Meeting Title: PoolPartsAI Regroup Date: 2025-07-09 Meeting participants: bencohen, Miguel de Veyra, Uttam Kumaran
WEBVTT
1 00:01:17.080 ⇒ 00:01:18.570 Miguel de Veyra: Hey! Beth! Morning.
2 00:01:19.390 ⇒ 00:01:20.419 bencohen: Hey! How are you?
3 00:01:20.830 ⇒ 00:01:24.400 Miguel de Veyra: Yeah. Doing? Okay? Let me just ping Uta.
4 00:01:24.800 ⇒ 00:01:25.410 bencohen: Okay.
5 00:03:01.760 ⇒ 00:03:02.529 Uttam Kumaran: Hey! Ben!
6 00:03:03.570 ⇒ 00:03:04.140 bencohen: What’s up, man?
7 00:03:04.140 ⇒ 00:03:04.950 Uttam Kumaran: Hey?
8 00:03:05.350 ⇒ 00:03:09.289 Uttam Kumaran: Sorry, my Zoom just decided to like update right? As I joined.
9 00:03:10.240 ⇒ 00:03:12.520 bencohen: All good, no worries at all.
10 00:03:12.850 ⇒ 00:03:18.080 Uttam Kumaran: How’s everything? So I’m a little bit under the weather today. So I’m like.
11 00:03:18.230 ⇒ 00:03:19.599 bencohen: You want to do this tomorrow.
12 00:03:19.600 ⇒ 00:03:23.429 Uttam Kumaran: No, no, I’m good. I’m good. Just may not go on video just like.
13 00:03:23.430 ⇒ 00:03:25.319 bencohen: That’s fine. I can shut off, too. So.
14 00:03:27.030 ⇒ 00:03:28.599 Uttam Kumaran: How’s everything with you?
15 00:03:29.220 ⇒ 00:03:30.640 bencohen: Real good.
16 00:03:31.980 ⇒ 00:03:38.250 bencohen: feel like summer’s in full swing. Now we’ve got some drama with Amazon. One of our competitors
17 00:03:38.840 ⇒ 00:03:45.629 bencohen: filed some bullshit claim on us. So they pulled a few of our key Amazon listings yesterday. So we’re just.
18 00:03:46.600 ⇒ 00:03:47.450 Uttam Kumaran: Oh, my God!
19 00:03:47.450 ⇒ 00:03:56.469 bencohen: I mean, it’s not based in reality, but it’s still a kick in the nuts. So we’re trying to deal with it. But other than that kind of business, as usual.
20 00:03:57.340 ⇒ 00:04:05.369 Uttam Kumaran: Cool. Yeah, we’re we’ve we got a few new clients. Some in like, Ecom, a few in like
21 00:04:06.126 ⇒ 00:04:12.230 Uttam Kumaran: in in Sas. And yeah, we’re we’re continuing to roll, so doing a lot more
22 00:04:12.460 ⇒ 00:04:16.459 Uttam Kumaran: in AI as well. I think over the last like, really like 6 months
23 00:04:16.579 ⇒ 00:04:19.800 Uttam Kumaran: since we started kind of going to market with AI stuff.
24 00:04:20.303 ⇒ 00:04:25.010 Uttam Kumaran: I mean, we we’ve been doing a ton of AI stuff internally for a while.
25 00:04:25.220 ⇒ 00:04:32.369 Uttam Kumaran: and then sort of. Tried to transition some of that to offer as a service. That’s probably been the most fun.
26 00:04:33.561 ⇒ 00:04:35.224 Uttam Kumaran: You know, of course.
27 00:04:35.840 ⇒ 00:04:43.129 Uttam Kumaran: but also like very brand new. So it’s the like, the tools and everything. It’s sort of changing almost every
28 00:04:43.360 ⇒ 00:04:46.149 Uttam Kumaran: few weeks. Yeah, like, really intense.
29 00:04:46.540 ⇒ 00:04:50.070 bencohen: Yeah, that’s the issue. The pace is just too much.
30 00:04:50.360 ⇒ 00:04:56.300 Uttam Kumaran: The the pace is quite a bit and if you know, it’s just like
31 00:04:56.530 ⇒ 00:05:00.569 Uttam Kumaran: if even our company is having a hard time keeping up, I think.
