Meeting Title: MarTech Discussion Reschedule Date: 2026-01-29 Meeting participants: Robert Tseng, Zoran Selinger
WEBVTT
1 00:02:03.820 ⇒ 00:02:05.339 Robert Tseng: It’s been a good game, Ben.
2 00:02:05.650 ⇒ 00:02:16.780 Zoran Selinger: Yeah, actually, it wasn’t a great game, but we’ve done a bunch. I expected 30 goals, around 30 goals, but we only got 27.
3 00:02:17.260 ⇒ 00:02:33.130 Zoran Selinger: Yeah, but yeah, it’s, it was always my dream to do those kinds of things with my kids, so now it’s happening, I’m really enjoying it. And I’m happy that he’s also training. Sara is in gymnastics at the moment.
4 00:02:33.700 ⇒ 00:02:38.800 Zoran Selinger: And he’s doing martial arts. Like, he just won recently,
5 00:02:39.960 ⇒ 00:02:45.710 Zoran Selinger: like, nationals, like, he had… he got silver and bronze. Silver.
6 00:02:45.710 ⇒ 00:02:46.380 Robert Tseng: Wow.
7 00:02:46.380 ⇒ 00:02:49.789 Zoran Selinger: in his… In his, age.
8 00:02:50.340 ⇒ 00:02:58.319 Zoran Selinger: And in… and we also put him in with the older, kids, and he got bronze there, so he’s doing well.
9 00:02:58.320 ⇒ 00:02:58.930 Robert Tseng: Yeah.
10 00:02:58.930 ⇒ 00:03:10.060 Zoran Selinger: just needs to… they’re all a little bit more spoiled than we were when we… when we were younger, you know? So it’s… you… you have to push them a little bit more, like.
11 00:03:10.650 ⇒ 00:03:29.670 Zoran Selinger: a little bit more reinforcements than we used to need. We had our heroes, like Bruce Lee, and that was enough for us. We didn’t need more than that. These kids, these days, they’re a little bit more… you have to push them a little bit.
12 00:03:29.670 ⇒ 00:03:32.940 Robert Tseng: Yeah. Oh, that’s kind of strange.
13 00:03:32.940 ⇒ 00:03:41.070 Zoran Selinger: All good. Are you, I think I noticed somewhere, like, you got a little bit of traps going as well.
14 00:03:41.070 ⇒ 00:03:42.969 Robert Tseng: Oh, yeah?
15 00:03:43.380 ⇒ 00:03:48.650 Robert Tseng: I… I probably… I worked out a lot more. I mean, I used to play college tennis,
16 00:03:48.770 ⇒ 00:04:00.450 Robert Tseng: Yeah, so, I mean, I trained my whole childhood, you know, I practiced, like, 3 hours a day until I was, like, 18, and then in college, I kind of switched to…
17 00:04:00.450 ⇒ 00:04:15.150 Robert Tseng: I mean, I actually did enjoy playing at the collegiate level, so I switched to doing other sports. I just tried a bunch of things, so… I just like being active, feel like I need to get out more. Definitely, in the past year, doing this Brainforge thing has…
18 00:04:15.220 ⇒ 00:04:21.319 Robert Tseng: I’ve, just been doing the bare minimum, at least trying to get to the gym every day, that’s about it. Yeah.
19 00:04:21.329 ⇒ 00:04:31.399 Zoran Selinger: No, that’s good. If you’re… if you’re getting in there four times a week, that’s… I mean, three times a week is good, but four times and more is great.
20 00:04:31.420 ⇒ 00:04:32.300 Robert Tseng: Yeah, yeah.
21 00:04:32.380 ⇒ 00:04:41.850 Zoran Selinger: That was kind of my rule, I always wanted to train more days than not, so if I’m doing 4 days, that’s great.
22 00:04:41.850 ⇒ 00:04:42.250 Robert Tseng: Yeah.
23 00:04:42.250 ⇒ 00:04:42.870 Zoran Selinger: Excellent.
24 00:04:44.940 ⇒ 00:04:47.500 Zoran Selinger: Yeah, these days, I just,
25 00:04:48.090 ⇒ 00:04:52.799 Zoran Selinger: Do… do… basically bare minimum… bare minimum.
26 00:04:52.800 ⇒ 00:04:53.250 Robert Tseng: Yeah.
27 00:04:53.250 ⇒ 00:04:59.049 Zoran Selinger: I can’t, like, these meetings in the evening that we now have, I can’t go into the session.
28 00:04:59.410 ⇒ 00:05:01.150 Zoran Selinger: Into the MMA session.
