwini Sharma: This is just one tab, and this is the data, okay?

268 00:31:08.500 ⇒ 00:31:10.440 Katherine Bayless: Yeah, yeah, yeah, okay.

269 00:31:12.330 ⇒ 00:31:22.529 Katherine Bayless: Yeah, and so, in this dataset, like, the way I’ve been working with it, I’ve ignored the address and phone information, because it’s redundant with the…

270 00:31:22.530 ⇒ 00:31:37.600 Katherine Bayless: registration data set, although I should caveat that and say I would not be surprised to find out that they are different, but I’m not worried about people’s mailing addresses and phone numbers right yet. So I was just kind of ignoring those pieces, but maintaining,

271 00:31:37.680 ⇒ 00:31:42.760 Katherine Bayless: Yeah, really just kind of the ID, the dates, times, statuses…

272 00:31:43.300 ⇒ 00:31:48.839 Katherine Bayless: Emails, reg types, yeah, so it’s, like, just eliminating the contact information except email, basically.

273 00:31:49.840 ⇒ 00:31:53.310 Katherine Bayless: This is data that… oh, go ahead, sorry.

274 00:31:53.460 ⇒ 00:31:58.009 Kyle Wandel: I was gonna say, the only way to tie this to registration data is the email, then, it looks like.

275 00:31:58.350 ⇒ 00:32:01.070 Kyle Wandel: Or is that ID associated with a badge number?

276 00:32:01.620 ⇒ 00:32:04.730 Katherine Bayless: Yeah, yeah, the registrant ID is the badge number.

277 00:32:04.930 ⇒ 00:32:05.670 Kyle Wandel: Okay, cool.

278 00:32:06.620 ⇒ 00:32:09.930 Katherine Bayless: Conveniently. One rare convenience.

279 00:32:09.930 ⇒ 00:32:15.330 Kyle Wandel: Yeah, I was gonna say, I would think… I know that we use different vendors, but I would think that they should be the same, because you’re getting the badge.

280 00:32:15.330 ⇒ 00:32:16.820 Ashwini Sharma: Is this a badge number?

281 00:32:18.260 ⇒ 00:32:18.750 Kyle Wandel: commitments.

282 00:32:18.750 ⇒ 00:32:21.310 Ashwini Sharma: The first one? Okay, this? Yes. Okay.

283 00:32:21.850 ⇒ 00:32:23.890 Kyle Wandel: So you really only need, like, I mean.

284 00:32:24.870 ⇒ 00:32:31.279 Kyle Wandel: I mean, I guess you want all the data, but, I mean, from a dbt model, you really only need the ID and then the conference name.

285 00:32:33.290 ⇒ 00:32:37.780 Katherine Bayless: Yeah, I would, I would keep the email address only.

286 00:32:37.780 ⇒ 00:32:38.180 Kyle Wandel: Yeah.

287 00:32:39.110 ⇒ 00:32:57.639 Katherine Bayless: just… well, actually, part of me was thinking too, Kyle, like, those registrant IDs, because they are just simple, like, six-digit numbers at the end of the day, like, probably they get recycled, like, we know they get recycled year to year. I doubt they’re the same person year to year, and so we’ll probably wind up needing some kind of, like.

288 00:32:57.820 ⇒ 00:33:11.109 Katherine Bayless: composite key of a sort that’s, like, your reg ID and the CES year, and then the email might just kind of help us with being able to disambiguate, if any questions came up.

289 00:33:11.110 ⇒ 00:33:21.240 Katherine Bayless: Regarding the, like, session code, config code, and full name fields, I did check them. They are all unique, so, like, there aren’t…

290 00:33:21.360 ⇒ 00:33:32.559 Katherine Bayless: repeats in config code, where you would need the full name to figure out which one was which kind of thing. But, it does also seem like the session codes are, like.

291 00:33:33.140 ⇒ 00:33:50.179 Katherine Bayless: sometimes they make sense and are helpful, and other times they’re like RSS101, and you’re like, I don’t know what the hell that could be. So, I would encourage keeping session code, config code, and full name, because I definitely needed all three in different cases where things weren’t super clear based on just the short codes.

292 00:33:53.800 ⇒ 00:33:59.670 Ashwini Sharma: This is only for… okay, this is just a scanner thing, right? And, where did it come? Sorry.

293 00:34:01.920 ⇒ 00:34:03.230 Ashwini Sharma: Engagement…

294 00:34:06.760 ⇒ 00:34:24.800 Katherine Bayless: Yeah, I think the engagement stuff, this is kind of like what Kyle was talking about, where we’ll export the old table. It might take some time to figure out exactly what all of the various engagement options are and where they came from. The scanners would provide some of them, but definitely not all of these.

295 00:34:31.840 ⇒ 00:34:32.880 Katherine Bayless: Actually, Kyle.

296 00:34:33.730 ⇒ 00:34:44.450 Katherine Bayless: I mean, we could take this and research it afterwards, I guess, but is there a chance that engagements might actually have come from, like, staff entering stuff into remembers, like, on the record?

297 00:34:45.650 ⇒ 00:34:52.650 Kyle Wandel: No, I don’t think so, actually. So, I mean, looking at how they used to do it,

298 00:34:53.210 ⇒ 00:35:05.839 Kyle Wandel: they have one big flat… like, one file that it feeds off of. Looking at the Power BI model, you can kind of see a little bit differently. But it’s, like, one big table is where they kind of, like, start, and that table is…

299 00:35:06.470 ⇒ 00:35:11.490 Kyle Wandel: Not calculated, but it is, created… it was… it was created…

300 00:35:11.680 ⇒ 00:35:31.639 Kyle Wandel: ad hoc-ly, I believe, in either Excel or something. I don’t know exactly how they were doing it, but that was, like, the one big table, and then all these, like, the history tables would feed off of that. It was almost like their account management, like, flat file, and then it had a bunch of, like, information about that account management, or that account, basically, member account.

301 00:35:31.640 ⇒ 00:35:39.100 Kyle Wandel: And then you would kind of reference the events history, the scan history, the CES history, and the exhibitor booth history.

302 00:35:39.100 ⇒ 00:35:41.210 Kyle Wandel: To tie it into…

303 00:35:41.720 ⇒ 00:35:53.989 Kyle Wandel: everything. And so, like, even for, like, research downloads, I mean, that’s straight from Shopify, like, that’s also flat file. And I don’t know where they’ve… have that lived in the past. I know…

304 00:35:53.990 ⇒ 00:36:05.129 Kyle Wandel: we have a very robust data set in MRD, but I don’t know how marketing does it, or how they used to do it. So, I think it’s just… it’s just… I think that when we talked about doing this.