32 00:05:01.000 ⇒ 00:05:08.308 Uttam Kumaran: I mean, it may be a good thing for us, so that we sort of come in. We partner with folks and we help them make decisions. But
33 00:05:08.930 ⇒ 00:05:13.998 Uttam Kumaran: it’s moving really fast, although the tech is like here like it works. And
34 00:05:14.890 ⇒ 00:05:16.799 Uttam Kumaran: you know, it’s just a matter of like
35 00:05:17.550 ⇒ 00:05:28.009 Uttam Kumaran: adoption. I feel like for a lot of companies like we we do a lot of. We don’t do as much work building like customer facing stuff. A lot of it is like internal co-pilots or internal agents.
36 00:05:28.010 ⇒ 00:05:28.760 bencohen: Yep.
37 00:05:28.760 ⇒ 00:05:30.899 Uttam Kumaran: But the adoption is the hardest part.
38 00:05:31.380 ⇒ 00:05:32.390 bencohen: Yeah.
39 00:05:32.820 ⇒ 00:05:36.350 Uttam Kumaran: Like getting everybody to use it, and
40 00:05:36.700 ⇒ 00:05:40.089 Uttam Kumaran: like cause cause it’s like the average employee.
41 00:05:41.082 ⇒ 00:05:45.009 Uttam Kumaran: you know, like, may not even be using chat, gpt.
42 00:05:45.010 ⇒ 00:05:47.185 bencohen: Oh, I know we we
43 00:05:47.890 ⇒ 00:06:02.880 bencohen: one. Our content person made like a really simple gpt for blog creation and we wanted to do was we wanted to put our customer service team, especially the ones who have technical experience with pools to basically
44 00:06:03.240 ⇒ 00:06:16.830 bencohen: generate an article and then like, go through it with, like a, you know, like an experienced hand. Just make some simple edits and just kind of ship it out. One of the guys has never used chat, gpt, and he refuses to. And it’s like
45 00:06:17.680 ⇒ 00:06:20.709 bencohen: it’s I already understand, like how to even like talk to someone like that. It’s like.
46 00:06:20.710 ⇒ 00:06:34.799 Uttam Kumaran: Oh, no, yeah. I mean, I felt like this for like since I started using it 2 years ago. And so even if I hear that now it’s really tough. I mean, I feel like what’s changed is that folks at the top are sold.
47 00:06:35.692 ⇒ 00:06:45.119 Uttam Kumaran: and so that really allows adoption to happen. Well, base, I mean, you either make it sort of mandatory, or you find different ways of like
48 00:06:45.570 ⇒ 00:06:49.449 Uttam Kumaran: sort of promoting it. But that’s the toughest part is like
49 00:06:51.810 ⇒ 00:07:00.620 Uttam Kumaran: It’s just making sure everybody can get trained. And using that, I think that’ll change over time. But it’s an advantage if you can figure it out. Now, you know.
50 00:07:00.930 ⇒ 00:07:06.509 bencohen: Oh, God, yeah. Oh, yeah. The just real quick. On introducing my friend.
51 00:07:08.180 ⇒ 00:07:20.729 bencohen: nefco, they took on private equity a few years ago, and they have been buying like everything in their space. So they’re like in like a huge growth mode, because I think the private equity firm wants them to
52 00:07:21.150 ⇒ 00:07:25.379 bencohen: be acquired at, you know, 2027, 2028, something like that. So
53 00:07:25.530 ⇒ 00:07:31.680 bencohen: they’re like basically doing everything they can to grow faster. And I think, reading between the lines.
54 00:07:32.570 ⇒ 00:07:40.726 bencohen: They want to get more efficient and not hire like too much staff at the Hq. He didn’t say that, but I kind of felt that
55 00:07:41.430 ⇒ 00:08:05.699 bencohen: because they’re just buying like whole companies, like, even like some of them, are like almost the size of theirs. And it’s just a lot of it’s a lot. It’s just a lot. So I know the employee he put in charge of like spearheading AI, and it’s basically it sounds almost like a government thing where they’re like. All right, you know, this guy is going to be like the Aisar, and he’s not like he’s just younger. He’s not like A
56 00:08:06.150 ⇒ 00:08:10.060 bencohen: doesn’t come from. At least I don’t think it comes from data or anything like that. So
57 00:08:10.770 ⇒ 00:08:32.899 bencohen: my thought was that they need, and he was asking me like, do you think we should stick with Chat Gpt, or I’m sorry Copilot or Chatgpt, and he was like, you know, we’re already on Microsoft suite, and I was like, I don’t think there’s anything that copilot would be better at than Gpt, but I really don’t know, I said. I’ll think about it like I don’t know why I would even bother thinking about it. I’ll just introduce you. And you
58 00:08:34.450 ⇒ 00:08:38.900 bencohen: cause I think what is gonna happen is, I think that they are gonna need to.