29 00:05:01.260 ⇒ 00:05:02.960 Robert Tseng: Oh, thank you.
30 00:05:02.960 ⇒ 00:05:13.069 Zoran Selinger: I’m trying to get… I’m trying to get a few people to do morning sessions, but it just… we live in a place that is too small for that, we just…
31 00:05:13.190 ⇒ 00:05:15.249 Zoran Selinger: Don’t have enough people to…
32 00:05:15.460 ⇒ 00:05:19.940 Robert Tseng: Yeah, well, if you want to do a day where you just go and do… I mean, yeah, just block off the time. I know these…
33 00:05:19.940 ⇒ 00:05:20.270 Zoran Selinger: Yeah.
34 00:05:20.270 ⇒ 00:05:26.859 Robert Tseng: Kind of get on top of you. So, you just gotta… you gotta protect your time, do what… do what makes you feel…
35 00:05:26.860 ⇒ 00:05:27.220 Zoran Selinger: Yeah.
36 00:05:27.220 ⇒ 00:05:37.410 Robert Tseng: like, you can actually do this sustainably. So, yeah, I mean, I block off meetings on usually Wednesday afternoons, I don’t take meetings, but
37 00:05:38.280 ⇒ 00:05:45.270 Robert Tseng: Yeah, I feel like I have to have at least one day with… one chunk with no meetings to, like, just give myself that space.
38 00:05:45.270 ⇒ 00:05:50.830 Zoran Selinger: You split the week that way, and it refreshes you, that one day refreshes you, really.
39 00:05:50.990 ⇒ 00:05:51.620 Robert Tseng: Yeah.
40 00:05:51.920 ⇒ 00:05:52.440 Zoran Selinger: Yeah.
41 00:05:52.780 ⇒ 00:05:53.610 Zoran Selinger: Yeah. Okay.
42 00:05:53.610 ⇒ 00:06:08.629 Robert Tseng: Well, with this call, I know we got rescheduled a bunch, I think I was definitely a lot more excited about it, like, end of last year. It’s been a while since I’ve come to this doc, so… I mean, I don’t even know if you’ve finished reading it, but yeah, I was hoping we could just.
43 00:06:08.630 ⇒ 00:06:10.589 Zoran Selinger: Yes, I did go through it.
44 00:06:10.990 ⇒ 00:06:12.500 Robert Tseng: Okay. Well, I think…
45 00:06:12.500 ⇒ 00:06:20.480 Zoran Selinger: Honestly, I was surprised a little bit, because I don’t hear those conversations, maybe in, like.
46 00:06:20.610 ⇒ 00:06:24.740 Zoran Selinger: In the clients that we, that we talk about, like.
47 00:06:24.740 ⇒ 00:06:25.380 Robert Tseng: Yeah.
48 00:06:25.640 ⇒ 00:06:29.550 Zoran Selinger: We’re not directly mentioning, kind of.
49 00:06:29.760 ⇒ 00:06:36.369 Zoran Selinger: AI that much, and how you’re gonna apply it in, like, those three different categories, right?
50 00:06:36.370 ⇒ 00:06:36.710 Robert Tseng: Yeah.
51 00:06:38.810 ⇒ 00:06:39.420 Zoran Selinger: So, I…
52 00:06:39.420 ⇒ 00:06:42.369 Robert Tseng: But I would like to more. So, like, I think,
53 00:06:43.310 ⇒ 00:07:00.680 Robert Tseng: Yeah, I feel like we… you know, the whole point of doing this Brain Forge thing is to kind of, try to reimagine everything being very AI-forward, or kind of pushing the envelope. I think we’re… we get to go and… and… and think
54 00:07:00.680 ⇒ 00:07:15.400 Robert Tseng: and see how we can apply the latest tech technology. So, yeah, I mean, you’re right, I think for the most part, it’s not really been… it’s usually not what people are asking for, because they don’t really know what it can do. So we… so we have to kind of be…
55 00:07:15.400 ⇒ 00:07:19.559 Zoran Selinger: More or less the subject matter expert to let them know what it can do.
56 00:07:20.290 ⇒ 00:07:29.259 Robert Tseng: So for this call, I was hoping we could maybe… I mean, I don’t even know how to download this PDF, so I think, trying to…
57 00:07:30.260 ⇒ 00:07:33.950 Robert Tseng: This is maybe not the best way to view it, but .
58 00:07:34.890 ⇒ 00:07:37.149 Zoran Selinger: I was able to download it somehow.
59 00:07:37.570 ⇒ 00:07:38.760 Robert Tseng: You are?