305 00:36:05.430 ⇒ 00:36:14.679 Kyle Wandel: asked for this requirement. It was going to be a big one, because it touched everything, and we knew that not everything was clean, quite frankly.

306 00:36:14.790 ⇒ 00:36:27.380 Kyle Wandel: So that’s why it’s been challenging, but I think we definitely need to start getting together the documentation of what each one of these areas’ data sets are, and how do they contribute to this member engagement report.

307 00:36:28.580 ⇒ 00:36:47.510 Katherine Bayless: Yeah, yeah, and I think if we build it out, kind of, as we go, so, like, all the stuff that we’ve figured out, where it comes from, pushing that out to the team, and then adding the pieces as we sort them out, that way they’re not kind of waiting for everything before they get anything. But yeah, it might take a little time to track down some of the loose ends.

308 00:36:55.640 ⇒ 00:37:05.100 Katherine Bayless: And another thing we should add to this report is the Marketing Cloud data, as soon as we have that, so that we could include, like.

309 00:37:05.170 ⇒ 00:37:22.769 Katherine Bayless: something around, like, you know, number of deliverable contacts, number of engaged contacts, right? So they could pull up a company record and say, like, oh, it looks like none of our emails are going through, or, you know, we’ve got 219 contacts on file, but these are the 3 people that actually engage with our emails when we send them out, kind of thing.

310 00:37:24.940 ⇒ 00:37:40.450 Katherine Bayless: Because one of the big pushes this year, well, probably, it’ll be forever work, but, like, one of the big things I’m trying to push for is, like, we do need everybody in the organization to send emails via the marketing team in Marketing Cloud. Unless, of course, they’re, you know, personal correspondents.

311 00:37:40.560 ⇒ 00:37:54.410 Katherine Bayless: And so the more we can, like, surface that data in places like this, and then people will say, like, oh, well, how come I don’t see blah blah blah? And we’ll say, aha, because that team went rogue, and we don’t have the data. So yeah, kind of nudging people in that direction.

312 00:37:55.410 ⇒ 00:37:55.930 Uttam Kumaran: Okay.

313 00:37:57.000 ⇒ 00:37:57.500 Ashwini Sharma: So, yeah.

314 00:37:57.950 ⇒ 00:38:02.299 Ashwini Sharma: The marketing cloud data won’t be available until early February, so…

315 00:38:02.890 ⇒ 00:38:03.580 Katherine Bayless: Okay, okay.

316 00:38:03.580 ⇒ 00:38:06.740 Ashwini Sharma: Yeah, that’s when the connector is getting ready.

317 00:38:06.740 ⇒ 00:38:07.260 Katherine Bayless: Kevin.

318 00:38:08.440 ⇒ 00:38:09.610 Katherine Bayless: Good to know, though.

319 00:38:13.120 ⇒ 00:38:19.589 Uttam Kumaran: Cool. So I feel like… yeah, maybe, can we… should we try to schedule something for Monday for, like, Power BI?

320 00:38:20.270 ⇒ 00:38:22.140 Uttam Kumaran: Figure out session.

321 00:38:23.340 ⇒ 00:38:24.830 Ashwini Sharma: Yeah, I think.

322 00:38:24.950 ⇒ 00:38:27.719 Katherine Bayless: Oh, actually, are we open Monday?

323 00:38:28.050 ⇒ 00:38:29.029 Kyle Wandel: Oh, yeah, we are.

324 00:38:29.030 ⇒ 00:38:32.300 Uttam Kumaran: Oh, yeah, I don’t… MLK? Yeah, I think some… yeah.

325 00:38:32.520 ⇒ 00:38:34.189 Kyle Wandel: We, we, we are open.

326 00:38:34.380 ⇒ 00:38:35.990 Katherine Bayless: Okay, perfect, then yes.

327 00:38:39.150 ⇒ 00:38:39.700 Katherine Bayless: Oops.

328 00:38:39.830 ⇒ 00:38:45.430 Uttam Kumaran: Like, how is, like… Sometime in the morning, like…

329 00:38:46.760 ⇒ 00:38:48.919 Katherine Bayless: Like, 11 East…

330 00:38:50.010 ⇒ 00:38:51.330 Uttam Kumaran: You go up at 11am?

331 00:38:53.080 ⇒ 00:38:55.599 Uttam Kumaran: Okay, maybe we do… noon?

332 00:38:57.000 ⇒ 00:38:58.300 Katherine Bayless: Or…

333 00:38:58.640 ⇒ 00:39:00.880 Uttam Kumaran: Yeah, so maybe let’s plan on that.

334 00:39:03.450 ⇒ 00:39:04.359 Kyle Wandel: That works.

335 00:39:05.260 ⇒ 00:39:05.890 Uttam Kumaran: Okay.

336 00:39:07.290 ⇒ 00:39:10.389 Uttam Kumaran: I have, actually, I have a… I have a 1230.

337 00:39:10.900 ⇒ 00:39:16.759 Uttam Kumaran: Maybe can we… how’s… how’s, like, 1.30?

338 00:39:17.790 ⇒ 00:39:19.870 Katherine Bayless: Yeah, we can do that too, that works.

339 00:39:19.870 ⇒ 00:39:20.470 Uttam Kumaran: Okay.

340 00:39:21.340 ⇒ 00:39:24.340 Uttam Kumaran: Oh, okay. Okay, perfect. I will send that.

341 00:39:28.940 ⇒ 00:39:29.560 Uttam Kumaran: Okay.

342 00:39:30.580 ⇒ 00:39:48.859 Uttam Kumaran: So we talked about Power BI, we talked about the existing report, so I think we’re just gonna kind of drive towards that. I think, Ashwini, we can talk about how we’re gonna stitch historicals, but priority is just this past year, really. So, once we nail that, we can start to weave in

343 00:39:49.480 ⇒ 00:39:51.499 Uttam Kumaran: The, like, backfills?

344 00:39:51.670 ⇒ 00:40:03.300 Uttam Kumaran: I guess I wanted to also just chat about, like, ways of working, like, kind of for the next few months. I mean, we just have this meeting currently booked. I think it would be helpful to at least have, like.