59 00:08:39.049 ⇒ 00:08:42.410 bencohen: This guy is gonna need to basically
60 00:08:42.840 ⇒ 00:08:47.350 bencohen: assess a stack and also use cases. They they probably don’t even know all of what.
61 00:08:47.350 ⇒ 00:08:47.880 Uttam Kumaran: Yeah.
62 00:08:47.880 ⇒ 00:08:49.059 bencohen: What they can do.
63 00:08:49.510 ⇒ 00:08:50.500 bencohen: So
64 00:08:52.370 ⇒ 00:08:56.839 bencohen: I think that they need on the top level to work with a team like yours.
65 00:08:57.420 ⇒ 00:09:18.770 bencohen: and then make a plan for how to make it, you know, adopted by the company, but I think at the highest level they could probably just harmonize data and do some simple things that’ll make their lives easier and then getting their people. I mean, it’s like an old fashioned business. It’s not. They bought one E-com company, but for the most part it’s kind of order taking and order making.
66 00:09:18.770 ⇒ 00:09:19.220 Uttam Kumaran: Yeah.
67 00:09:19.220 ⇒ 00:09:23.590 bencohen: Was not like, it’s old school. So
68 00:09:23.760 ⇒ 00:09:27.596 bencohen: they have a lot of data, though. So I think it would be something.
69 00:09:28.170 ⇒ 00:09:35.850 Uttam Kumaran: Yeah, we I mean, we talked to a lot of folks kind of in that world where they’re like this is here and we don’t. We can’t miss the boat, and, like.
70 00:09:36.180 ⇒ 00:09:42.110 Uttam Kumaran: you know, even even for some of our clients, I said. Like, Look. Previously your options were like, try to grow
71 00:09:42.340 ⇒ 00:09:43.730 Uttam Kumaran: revenue
72 00:09:43.770 ⇒ 00:09:58.910 Uttam Kumaran: and like, try to keep costs low while you’re doing that. But like there’s, there’s not many options for like optimization. And there just happens to now be this technology that you can actually, basically ideally prevent a next hire or
73 00:09:59.204 ⇒ 00:10:15.980 Uttam Kumaran: have your existing hires take on 20 to 40% more or work basically work 20 to 40% more efficiently. And that’s the goal. Like I, I don’t go on and say, like, we’re here to replace everybody, or I don’t go. Say, you’re gonna like 10 x everything I’m like, if you could get
74 00:10:15.990 ⇒ 00:10:28.870 Uttam Kumaran: if you could aim for 20 to 40% more efficiency like, that’s pretty swell, you know. And so for people that are very practical, we, we basically try to break down. Okay, what are like the core
75 00:10:29.330 ⇒ 00:10:43.620 Uttam Kumaran: revenue facing bottlenecks in the business, whether it’s like people, whether it’s process. And then we sort of like, okay, is there a proof of concept here, or several proof of concepts that we can build towards, whether it’s on the sales side, like a lot of people aren’t doing basic like
76 00:10:43.890 ⇒ 00:11:01.920 Uttam Kumaran: using AI to filter down leads and and generate. You know, emails and stuff like that like, that’s a very simple thing to increase your like deal volume. It could be like processing or qualification. And then there’s there’s stuff on like, okay, post. Once you close like, what are the steps that need to get taken care of?
77 00:11:02.538 ⇒ 00:11:14.150 Uttam Kumaran: It could also be like, Hey, there’s a big step where a lot of people are spending time looking through documents to figure something out great. We should build something on top of a bunch of documents that allows them to search those common questions.
78 00:11:15.520 ⇒ 00:11:21.170 Uttam Kumaran: but but again, it’s like for us, like I try to come to the table with like, Okay, here’s like, kind of like, 5 use cases.
79 00:11:21.676 ⇒ 00:11:32.750 Uttam Kumaran: And I tell them, like, Look, we want to help you prevent your ideally like, not hire at the same rate, and get your people to be more efficient with their time.
80 00:11:33.020 ⇒ 00:11:34.750 bencohen: Yeah, yeah.
81 00:11:34.750 ⇒ 00:11:44.739 bencohen: I think that they I think he I mean Matt’s the CEO. His father founded the company, and then he took it over around the time of the deal. He basically, I basically said.
82 00:11:44.860 ⇒ 00:12:02.203 bencohen: you know, this doesn’t replace humans, it just makes you guys more efficient, you know. So if you guys can each accomplish 30% more work, that’s a good thing. And he was like, you know, I think he’d love that. So reply and get some time, and then I think you’ll speak to him. And this guy named Jan, who I’ve met before. And
83 00:12:03.090 ⇒ 00:12:04.760 bencohen: yeah, you know what to do. I think.