60 00:07:39.230 ⇒ 00:07:41.579 Robert Tseng: Okay. Well, I was hoping I would just kind of click.
61 00:07:41.580 ⇒ 00:07:43.209 Zoran Selinger: But yeah, but this is fine.
62 00:07:43.410 ⇒ 00:07:54.739 Robert Tseng: And, yeah, when we get to certain topics, we’ll just kind of pause there and kind of have a conversation about it. Then I want, basically, Luke to go and
63 00:07:54.850 ⇒ 00:08:00.269 Robert Tseng: Crawl this transcript, and… Pick out pieces that we want to…
64 00:08:00.530 ⇒ 00:08:17.639 Robert Tseng: kind of either do a deeper dive into, or maybe we’ve… we… we just… we had an interesting discussion that can be… can be repurposed into some pitch, proposal, or… or something like that. So, I think, yeah, I didn’t… I didn’t… this was a lot more fresh on my mind, maybe…
65 00:08:17.870 ⇒ 00:08:20.659 Robert Tseng: last month, so I’ll probably just kinda…
66 00:08:21.080 ⇒ 00:08:29.860 Robert Tseng: We’ll just… we’ll just kind of click through and see how much we can get… get done in, like, 15-20 minutes, and after that, we’ll… we can call it. Maybe we’ll do another session.
67 00:08:30.180 ⇒ 00:08:31.190 Robert Tseng: So…
68 00:08:31.670 ⇒ 00:08:41.959 Robert Tseng: Cool, so I think, like, yeah, this idea of, like, AI agents by domain, I thought, was a good way of breaking it down. So, agents for marketers, customers, and for cus… and…
69 00:08:42.090 ⇒ 00:08:47.519 Robert Tseng: well, I guess this is… this is pretty much the… I think that’s probably a typo, similar things, but the idea is that…
70 00:08:47.520 ⇒ 00:08:49.079 Zoran Selinger: No, it’s not. No, it’s not.
71 00:08:49.200 ⇒ 00:08:50.409 Zoran Selinger: No, no, no. Or agents.
72 00:08:50.410 ⇒ 00:08:53.360 Robert Tseng: Of customers. Oh, got it, got it. Four and four and above.
73 00:08:53.860 ⇒ 00:09:11.689 Zoran Selinger: So, yeah, that’s, like, generative search stuff, and… and every… like, that’s things that are on their side, and we don’t really have control over it much, right? Yeah. So we have to think about that. For example, generative search is a good example of agents, of customers.
74 00:09:12.900 ⇒ 00:09:15.880 Robert Tseng: Yeah, I think it’d be interesting to…
75 00:09:16.100 ⇒ 00:09:26.529 Robert Tseng: better understand how customers are using agents and how that impacts the buyer journey. Like, I feel like, you know, basic things like
76 00:09:26.750 ⇒ 00:09:43.660 Robert Tseng: well, now there… or maybe it’s not so basic, but now, you know, customers can… can shop out of LLMs, and so there’s less reliance on search, so search engine optimization has definitely decreased. I spoke to, kind of, The Knot, which is a…
77 00:09:43.660 ⇒ 00:09:47.550 Robert Tseng: It’s like a wedding website company,
78 00:09:47.600 ⇒ 00:09:54.870 Robert Tseng: Probably the biggest one in the U.S, and over the past 2 years, their search volume has dropped by 40%.
79 00:09:56.350 ⇒ 00:10:15.909 Robert Tseng: And so, I mean, I think it’s a lot of different things, maybe other competitors in their space, but they’re saying a lot of that is because of AI, and people relying less on their blogs and the content they create to do wedding planning. Like, they have, like, a lot of good thought leadership around
80 00:10:15.940 ⇒ 00:10:32.720 Robert Tseng: you know, just a lot of… it’s a very niche topic about weddings, but now people just kind of chat with ChatGPT, and then it redirects them, or they’re satisfied with the answer, and don’t even really need to go into their content anymore. Yeah. So, so that’s an interesting, kind of insight from… from the customer perspective.
81 00:10:32.880 ⇒ 00:10:40.530 Robert Tseng: And then… yeah, I think, like, agents that interact with prospects and customers, like, I feel like this is probably…
82 00:10:40.710 ⇒ 00:10:45.049 Robert Tseng: This is, to me, is the most risky, because
83 00:10:45.250 ⇒ 00:10:47.870 Robert Tseng: Yeah, I mean, we… we can do…
84 00:10:48.170 ⇒ 00:10:57.349 Robert Tseng: Yeah, it just requires a lot of maintenance to make sure that the agent is representing our clients well, if we were to build something like this for them. I think about, like.