345 00:40:03.460 ⇒ 00:40:04.700 Uttam Kumaran: one, like.

346 00:40:04.840 ⇒ 00:40:13.920 Uttam Kumaran: Typically, what we’ve done is just, like, have an open working session. Seems like we’ve had a lot of things that we’ve just gone back and forth on on Slack, and I think it’d be good to just have

347 00:40:14.030 ⇒ 00:40:15.630 Uttam Kumaran: At least one of those.

348 00:40:16.140 ⇒ 00:40:20.450 Uttam Kumaran: Like, midweek, like, maybe, like, on Tuesdays.

349 00:40:20.680 ⇒ 00:40:34.739 Uttam Kumaran: Typically, like, on Mondays, we, like, kind of regroup and work on a bunch of stuff. I think we’ll have some questions, and then that way, we have this meeting on Friday. It could be this crew and Kai, and then, sort of as needed, people can kind of come in and out.

350 00:40:34.890 ⇒ 00:40:38.609 Uttam Kumaran: Does that sound like a good plan?

351 00:40:39.060 ⇒ 00:40:41.190 Katherine Bayless: Yeah, I think so. Kyle, what do you think?

352 00:40:41.190 ⇒ 00:40:41.870 Uttam Kumaran: Okay.

353 00:40:41.870 ⇒ 00:40:43.120 Kyle Wandel: Okay, that sounds good to me.

354 00:40:44.110 ⇒ 00:40:53.709 Uttam Kumaran: That way, this meeting, we can talk… we could keep more, like, summary, like, high level, and then we can get a lot of, like, pair programming stuff done. And then, Nashweeny, I’ll let you sort of schedule

355 00:40:54.000 ⇒ 00:41:01.050 Uttam Kumaran: sessions as… as, like, you need, so… Sure, yeah. Just want to get more touchpoints and momentum, so…

356 00:41:02.470 ⇒ 00:41:08.880 Katherine Bayless: Yeah, I’m always a big fan of having something that’s, like, already on the calendar, but don’t feel like you gotta wait until Tuesday, right?

357 00:41:09.050 ⇒ 00:41:10.050 Uttam Kumaran: Exactly.

358 00:41:10.500 ⇒ 00:41:24.010 Katherine Bayless: I think… I was gonna ask, too, so I wanted to start, I was gonna send out a, like, a daily stand-up invite, and then a weekly, sort of, or maybe bi-weekly, I don’t know, I haven’t kind of decided yet, planning. I think…

359 00:41:24.010 ⇒ 00:41:40.940 Katherine Bayless: Stand-up, it’s okay if it’s, like, our team internally. I don’t know if you guys need to join a daily stand-up. Oh, you’re more than welcome to, of course. But planning-wise, if you guys are already, sort of, planning on a particular day, it probably makes sense to synchronize the cadence, unless it just is, like, the one spot that won’t work for us, but…

360 00:41:40.940 ⇒ 00:41:41.770 Katherine Bayless: When do you guys…

361 00:41:41.770 ⇒ 00:41:53.589 Uttam Kumaran: Yeah, so we… most of our teams, like, we’re… we’re operating internally, like, kind of like one-week sprints, which is, like, wake up, like, on Monday, what are we aiming? We look… we’ll look multiple weeks ahead.

362 00:41:53.590 ⇒ 00:42:08.319 Uttam Kumaran: But I feel like, after being an engineer for a while, it’s nice to be like, wait, what’s gonna… what are we gonna have by Friday? You know? Versus, like, two weeks is almost, like, too much time. So we can… again, Mondays is usually our day to sort of do a lot of that. If you want to just…

363 00:42:08.940 ⇒ 00:42:12.290 Uttam Kumaran: we could just basically have a planning session on Monday.

364 00:42:12.480 ⇒ 00:42:16.149 Uttam Kumaran: And it’s up to you if you want to do two-week sprints or not, but.

365 00:42:16.680 ⇒ 00:42:17.270 Katherine Bayless: Yeah, one week.

366 00:42:17.270 ⇒ 00:42:19.369 Uttam Kumaran: I also don’t know what the internal, like.

367 00:42:19.540 ⇒ 00:42:24.629 Uttam Kumaran: how heavy you’re thinking about, sort of, agile and the sprint process, so… happy to align.

368 00:42:25.370 ⇒ 00:42:41.690 Katherine Bayless: Yeah, my style is usually, like, 10-15 minute kind of daily stand-up, keep everybody just, you know, in touch and on the same page, and then, like, an hour, or maybe 90 minutes, maybe while we’re getting started, like, weekly planning, because, yeah, I’m the same way, like, 2 weeks is a little too long.

369 00:42:41.690 ⇒ 00:42:47.239 Katherine Bayless: One week, every now and then feels like a little too tight, but it’s better to be too tight sometimes than too long.

370 00:42:47.240 ⇒ 00:43:04.989 Katherine Bayless: most of the time. And then I do want to try and get, like, maybe monthly retros on the calendar, but I think… Okay, great. …in good flow right now, so there’s a lot of, like, feedback and back and forth happening. Like, I feel like retros become more essential when you’re like, no, no, I really need to pull these things out of your brains.

371 00:43:04.990 ⇒ 00:43:05.750 Uttam Kumaran: Yeah.

372 00:43:06.170 ⇒ 00:43:07.359 Katherine Bayless: Yeah, so I think Monday’s true.

373 00:43:07.360 ⇒ 00:43:14.209 Uttam Kumaran: Monthly retro… yeah, monthly retro could be good. I feel like, on our side, that’s sort of our cadence, either, like.

374 00:43:14.320 ⇒ 00:43:26.199 Uttam Kumaran: as a… as a… if a client is, like, just starting, we almost are, like, on a weekly basis, like, how did it go this week? Like, what adjustments? And usually we end up, like, bi-weekly or monthly, like.

375 00:43:26.310 ⇒ 00:43:46.300 Uttam Kumaran: okay, how do we feel comfortable with everything? For me, it’s, like, sort of, like, one, to keep, like, what is going on in the week… immediate week next week in your brain, and then have another session where we’re talking, like, 3 months out. And so, what I’m pushing a lot of our teams to do is to, like, at least try to get a view of 3 months, so that we have a… we just, like.

376 00:43:46.520 ⇒ 00:43:56.950 Uttam Kumaran: see the week in 2 weeks, and then we see it in a broader plan, and then you can kind of zoom in and out. And I think beyond 3 months, we sort of have, like, rough, big rocks that we’re getting to.