85 00:10:57.440 ⇒ 00:11:12.780 Robert Tseng: ABC and our approach there. They wanted us to build, like, a customer-facing chatbot. I don’t know if you’ve… you’ve looked at our Andy product that we’ve built for them, but actually what we built was more of an internal-facing agent.
86 00:11:12.780 ⇒ 00:11:18.669 Robert Tseng: It’s a co-pilot for their customer success reps. It’s not something that their end customer chats with. Yeah.
87 00:11:18.670 ⇒ 00:11:25.769 Zoran Selinger: And why did that happen? Why did you… did you switch intentionally, or it just ended up serving that purpose better?
88 00:11:25.950 ⇒ 00:11:34.960 Robert Tseng: Well, we decided it was too… it was too risky to build the end customer-facing agent, agent, chatbot, so we felt that…
89 00:11:34.960 ⇒ 00:11:49.529 Zoran Selinger: word, risky, because there is one concept that I’ve never heard in these few months that I’ve been with you. When you say risky, you’re talking about being represented properly in front of the customers.
90 00:11:49.690 ⇒ 00:11:55.130 Zoran Selinger: I’m really interested, because I was watching a talk on DEF CON.
91 00:11:55.130 ⇒ 00:11:55.560 Robert Tseng: Okay.
92 00:11:55.560 ⇒ 00:11:57.849 Zoran Selinger: exploits, in chatbots.
93 00:11:57.850 ⇒ 00:11:58.820 Robert Tseng: Oh, yeah.
94 00:11:58.820 ⇒ 00:12:04.019 Zoran Selinger: super scary, and I don’t hear us talking about that at all.
95 00:12:04.280 ⇒ 00:12:04.910 Robert Tseng: Yeah.
96 00:12:05.620 ⇒ 00:12:14.249 Zoran Selinger: So I… that’s also, like… thing that I think about when we’re talking about agents for customers.
97 00:12:14.410 ⇒ 00:12:29.210 Robert Tseng: Yeah, so, I mean, by default, whether we’re working with marketers, I think we… we default to making… building agents for our… for our clients, like op… like operators, because we still keep a human in the loop,
98 00:12:29.700 ⇒ 00:12:38.910 Robert Tseng: Like, yeah, we… we make it so the decision is still, like, on the… on the client side, client’s team in order to… to push something through.
99 00:12:39.280 ⇒ 00:12:49.049 Robert Tseng: And, you know, we felt that that was a safer approach. You know, given that, you know, we build… we build for speed, and…
100 00:12:49.380 ⇒ 00:12:52.029 Robert Tseng: Yeah, I just think that, you know.
101 00:12:52.220 ⇒ 00:13:04.930 Robert Tseng: I have, I have other… I have other friends who are… who are, at bigger consultancies working on customer-facing chatbots, and yeah, just, like, the amount of additional effort to… to go and
102 00:13:04.940 ⇒ 00:13:14.719 Robert Tseng: always have somebody who’s monitoring, kind of, the logs, making sure there’s no exploits, like, it’s just… it’s just a lot of work for not so much more value, in my opinion.
103 00:13:16.290 ⇒ 00:13:17.090 Robert Tseng: Yeah.
104 00:13:17.200 ⇒ 00:13:33.339 Robert Tseng: So, yeah, because customers don’t directly interact with these agents, we get to be more integrated with our clients’ workflows, and yeah, I think we get to assist in this whole kind of conceptualizing, like, what
105 00:13:34.070 ⇒ 00:13:42.439 Robert Tseng: We’re reinventing the workflow with them, and then we’re also analyzing their work, helping… and that’s, you know, the analysis part has always been our strong suit.
106 00:13:42.440 ⇒ 00:14:05.840 Robert Tseng: being able to take what already exists and assess the performance. I think where our team is really being pushed to do is to conceptualize, okay, if we give them new capabilities, what should our clients be able to do with that? And you’re kind of getting a taste of that on, especially on Eden, right? As you’re kind of sharing, like, how some of the new data models we’ve built that they’ve never seen before, how do they actually
107 00:14:05.840 ⇒ 00:14:09.190 Robert Tseng: Use it in their… in their lifecycle marketing workflows.
108 00:14:11.840 ⇒ 00:14:22.470 Zoran Selinger: Yeah, so, yeah, yeah. I don’t know how, how that exactly looks like, for, like, these agents for, for, marketers. Let’s…
109 00:14:22.780 ⇒ 00:14:25.119 Zoran Selinger: I’m thinking about, okay, let…
110 00:14:26.300 ⇒ 00:14:34.319 Zoran Selinger: we have an agent just for Eden that looks at our warehouse and knows the… understands the schema.
111 00:14:34.630 ⇒ 00:14:35.190 Robert Tseng: Yeah.
112 00:14:35.670 ⇒ 00:14:43.109 Zoran Selinger: And we just, yeah, asking questions, based on it understanding what’s in BigQuery.
113 00:14:43.430 ⇒ 00:14:44.190 Robert Tseng: Yeah.
114 00:14:45.110 ⇒ 00:14:55.309 Robert Tseng: So we’re doing that on two fronts. Like, one is, I think, obviously, we have our data platform documentation, which is meant to serve as, like, the guardrails, or, like, kind of the…
115 00:14:55.310 ⇒ 00:15:08.569 Robert Tseng: the, like, the outline for an LLM to follow, because it has the standardized list of definitions. It needs to be updated. That format is kind of outdated, and we haven’t been doing a good job of keeping that updated. So, I’m kind of trying to think through how do we…
116 00:15:08.630 ⇒ 00:15:12.400 Robert Tseng: how do we… how do we… how do we do that? And then, like.
117 00:15:12.540 ⇒ 00:15:20.570 Robert Tseng: But that’s, like, just from, like, a… that’s for human consumption, or human use, to just understand, to chat with, like, how do we really, like, yeah, like…
118 00:15:20.570 ⇒ 00:15:44.570 Robert Tseng: what data is available to us? Like, that’s really, like, the main use case to solve for that. And then the second part is, from a tooling perspective, like, right now, our tools don’t really tap into BigQuery very, very well. There’s no tool that we have set up for Eden that has, like, a built-in MCP, which is why we’re moving them off Tableau into Omni, because Omni should be able to do that.
119 00:15:44.570 ⇒ 00:15:53.309 Robert Tseng: So, yeah, then the question there is, like, okay, can we build reports, you know, out of, out of Omni just by using, like, kind of.
120 00:15:53.310 ⇒ 00:16:10.279 Robert Tseng: natural language, and I think that should reduce the number of ad hoc requests we get, and so we got approval to go and pursue that migration, you know, in the next month before, you know, the Tableau contract is up. So I think that’s… those are two ways how we’re trying to introduce,
121 00:16:10.510 ⇒ 00:16:16.979 Robert Tseng: it’s not necessarily agents for just marketers, but agents for, like, just, the people at Eden overall.
122 00:16:17.390 ⇒ 00:16:31.949 Zoran Selinger: Yeah, I’m really excited about Omni. Unfortunately, we didn’t get to certify, and we really wanted to do that, and I started learning, but just… we all got stuck.
123 00:16:32.400 ⇒ 00:16:33.640 Zoran Selinger: So…
124 00:16:34.110 ⇒ 00:16:52.950 Zoran Selinger: it looks like such a great tool. It’s gonna help me as well, that you have these 3 different ways to work with it, like, it’s an amazing, amazing tool. I love it and can’t wait to use it. I think I’m gonna be… I’m gonna be much faster with it.
125 00:16:53.510 ⇒ 00:16:57.400 Robert Tseng: Great. Well, you’ll definitely be involved in this migration, so…
126 00:16:57.400 ⇒ 00:16:58.170 Zoran Selinger: Yeah, true.
127 00:16:58.170 ⇒ 00:16:59.180 Robert Tseng: You’re learning well, yeah.
128 00:16:59.180 ⇒ 00:16:59.860 Zoran Selinger: from.
129 00:17:00.330 ⇒ 00:17:01.060 Robert Tseng: Okay.
130 00:17:01.350 ⇒ 00:17:15.169 Robert Tseng: Yeah, I think this was interesting to me. Content… producing content was not surprising to me. Yeah, customer service, so yeah, we definitely hit these top 3 use cases. This customer journey builder agent is interesting. I feel like…
131 00:17:15.180 ⇒ 00:17:24.249 Robert Tseng: that’s not something we’ve done, before, and seems like, you know, I’d be interested in actually seeing the ROI of, like.
132 00:17:24.280 ⇒ 00:17:26.349 Robert Tseng: How valuable that is.
133 00:17:28.800 ⇒ 00:17:37.519 Zoran Selinger: I haven’t seen anything on that front either. I’m thinking about where are we assisting our clients?
134 00:17:38.330 ⇒ 00:17:42.300 Zoran Selinger: So, they’re producing agents…
135 00:17:42.440 ⇒ 00:17:48.249 Zoran Selinger: like, content production, they do that themselves, right? They all know how to do how to do that.