377 00:43:57.120 ⇒ 00:44:03.200 Uttam Kumaran: But 3 months is a… like, the quarter or 3 months is, like, a nice chunk to at least try to, like, be like, okay, we have…

378 00:44:03.370 ⇒ 00:44:06.230 Uttam Kumaran: You know, whatever, 12 weeks to work with here, so…

379 00:44:06.360 ⇒ 00:44:22.240 Uttam Kumaran: that’s… that’s something that’s kind of reflected in our… in our Gantt chart as well, is, like, trying to get to that… that point. So totally. So Catherine, if you just want to book that on Mondays and… and add both of us, we’re happy to be there, and then you can certainly add us as optional to the daily stand-ups as well.

380 00:44:22.420 ⇒ 00:44:27.020 Uttam Kumaran: I can… Yeah, we can kind of… Okay, yeah.

381 00:44:27.610 ⇒ 00:44:28.880 Katherine Bayless: Okay. Yeah.

382 00:44:28.880 ⇒ 00:44:29.679 Uttam Kumaran: And then I’ll send…

383 00:44:29.680 ⇒ 00:44:30.210 Katherine Bayless: Later.

384 00:44:30.850 ⇒ 00:44:31.400 Katherine Bayless: Sorry, dude.

385 00:44:31.400 ⇒ 00:44:36.609 Uttam Kumaran: Yeah, and then I’ll send the… the call for… for Monday, just, like.

386 00:44:36.830 ⇒ 00:44:41.270 Uttam Kumaran: Power BI, powwow, I’ll send the call for, like, a Tuesday working session.

387 00:44:41.380 ⇒ 00:44:44.430 Uttam Kumaran: we have these Friday things, but also if you want to…

388 00:44:44.680 ⇒ 00:44:49.639 Uttam Kumaran: own this meeting, or, like, change it up. We can also have this be, like, a weekly retro.

389 00:44:49.880 ⇒ 00:44:50.650 Uttam Kumaran: Bye.

390 00:44:50.650 ⇒ 00:44:51.879 Katherine Bayless: Yeah, that’s actually…

391 00:44:51.880 ⇒ 00:44:53.400 Uttam Kumaran: Farm side, or… yeah.

392 00:44:53.760 ⇒ 00:45:02.020 Katherine Bayless: kind of reading my mind, I was gonna ask, like, I… this is something that’s been on my brain since, like, late fall, like, you know, now that

393 00:45:02.340 ⇒ 00:45:19.289 Katherine Bayless: there’s this, like, AI backbone to every interaction. It probably does make sense for us to, like, own and schedule all the calls, make sure the AI companion is on, and always park the, you know, or outputs of the call in a central location. It just seems like we’re probably passing.

394 00:45:19.290 ⇒ 00:45:21.250 Uttam Kumaran: We are doing that on our side.

395 00:45:21.250 ⇒ 00:45:21.580 Katherine Bayless: Yeah.

396 00:45:21.580 ⇒ 00:45:26.430 Uttam Kumaran: So we are… we’re putting that into a repo on our side that helps us sort of, like.

397 00:45:26.820 ⇒ 00:45:28.949 Uttam Kumaran: Because also, I have, sometimes I’m like.

398 00:45:29.090 ⇒ 00:45:35.840 Uttam Kumaran: Okay, tell me, like, I’m interested in, like, what are some open questions that we’ve talked about, like, multiple times?

399 00:45:36.100 ⇒ 00:45:40.530 Katherine Bayless: That’s something that you want to scan across, like, all of your transcripts for.

400 00:45:41.490 ⇒ 00:45:48.650 Uttam Kumaran: So, I mean, we already do that, because it helps us stay organized and keep moving. So yeah, I’m happy to…

401 00:45:48.660 ⇒ 00:46:05.520 Uttam Kumaran: switch over to yours, and then just, like, even help facilitate that and show you kind of how… what are the ways that we’re using, because we… we basically, it’s not automated right now, but we are working on it, where we have, sort of, Zoom recording. We plug into, like, the Zoom SDK on… on our side, and.

402 00:46:05.520 ⇒ 00:46:08.309 Katherine Bayless: We’re gonna… we kind of pull the transcripts into…

403 00:46:08.310 ⇒ 00:46:10.790 Uttam Kumaran: A central place for each client.

404 00:46:10.900 ⇒ 00:46:23.369 Uttam Kumaran: For us, a lot of our engine uses Cursor, so we’re… we’re gonna be planning on just, like, committing that to a repo, so that you can pull that down and just basically, like, use Cursor to chat, like.

405 00:46:23.680 ⇒ 00:46:27.860 Uttam Kumaran: With, like, okay, help me produce, like, an end-of-week summary.

406 00:46:28.090 ⇒ 00:46:39.249 Uttam Kumaran: Tell me… tell me about, like, given everything we’ve talked about in this engagement, like, what else, like, can we do, or, like, what are… what are things we’re maybe missing, or… or not considering, and, like.

407 00:46:39.420 ⇒ 00:46:46.150 Uttam Kumaran: It’s, like, insane, like, never before possible to, like, be able to do stuff like that. So, like, highly encourage that, yeah.

408 00:46:46.750 ⇒ 00:47:00.970 Katherine Bayless: Yeah, I would love if you could kind of, like, do a little, like, show and tell, because, sure. Yeah, I’ve been sort of ad hoc and formally doing some of this stuff, but I do want to kind of make it an actual, like, this is a pattern, this is a pipeline, this is a thing we do.

409 00:47:00.970 ⇒ 00:47:05.570 Uttam Kumaran: What do you guys use as a companion? Or, like, the recording.

410 00:47:06.370 ⇒ 00:47:26.360 Katherine Bayless: So, to be totally honest, I have not gotten in the habit of using it. Like, I mean, so Jay set me up with the, like, the fancy Zoom license so that I can get into the admin side and, like, tinker with the transcript, like, templates and stuff like that. So I have all the permissions, I just haven’t actually sat down to do it, and so, yeah.

411 00:47:26.360 ⇒ 00:47:30.260 Uttam Kumaran: Yeah, I mean, we could give you guys the… what we wrote, and basically put that on a…

412 00:47:30.560 ⇒ 00:47:31.829 Uttam Kumaran: on a trigger.