136 00:17:48.580 ⇒ 00:17:57.290 Zoran Selinger: Yeah, so AI-optimized content… I mean, we are not really… we don’t do much SEO, right? With…
137 00:17:57.560 ⇒ 00:18:02.949 Zoran Selinger: I mean, in this case, with Qatar and Tigran specifically, we do have that.
138 00:18:04.320 ⇒ 00:18:08.580 Zoran Selinger: that muscle in the company now, if they’re gonna be here. Yeah.
139 00:18:10.530 ⇒ 00:18:15.310 Zoran Selinger: But then chatbots and everything else, yeah, that’s what… that’s what we do, yeah.
140 00:18:16.350 ⇒ 00:18:17.660 Zoran Selinger: What’s perfect searching.
141 00:18:17.860 ⇒ 00:18:18.970 Zoran Selinger: Yeah, okay.
142 00:18:18.970 ⇒ 00:18:22.469 Robert Tseng: I’m curious if there’s anything on here that we do that is not…
143 00:18:23.230 ⇒ 00:18:29.930 Robert Tseng: Sorry. Sorry if it’s, like, not that… I can’t really zoom in much more. Let’s see…
144 00:18:36.430 ⇒ 00:18:41.539 Robert Tseng: Oh, that’s interesting. An agent’s file, basically, on your website.
145 00:18:42.600 ⇒ 00:18:49.040 Robert Tseng: instructions for what LLMs that are crawling your site should do, the help of AEO, I guess.
146 00:18:49.210 ⇒ 00:18:54.249 Robert Tseng: In-product agent or co-pilot for tutorials.
147 00:18:54.670 ⇒ 00:19:02.930 Robert Tseng: Yeah, we haven’t really worked with any complicated SaaS products recently, so that’s not something we’ve had to do. That’s interesting, but we don’t do that.
148 00:19:03.330 ⇒ 00:19:10.179 Robert Tseng: Yeah, the BDR stuff is… they’re machine-readable feeds for product and pricing data.
149 00:19:14.660 ⇒ 00:19:19.270 Robert Tseng: of customers. So, this is, like, for customers. That’s interesting.
150 00:19:21.750 ⇒ 00:19:28.220 Robert Tseng: Yeah, I mean, I feel like we should… we should consider doing… doing that. Yeah, MCPs we do.
151 00:19:28.460 ⇒ 00:19:37.140 Robert Tseng: Yeah, we’ve done, like, a good amount of these, probably at least half of them. Live guidance for reps, this is exactly what Andy for,
152 00:19:37.480 ⇒ 00:19:40.089 Robert Tseng: or, ABC is.
153 00:19:40.640 ⇒ 00:19:50.249 Robert Tseng: Data enrichment, yes, that’s actually what we do. Yeah, and I guess the last thing, audience discovery slash segmentation. So.
154 00:19:50.370 ⇒ 00:19:58.730 Robert Tseng: Yeah, I mean, for the thing that you’re doing and supporting Judd and their team, I’m curious, like, how we… how you will kind of develop a perspective of how we can
155 00:19:58.960 ⇒ 00:20:05.360 Robert Tseng: You know, built… use… To turn that into an agentic workflow.
156 00:20:05.930 ⇒ 00:20:07.020 Zoran Selinger: Yeah, yeah.
157 00:20:08.420 ⇒ 00:20:16.339 Robert Tseng: Yeah, that… kind of the process right now, like, kind of, how are you discovering which segments to give… to give them?
158 00:20:17.940 ⇒ 00:20:21.899 Zoran Selinger: So here I was, so here I was simply,
159 00:20:21.930 ⇒ 00:20:29.580 Zoran Selinger: solving a specific question, getting an answer for a specific question. And that is, what does churn look like
160 00:20:29.580 ⇒ 00:20:42.769 Zoran Selinger: And how reactivations look like. And you might remember that I initially started doing modeling by transaction, because that’s what Jad was using.
161 00:20:43.020 ⇒ 00:20:47.640 Zoran Selinger: And then we figured out that this is not sufficient.
162 00:20:48.330 ⇒ 00:20:54.990 Zoran Selinger: To exactly define the status of every customer, and then we got to, you know, a treatment.
163 00:20:55.240 ⇒ 00:20:58.719 Zoran Selinger: I think you might have been the one to… to tell me. Yeah, you, you.
164 00:20:58.720 ⇒ 00:20:59.160 Robert Tseng: Yeah.
165 00:20:59.160 ⇒ 00:21:03.559 Zoran Selinger: They were the one to tell me that. And that was exactly it, right?
166 00:21:03.640 ⇒ 00:21:09.200 Robert Tseng: Yeah. When we look at the treatment level, we essentially see everything. We see.
167 00:21:09.200 ⇒ 00:21:29.049 Zoran Selinger: how… exactly how they interacted with us, with treatments, whether, like, they had updates to our treatment, how long it was pending, then canceled, completed, active, whatever. We see everything, we see the whole timeline for each customer of exactly what they were doing.
168 00:21:29.570 ⇒ 00:21:30.160 Robert Tseng: Yeah.
169 00:21:31.000 ⇒ 00:21:46.289 Zoran Selinger: So, yeah, that, that, that was, that was basically it. So, that, that grain of, that entity of treatment is exactly the way we should look into, we should look at, for this particular, client.
170 00:21:47.880 ⇒ 00:21:54.150 Zoran Selinger: And that was it. I mean, when, when I was, when I was looking, when I was looking at,
171 00:21:54.270 ⇒ 00:21:55.530 Zoran Selinger: and churn.
172 00:21:55.910 ⇒ 00:22:06.760 Zoran Selinger: And kind of looking at the volumes, simply the model, came out with… it… it simply gave me 4 different… 4 different…
173 00:22:06.830 ⇒ 00:22:22.790 Zoran Selinger: segments. And those are the four… I mean, the three of them were, like, logical, active, inactive, and recently inactive, because we are ignoring the first 60 days of inactivity, as we’re still not considering them churned.
174 00:22:23.230 ⇒ 00:22:28.960 Zoran Selinger: And then… It simply gave me the idea of, okay, these were…
175 00:22:29.110 ⇒ 00:22:33.750 Zoran Selinger: Customers that never had an active treatment.
176 00:22:33.950 ⇒ 00:22:39.009 Zoran Selinger: Which is very, very interesting. So it was purely from the agent.
177 00:22:40.580 ⇒ 00:22:41.230 Robert Tseng: Wow.
178 00:22:41.480 ⇒ 00:22:43.920 Robert Tseng: Yeah. I guess this was an interesting slide.
179 00:22:44.210 ⇒ 00:22:47.150 Robert Tseng: breaking it out by B2B versus B2C.
180 00:22:47.410 ⇒ 00:22:54.730 Robert Tseng: We can see that, like, what you were just describing, definitely, I mean, I would expect all of this to be higher on the B2C side.
181 00:22:55.030 ⇒ 00:23:05.040 Robert Tseng: Except for on prospecting, it’s interesting, there’s a few, a few similarities. Customer Journey Builder Agent seems like it’s more widely adopted in B2C than it is B2B.
182 00:23:07.450 ⇒ 00:23:23.230 Robert Tseng: So, yeah, I mean, I’d be curious to kind of see how we really, better deploy this. So, when you were saying that you discovered this using the agents, like, how… what exact workflow did you use, or kind of, like, what were we…
183 00:23:23.650 ⇒ 00:23:26.349 Zoran Selinger: So, so you gave me an idea of…
184 00:23:26.460 ⇒ 00:23:34.220 Zoran Selinger: You gave me an idea of treatments, so I was, I found…
185 00:23:34.660 ⇒ 00:23:39.540 Zoran Selinger: a table, table with treatments. I gave it, so…
186 00:23:40.630 ⇒ 00:23:43.319 Zoran Selinger: I use… I’m using, like,
187 00:23:44.000 ⇒ 00:23:48.720 Zoran Selinger: like, Jupyter notebooks, okay? So, I connected to the…
188 00:23:49.840 ⇒ 00:23:53.820 Zoran Selinger: To the… to BigQuery, I…
189 00:23:54.220 ⇒ 00:23:59.980 Zoran Selinger: Asked… asked to, the agent to pull the schema.
190 00:24:00.160 ⇒ 00:24:02.850 Zoran Selinger: And I gave… gave the output of… of…
191 00:24:03.070 ⇒ 00:24:07.400 Zoran Selinger: of running, looking at the schema, and I explained what I needed.
192 00:24:08.390 ⇒ 00:24:09.090 Robert Tseng: Nice.
193 00:24:09.330 ⇒ 00:24:11.580 Zoran Selinger: And I explained what I needed, and it…
194 00:24:11.840 ⇒ 00:24:22.380 Zoran Selinger: and it fired off. We were… we needed to go back and forth quite a bit, because it didn’t understand, that,
195 00:24:23.640 ⇒ 00:24:30.539 Zoran Selinger: Like, for example, I didn’t initially know that we have a status of completed, right? I only saw.
196 00:24:30.540 ⇒ 00:24:30.900 Robert Tseng: Yeah.