413 00:47:31.940 ⇒ 00:47:39.429 Uttam Kumaran: to just pull transcripts into somewhere. I know when we originally talked, also, we talked about, like, hey, a lot of the sales calls are being transcribed.

414 00:47:39.660 ⇒ 00:47:51.530 Uttam Kumaran: you know, again, I think it’s, like, however you want to handle that. For example, you can just say, like, any meeting that I’m in, or, like, or, like, you or any of the team, pull those transcriptions.

415 00:47:51.790 ⇒ 00:48:07.319 Uttam Kumaran: like, you could just basically shove it into Snowflake, or commit it to a repo, because it just… yeah, you could basically do whatever, so that could be something we could do. And yeah, we… I decided not to use Otter, Fireflies, or any of them, because I don’t like the interface.

416 00:48:07.410 ⇒ 00:48:18.849 Uttam Kumaran: I think, like… and the Zoom SDK is, like, hard to use, but it’s, like, it works really great, like, once you get it working. And it’s, like, seems… it’s pretty seamless, so…

417 00:48:19.080 ⇒ 00:48:26.149 Uttam Kumaran: If you guys are on Zoom, it helps prevent, like, get another tool that’s, like, trying to suck up your time and energy.

418 00:48:26.630 ⇒ 00:48:34.479 Katherine Bayless: Yeah, yeah, no, exactly. Like, I’m… given how many tools we already have, I’m all for, like, trying to use the ones we have. But yeah, so there’s.

419 00:48:34.480 ⇒ 00:48:34.860 Uttam Kumaran: Yeah.

420 00:48:34.860 ⇒ 00:48:52.410 Katherine Bayless: in the sales team and the membership team, membership team both for sales calls and their, like, committee and board calls, they want to kind of similarly set up this, like, Zoom transcript pipeline campaign. Oh, okay. Like, we go first, then we can help, you know, deploy that pattern to the other teams as well.

421 00:48:53.030 ⇒ 00:49:10.530 Uttam Kumaran: Okay, great. Yeah, I mean, I’d love to show you how we’re doing it, we’re thinking about this, because, it’s… it’s not only allowing us to just do the basics, which is just, like, keep track of stuff, which I feel like is table stakes, but really it’s the, like, chat over multiple transcripts, like.

422 00:49:10.640 ⇒ 00:49:14.800 Uttam Kumaran: those types… and give that access to anybody on the team to be able to do that, right? So…

423 00:49:14.980 ⇒ 00:49:18.029 Uttam Kumaran: There’s not debate about missing stuff, it’s actually just, like.

424 00:49:18.230 ⇒ 00:49:26.999 Uttam Kumaran: like, more proactive, like, it’s been really cool, so I’m happy to share that next week, like, a version of, like, how we’re doing that in Cursor and, like, what’s possible.

425 00:49:27.670 ⇒ 00:49:29.290 Katherine Bayless: Okay, yeah, yeah, that’d be awesome.

426 00:49:30.240 ⇒ 00:49:30.800 Uttam Kumaran: Cool.

427 00:49:31.750 ⇒ 00:49:40.410 Ashwini Sharma: Yeah, Catherine, I had set up a different meeting with you. I think that may not be needed now. I got the answers in this meeting itself, so I’ll just cancel that.

428 00:49:40.920 ⇒ 00:49:47.930 Katherine Bayless: Yeah, no worries, that works. I kind of figured, but yeah. Whatever time you need, I’m happy to, you know, be around. But if you got what you needed.

429 00:49:47.930 ⇒ 00:49:48.899 Ashwini Sharma: That works too.

430 00:49:51.140 ⇒ 00:49:56.820 Uttam Kumaran: And then how is, like… I guess one question, Catherine, like, how is it going on…

431 00:49:57.380 ⇒ 00:50:01.369 Uttam Kumaran: Box, like, SharePoint, like, and any of that initiative.

432 00:50:02.220 ⇒ 00:50:17.349 Katherine Bayless: Yeah, it’s funny, actually, I was gonna see if I could gently, poke Jay today on that topic, because he, earlier this week, was like, I am not ready to talk about 2026 yet. Which is fair, we’re all tired still. Yeah.

433 00:50:17.350 ⇒ 00:50:42.179 Katherine Bayless: But I would really like to start, like, pushing things that direction, and he’s expressed being on board, so I think once the, you know, brain fog of CES wears off, I’m sure we can resume the conversation, but I’m cautiously optimistic that we could start that, like, sooner than later, because I think probably your brain’s going the same place mine is. It’s like, if we’re going to set up this, you know, fancy-pants Zoom pipeline, it makes sense to park stuff in box at the end.

434 00:50:42.180 ⇒ 00:50:47.179 Katherine Bayless: rather than pushing stuff into SharePoint, only to then need it somewhere else, right?

435 00:50:47.180 ⇒ 00:50:48.759 Katherine Bayless: So yeah.

436 00:50:48.770 ⇒ 00:50:50.300 Katherine Bayless: Yeah, I’ll let you…

437 00:50:50.300 ⇒ 00:51:05.539 Uttam Kumaran: The modality of files that are gonna be in there is definitely gonna be different than, like, just text or structure. Like, you’re gonna have images, diagrams, and so, like, we do a lot of work with this company.

438 00:51:05.580 ⇒ 00:51:11.379 Uttam Kumaran: Contextual. The guy that started this company is, like, was on the…

439 00:51:11.690 ⇒ 00:51:26.029 Uttam Kumaran: basically was on the first paper about RAG, like, and so he started this company that’s, like, basically purely… they’re built on, like, the leading edge retrieval, and sort of, like.

440 00:51:27.100 ⇒ 00:51:35.419 Uttam Kumaran: basically, like, Q&A over documents is, like, their bread and butter, and they’re growing a lot, and their product… I’ve never seen a better product in, sort of, like.

441 00:51:35.660 ⇒ 00:51:39.229 Uttam Kumaran: question over documents. For example, it will literally, like.

442 00:51:39.490 ⇒ 00:51:48.029 Uttam Kumaran: go in and box the, like, diagram that it, like, picked from the PDF, and, like, show you the results, and you can, like, trace it all the way back.

443 00:51:48.150 ⇒ 00:51:54.849 Uttam Kumaran: So, like, in terms of speed to value with, like, hey, like, how can we just, like, get at…

444 00:51:55.290 ⇒ 00:51:58.319 Uttam Kumaran: subset of stuff that’s in SharePoint or Box.

445 00:51:58.540 ⇒ 00:52:03.289 Uttam Kumaran: in front of, like, an AI agent, like, this is, like, a path worth,

446 00:52:03.480 ⇒ 00:52:08.939 Uttam Kumaran: worth considering. The team’s great, the product is, like, is, like, out of this world.

447 00:52:09.050 ⇒ 00:52:12.819 Uttam Kumaran: In terms of, like, state-of-the-art retrieval.

448 00:52:12.930 ⇒ 00:52:17.470 Uttam Kumaran: For, like, multimodal stuff, like, from images to text.

449 00:52:17.580 ⇒ 00:52:20.780 Uttam Kumaran: And of course, I know it’s, like, a lot of contracts, PDFs.

450 00:52:20.950 ⇒ 00:52:23.629 Uttam Kumaran: They also do structured data, but…

451 00:52:24.380 ⇒ 00:52:27.799 Uttam Kumaran: I was just talking to them yesterday, I was like, okay, let me bring them up, so…

452 00:52:28.530 ⇒ 00:52:32.710 Katherine Bayless: Yeah, I’ve actually… it’s funny, I’ve had them in a tab on my phone, and maybe you’ve mentioned them before, I saw them.

453 00:52:32.710 ⇒ 00:52:33.350 Uttam Kumaran: Okay.

454 00:52:33.350 ⇒ 00:52:39.910 Katherine Bayless: This is, like, this is… I’ve had them, like, bookmarked for, like, I want to do this thing, for a while, so yeah.

455 00:52:39.910 ⇒ 00:52:55.290 Katherine Bayless: I think… I mean, yeah, I think we had had this call with the Glean Professional Services guy back in, like, beginning of December, and he was sort of offering, you know, the garden variety, like, we’ll train your staff on making agents, and I was like, I don’t think that’s what we need anymore, like, we’ve got.

456 00:52:55.290 ⇒ 00:52:55.680 Uttam Kumaran: Yeah.

457 00:52:55.960 ⇒ 00:53:18.200 Katherine Bayless: like, we actually… there’s a lot of agents people have created in Glean, but I was like, the muscle piece that’s missing is understanding how much that curation layer, or the curation of the context layer matters. I was like, that’s the fluency we need to build, and I don’t know what I’ll come back with, but to that end, I feel like if we… if we use Box as a way to advance that conversation, right?

458 00:53:18.200 ⇒ 00:53:18.670 Uttam Kumaran: Yeah.

459 00:53:18.670 ⇒ 00:53:26.959 Katherine Bayless: inbox, but we’re also trying to get you to behave differently around files and curating knowledge layers and stuff like that. Like, this is my.

460 00:53:26.960 ⇒ 00:53:42.039 Uttam Kumaran: Well, like, a good example, I was talking to them yesterday, and I was literally talking about, like, what metadata matters to provide the AI to understand. I was like, of course, like, the file name, but I’m like, what else do you get? And they have a really unique way where you can

461 00:53:42.110 ⇒ 00:54:01.460 Uttam Kumaran: you can actually add metadata to files, custom metadata. And so, for example, like, you could basically pre-process docs and say, like, okay, this is, like, this type of document, this is this type of document. So when the AI retrieves it, it’s not… doesn’t have to, like, go in, look at everything, and, like, kind of figure it out. You can basically pre-process.

462 00:54:01.640 ⇒ 00:54:11.920 Uttam Kumaran: Another thing that we’re doing is, like, we use Notion a lot internally, and, our current Notion setup, as this Rini will share with you, is, like, not great, because I just, like.

463 00:54:12.230 ⇒ 00:54:26.720 Uttam Kumaran: I’m… I like having things organized, but I didn’t have, sort of, like, a great model of thinking. I’ve never seen great Notion setups, so I generally, like, it worked, but immediately, just, like, our company grew. So, our operations team is, like, doing a revamp.

464 00:54:26.730 ⇒ 00:54:42.679 Uttam Kumaran: But I was telling them, like, actually what’s really important is that you have, like, a written structure on why we organize it in this way, and in the way that’s organized, like, describing the ontology of our Notion. Because when we hook up cursor MCP to Notion.

465 00:54:42.730 ⇒ 00:54:51.940 Uttam Kumaran: when it looks at Notion, it doesn’t see any of that, it just sees, like, a bunch of databases with stuff, and, like, maybe it confer the name, but there’s no… there’s… Notion doesn’t, like, give it…

466 00:54:52.090 ⇒ 00:54:52.880 Uttam Kumaran: like…

467 00:54:53.020 ⇒ 00:55:12.820 Uttam Kumaran: sort of text-based understanding of the relationships, and I was like, we have to define that in some document so that the AI sees the doc and is like, cool, I know quickly where to go to answer your question. Otherwise, it’s gonna scan, like, it doesn’t have, like, Notion is not very fixed. It’s like, Notion is used by every… any type of company, it’s not just for IT service businesses.

468 00:55:12.870 ⇒ 00:55:15.160 Uttam Kumaran: And so, they’re building that, like.

469 00:55:15.300 ⇒ 00:55:24.979 Uttam Kumaran: sort of, like, ontology of, like, what… how is Notion gonna be set up, so that when people use the Notion MCP, they actually get the results like they want, you know?

470 00:55:25.930 ⇒ 00:55:35.379 Katherine Bayless: Yeah, I think, yeah, ontologies are very much heavy on my brain. Like, how do we, like, infer them where we can, define them where we should, and also.

471 00:55:35.380 ⇒ 00:55:36.260 Uttam Kumaran: Yes.

472 00:55:36.260 ⇒ 00:55:39.270 Katherine Bayless: Not a word that only the data team is using, right?

473 00:55:39.800 ⇒ 00:55:44.180 Katherine Bayless: Everybody’s gonna need this skill. So yeah.

474 00:55:44.760 ⇒ 00:55:47.629 Katherine Bayless: Like, I think… yeah, I mean, you know…

475 00:55:48.160 ⇒ 00:55:55.960 Katherine Bayless: data, you know, for, I think, for a long time, we got to sort of be like, oh, you know, structured data, unstructured data, like, it’s all the same stuff now, like.

476 00:55:55.960 ⇒ 00:55:56.560 Uttam Kumaran: It’s all the same.

477 00:55:56.560 ⇒ 00:56:07.230 Katherine Bayless: Right? It’s like, the layer my team operates is no longer the actual data, it’s that layer of, like, okay, what is it, where is it, how do we use it? Because the robot can handle the rest.