197 00:24:31.930 ⇒ 00:24:40.179 Zoran Selinger: No, we needed to go back and forth, find, because at every step, I was essentially looking,
198 00:24:40.310 ⇒ 00:24:45.189 Zoran Selinger: It gave me samples of 20 people in each… in each category, right?
199 00:24:45.230 ⇒ 00:25:05.169 Zoran Selinger: And I would go in and look those up manually to see if they really satisfy the criteria that I said, that I wanted. And then there was probably only from 7 to 8 iteration, we got to the correct logic for those 4.
200 00:25:05.200 ⇒ 00:25:06.510 Zoran Selinger: Right? .
201 00:25:06.700 ⇒ 00:25:07.490 Robert Tseng: Yeah.
202 00:25:07.730 ⇒ 00:25:14.500 Zoran Selinger: Because he didn’t understand, like, all the… all the different statuses that there were. I should have probably,
203 00:25:15.370 ⇒ 00:25:19.259 Zoran Selinger: Also drill down to specific fields.
204 00:25:19.390 ⇒ 00:25:29.300 Zoran Selinger: And just see all the distinct values from specific fields, gave it exactly that, and then we would probably be faster with the logic.
205 00:25:29.530 ⇒ 00:25:32.060 Robert Tseng: Do you still have that, chat history?
206 00:25:32.060 ⇒ 00:25:33.360 Zoran Selinger: I have everything, yes.
207 00:25:33.360 ⇒ 00:25:38.230 Robert Tseng: Yeah, if you could send it to me, that might be helpful as I’m kind of rethinking the documentation piece.
208 00:25:38.230 ⇒ 00:25:39.480 Zoran Selinger: Sure, sure.
209 00:25:39.480 ⇒ 00:25:40.130 Robert Tseng: Yeah.
210 00:25:40.760 ⇒ 00:25:46.300 Zoran Selinger: I don’t know, do you have… I don’t know what you have on your side, because I’m on your account.
211 00:25:46.700 ⇒ 00:25:54.049 Zoran Selinger: For cursor. So I don’t know if you have access to all the, all the chats?
212 00:25:54.330 ⇒ 00:25:56.190 Zoran Selinger: On your side, or no?
213 00:25:56.370 ⇒ 00:26:00.300 Robert Tseng: Oh, I didn’t know you used… I didn’t know you used my cursor account.
214 00:26:01.300 ⇒ 00:26:07.820 Zoran Selinger: I mean, just the company, yeah, I got the… I got the account recently, maybe 2 weeks ago.
215 00:26:08.200 ⇒ 00:26:14.760 Robert Tseng: Oh, but it’s, like, on your account. Yeah, no, I… I guess I don’t… I don’t get to see what, you know, how you… how you’re prompting it, so…
216 00:26:14.920 ⇒ 00:26:20.589 Zoran Selinger: Okay, okay. I’ll send it to you. Is there an option to export everything, or…
217 00:26:20.590 ⇒ 00:26:21.850 Robert Tseng: Yeah, there should be.
218 00:26:21.850 ⇒ 00:26:22.530 Zoran Selinger: Okay.
219 00:26:23.370 ⇒ 00:26:23.990 Robert Tseng: Yeah.
220 00:26:24.210 ⇒ 00:26:26.200 Zoran Selinger: I’ll get that to you, I’ll get that to you.
221 00:26:26.200 ⇒ 00:26:32.879 Robert Tseng: Okay, cool. Actually, Mitesh and Josh are pinging me to kind of come on this call. They were going to walk through some
222 00:26:33.260 ⇒ 00:26:40.540 Robert Tseng: some data, like, metrics that they did. I mean, if you want, I can bring you into the call.
223 00:26:42.290 ⇒ 00:26:43.110 Zoran Selinger: There’s this AP.
224 00:26:43.110 ⇒ 00:26:43.470 Robert Tseng: seat.
225 00:26:43.470 ⇒ 00:26:44.950 Zoran Selinger: call as well.
226 00:26:44.950 ⇒ 00:26:49.220 Robert Tseng: Okay, you can… up to you, whichever one you want to join, but let me just,
227 00:26:49.580 ⇒ 00:26:54.539 Robert Tseng: I’ll create the meeting with them, and then I’ll invite you, so if you want to join.
228 00:26:54.960 ⇒ 00:26:56.019 Zoran Selinger: Okay, okay, thanks.
229 00:26:56.020 ⇒ 00:26:57.779 Robert Tseng: Okay, cool. Alright, talk to you later.
230 00:26:58.160 ⇒ 00:26:58.700 Zoran Selinger: Bye.