478 00:56:07.380 ⇒ 00:56:12.119 Uttam Kumaran: Yeah, and you’ll see a very similar, like, when we choose a BI tool,

479 00:56:12.300 ⇒ 00:56:17.109 Uttam Kumaran: Like, their ability to… for us to provide column level, table level.

480 00:56:17.200 ⇒ 00:56:30.460 Uttam Kumaran: documentation is, like, really the make or break for it working. Like, yeah, some of the things, like, members, it’s gonna understand, but there’s some stuff you can’t infer. And before, documentation sort of was, like, kind of useless.

481 00:56:30.460 ⇒ 00:56:45.160 Uttam Kumaran: Like, it was, like, good, and it was, like, but it was, like, kind of a luxury thing, like, if you, like, spend a lot of time on docs, it’s like, if you’re a good engineer, you write docs, but rarely do people use it. Now, it’s sort of like the linchpin to even, like, having AI be valuable.

482 00:56:46.970 ⇒ 00:56:52.359 Uttam Kumaran: So, yeah, and so, in the way you… when you go and do, like, a box-type project, it would be important to…

483 00:56:52.550 ⇒ 00:57:12.449 Uttam Kumaran: basically really narrate, like, where things are, and really consider the fact that when we add files, there is some pre-processing that, like, looks in a doc, and then maybe, like, attaches, like, a quick summary metadata, or, like, a category. So that way, when the AI is accessing it, it doesn’t have to access it, go through the whole file to infer it.

484 00:57:12.570 ⇒ 00:57:16.769 Uttam Kumaran: They could just, like, check the metadata quick and leave it or toss it, or bring it, you know?

485 00:57:16.990 ⇒ 00:57:21.629 Uttam Kumaran: But again, it’s, like, not the… the system should do it automatically, it doesn’t, because…

486 00:57:22.070 ⇒ 00:57:24.990 Uttam Kumaran: There wasn’t, sort of, like, programmatic.

487 00:57:25.240 ⇒ 00:57:29.140 Uttam Kumaran: accessing of files wasn’t… is not… wasn’t common, you know, I feel like, so…

488 00:57:29.870 ⇒ 00:57:40.679 Katherine Bayless: Yeah, yeah, yeah, I mean, right, like, we used to just kind of be going off of, you know, how many underscores, parentheses, ones, version 2, final, final, final, final, right? Yeah.

489 00:57:40.680 ⇒ 00:57:42.649 Uttam Kumaran: It’s just regex on the file name.

490 00:57:43.040 ⇒ 00:57:50.259 Uttam Kumaran: It’s all file name-based. Even for… even for data engineering, it’s all file-based, right? Like, when you’re doing file streaming and stuff like that, it’s just file name.

491 00:57:50.550 ⇒ 00:57:55.049 Uttam Kumaran: It’s kind of, yeah, it’s kind of stupid, and very, very fragile.

492 00:57:55.050 ⇒ 00:57:56.540 Katherine Bayless: Right, yeah.

493 00:57:56.540 ⇒ 00:57:57.860 Uttam Kumaran: Very limiting, yeah.

494 00:57:58.690 ⇒ 00:58:20.450 Katherine Bayless: You know, I took all the old documentation from the marketing team, and I put it into a knowledge base in Bedrock, because I just want to kind of play with it and see if I could, like, get some, like, forensic business logic out of it. And, I mean, I haven’t had a chance to circle back and play with it more, but the initial, like, results were pretty decent. I was able to get it to give me, like, pretty solid answers, and I was trying to kind of train it to understand that, like.

495 00:58:20.450 ⇒ 00:58:29.159 Katherine Bayless: The documentation is meaningful insofar as it reflects our business and the way it operates, but, like, please don’t refer to the actual, like.

496 00:58:29.160 ⇒ 00:58:44.689 Katherine Bayless: processes or code as good ideas, just look for the spirit of the exercise. So yeah, I think it’ll be interesting to see if we can get some good, like, I don’t know, legacy documentation sort of settled into a place where it’s useful, but not, like.

497 00:58:45.530 ⇒ 00:58:48.430 Katherine Bayless: Perpetuating certain paradigms going forward.

498 00:58:48.430 ⇒ 00:58:49.040 Uttam Kumaran: Yeah.

499 00:58:49.180 ⇒ 00:58:49.900 Uttam Kumaran: Yeah.

500 00:58:51.200 ⇒ 00:58:51.920 Katherine Bayless: Yeah.

501 00:58:51.920 ⇒ 00:58:52.620 Uttam Kumaran: Cool.

502 00:58:52.860 ⇒ 00:59:03.280 Uttam Kumaran: Okay, so let me send a little bit of summary of this call. I’ll book some of those meetings. I guess I’ll remove this one, Catherine, so you can host it, and then…

503 00:59:03.670 ⇒ 00:59:09.029 Uttam Kumaran: Yeah, I think, like, Monday we’ll just try to hit the ground running on trying to drive towards Friday, I guess.

504 00:59:09.230 ⇒ 00:59:11.730 Uttam Kumaran: I guess you could let us know, Catherine, like, what

505 00:59:11.940 ⇒ 00:59:23.710 Uttam Kumaran: when we want to try to review stuff, I do… we do our best to try to hit whatever deadline we can, but I think, like, it would be great to try to use this new process for this deliverable this year, so…

506 00:59:24.370 ⇒ 00:59:31.779 Katherine Bayless: Yeah, definitely, definitely. Yeah, I’ll try and get the meeting invites out today for everything, and then…

507 00:59:33.040 ⇒ 00:59:39.629 Katherine Bayless: move some of those files around in S3, work with Kyle on some of that, and then, yeah. Yeah, yeah, yeah, yeah.

508 00:59:40.760 ⇒ 00:59:41.370 Uttam Kumaran: Okay.

509 00:59:41.530 ⇒ 00:59:52.690 Katherine Bayless: Oh, the scope of work. If you have the… or if whenever you get a chance to make the couple, like, little edits to it, go ahead and put that in the queue, so that it can start getting reviewed and processed.

510 00:59:53.350 ⇒ 00:59:54.630 Uttam Kumaran: Okay, okay, perfect.

511 00:59:55.800 ⇒ 00:59:56.510 Katherine Bayless: Yeah.

512 00:59:57.050 ⇒ 00:59:58.160 Katherine Bayless: I’m excited.

513 00:59:58.160 ⇒ 00:59:58.840 Uttam Kumaran: Okay.

514 00:59:59.270 ⇒ 00:59:59.890 Uttam Kumaran: Alright.

515 01:00:00.360 ⇒ 01:00:01.149 Katherine Bayless: Thank you guys so much.

516 01:00:01.150 ⇒ 01:00:01.830 Uttam Kumaran: March.

517 01:00:01.830 ⇒ 01:00:02.639 Katherine Bayless: So cute.

518 01:00:04.310 ⇒ 01:00:05.709 Uttam Kumaran: His name is Finn.

519 01:00:06.260 ⇒ 01:00:11.429 Katherine Bayless: Okay, I like this. He just, he always looks very, like, astute and business-y, like…

520 01:00:11.430 ⇒ 01:00:26.219 Uttam Kumaran: He is very business, he’s all business, all serious, all the time. He says, what are you looking at me for? Turn around, get back to typing, let’s see the words per minute, start to get… get… hit the… hit our KPIs again.

521 01:00:27.840 ⇒ 01:00:28.680 Uttam Kumaran: Exactly.

522 01:00:28.830 ⇒ 01:00:35.709 Uttam Kumaran: In about an hour or two, he’ll be like, I’m hungry, lunch, and…

523 01:00:36.120 ⇒ 01:00:45.250 Uttam Kumaran: That’s his job. His job is to break up my meetings with laughter and cuteness. That’s a full-time gig.

524 01:00:47.030 ⇒ 01:00:48.359 Katherine Bayless: Important work, yes.

525 01:00:48.580 ⇒ 01:00:49.320 Uttam Kumaran: Yeah.

526 01:00:49.870 ⇒ 01:00:52.770 Kyle Wandel: How big is, how big are these? Like, 100 pounds? 110?

527 01:00:52.770 ⇒ 01:00:55.589 Uttam Kumaran: He is 125 pounds.

528 01:00:55.590 ⇒ 01:00:57.790 Kyle Wandel: He’s made himself extra small, but…

529 01:00:57.790 ⇒ 01:01:09.020 Uttam Kumaran: That couch is, like, pretty far away. I don’t… this is not showing the depth really well. But he’s a big… yes, he’s a big boy. He’s 6 years old, or… yeah, 6 years old.

530 01:01:09.440 ⇒ 01:01:13.990 Uttam Kumaran: We… He’s very, very kind. He’s a German Shepherd, Yellow Lab, great.

531 01:01:13.990 ⇒ 01:01:14.370 Kyle Wandel: Excuse me.

532 01:01:14.370 ⇒ 01:01:15.030 Uttam Kumaran: X.

533 01:01:15.290 ⇒ 01:01:17.879 Katherine Bayless: That is exactly what it looks like.

534 01:01:17.940 ⇒ 01:01:32.480 Kyle Wandel: Yeah, my wife… my wife… well, growing up, I had a black lab, yellow lab mix, and she was, like, 120, so I definitely get the size. Right now, we also have a Shepard, and he’s, like, 90 pounds, 95 pounds?

535 01:01:33.110 ⇒ 01:01:34.160 Uttam Kumaran: Great.

536 01:01:34.160 ⇒ 01:01:34.990 Kyle Wandel: It was big.

537 01:01:35.580 ⇒ 01:01:41.350 Uttam Kumaran: I can’t really pick him up, and he’s also, like, really… he’s just all dense, so he doesn’t… he’s not, like, fat.

538 01:01:41.500 ⇒ 01:01:45.979 Uttam Kumaran: But also, when he, like, when he pivots, all of his muscles sort of activate.

539 01:01:46.130 ⇒ 01:01:57.400 Uttam Kumaran: And I’m like, you don’t even go to the gym, like, how is this, like, how are you having any muscles? You have, like, a… you don’t even eat, like, you eat the same thing every day. Don’t just eat dog food.

540 01:01:57.400 ⇒ 01:01:58.199 Kyle Wandel: You don’t matter.

541 01:01:58.200 ⇒ 01:01:59.939 Uttam Kumaran: Super brawlick.

542 01:01:59.940 ⇒ 01:02:02.649 Kyle Wandel: Labs are just dense. Like, absolutely just dense.

543 01:02:02.650 ⇒ 01:02:18.319 Uttam Kumaran: Yeah, yeah. Like, my girlfriend doesn’t walk, can’t walk him anymore, because he’s just, like, he just, like, turns, and he’s just like, oh, you just, like, go with him. So I have, like, my hip belt strapped in, and…

544 01:02:18.320 ⇒ 01:02:19.270 Katherine Bayless: Yeah.

545 01:02:19.950 ⇒ 01:02:25.559 Kyle Wandel: It’s always fun when you walk in and, like, people will, like, go on the other side of the street because they’re just a big dog, basically.

546 01:02:25.560 ⇒ 01:02:31.120 Uttam Kumaran: I know, I know, which I’ve, I’m honestly fine with. That’s okay, I’m okay with that, like…

547 01:02:31.340 ⇒ 01:02:35.339 Uttam Kumaran: Yeah. He also gets nervous, too.

548 01:02:35.700 ⇒ 01:02:41.209 Uttam Kumaran: Like, yeah, well, you’re both nervous. You’re both actually lots more similar than you think. Yeah, you know.

549 01:02:41.210 ⇒ 01:02:50.270 Kyle Wandel: Yep, that’s how our shepherd is. Our shepherd’s a big, scaredy cat, and even though he’s 95 pounds, and he’s all muscle, but he just is scared. He’s reactive, so…

550 01:02:50.930 ⇒ 01:02:51.830 Katherine Bayless: Yeah.

551 01:02:53.750 ⇒ 01:02:54.850 Katherine Bayless: Adorable.

552 01:02:56.290 ⇒ 01:03:00.859 Uttam Kumaran: Okay. Well, thank you all, have a happy Friday, happy weekend, hope you both feel better.

553 01:03:00.970 ⇒ 01:03:02.480 Uttam Kumaran: Coming in Monday, so…

554 01:03:02.480 ⇒ 01:03:03.410 Katherine Bayless: Yeah.

555 01:03:03.410 ⇒ 01:03:04.700 Kyle Wandel: Toso.

556 01:03:05.770 ⇒ 01:03:06.500 Uttam Kumaran: Okay.

557 01:03:06.760 ⇒ 01:03:09.070 Uttam Kumaran: Talk to you soon. Bye.