84 00:12:04.760 ⇒ 00:12:05.490 Uttam Kumaran: Yeah, yeah.
85 00:12:05.490 ⇒ 00:12:11.685 bencohen: They’ll get it. And and I think it, it’s probably a pretty high likelihood that they’ll engage, because
86 00:12:12.160 ⇒ 00:12:14.669 bencohen: it’s like officially, a priority.
87 00:12:15.220 ⇒ 00:12:17.690 Uttam Kumaran: Cool. Yeah, they have a facility here in Austin.
88 00:12:18.260 ⇒ 00:12:22.079 bencohen: Everywhere and like, even if they don’t have a facility they may have bought
89 00:12:22.573 ⇒ 00:12:27.580 bencohen: like a whole, like regional company that has like like they like would buy like something that has, like
90 00:12:27.800 ⇒ 00:12:29.580 bencohen: the whole southeast covered.
91 00:12:29.985 ⇒ 00:12:30.390 bencohen: Okay?
92 00:12:30.390 ⇒ 00:12:39.569 bencohen: And in some cases I don’t even know if they rebrand I’m not even sure how that it’s. It’s it’s it’s growing. He’s always flying somewhere, and they’re buying something. It’s it’s like crazy.
93 00:12:40.040 ⇒ 00:12:53.439 Uttam Kumaran: Okay, cool. Yeah, I appreciate it. So in terms of like that, I think. You know where it’s actually good timing we’re we were planning on reaching out to you and some people. But we’re planning sort of like a referral. Pretty simple. Just referral program
94 00:12:53.843 ⇒ 00:13:07.790 Uttam Kumaran: where we can equip certain folks in our network with just like, Hey, here’s a bunch of materials on our side, and sort of how to talk to Brainforge if you need it. And then we probably just try to set up like a quarterly or something to just talk about
95 00:13:08.440 ⇒ 00:13:13.560 Uttam Kumaran: potential opportunities. And we can also share sort of the clients that we have worked with.
96 00:13:14.083 ⇒ 00:13:20.229 Uttam Kumaran: And I know you’re you’re pretty well connected and and have your sort of toes in a bunch of
97 00:13:20.420 ⇒ 00:13:32.949 Uttam Kumaran: different areas. And we have a couple of sort of friends of brain forge that are like that. So was hoping to get something like that set up. We’re just like finalizing sort of what that contract would look like but would love to
98 00:13:33.070 ⇒ 00:13:34.700 Uttam Kumaran: sort of make that happen.
99 00:13:34.700 ⇒ 00:13:36.210 bencohen: Yeah, that’s a great idea.
100 00:13:36.210 ⇒ 00:13:36.730 Uttam Kumaran: Cool.
101 00:13:36.730 ⇒ 00:13:46.009 bencohen: Great idea, I think, as I’m getting more and more into conversations outside of the pool world. Once I get into these orgs, I’ll probably start
102 00:13:46.850 ⇒ 00:14:03.979 bencohen: having an opportunity to like bring in more brains. So I think it’s only going to get bigger. But I think I mean, this is a really good 1st one, and then I imagine there’ll be like a regular cadence, because everyone’s thinking in this direction, and they don’t know what to do. And they like come to someone like me. That they know is like a little bit tech forward.
103 00:14:03.980 ⇒ 00:14:05.130 Uttam Kumaran: Oh, yeah.
104 00:14:05.130 ⇒ 00:14:13.360 bencohen: And they’re like, what do you think? And it’s like, I use some of this stuff. But like, I really can’t like connect the dots for you. So there needs to be an answer.
105 00:14:13.720 ⇒ 00:14:27.679 Uttam Kumaran: Yeah. And it’s also, you know, I actually, I actually am really happy. If if folks have used at least chatgpt before, because then they sort of get the magic like. I don’t have to explain to them like that, because that’s kind of it’s like not really easy to.
106 00:14:27.880 ⇒ 00:14:48.170 Uttam Kumaran: So so if they’re using that, then I can quickly show like, okay, here’s how. Here’s the reason why you need the most up to date context. You need this in slack, or your teams, or wherever you you need it. You need it over your documents. You know, you need to automate it in the background like. There’s a lot of ways for me to explain the use cases, but
107 00:14:48.270 ⇒ 00:14:57.969 Uttam Kumaran: we still talk to folks who haven’t used it at all, and they’re like we need to. I’m like to start. You should just start by using in your personal life instead of Google, and then call me back once you kind of get it
108 00:14:58.580 ⇒ 00:15:00.249 Uttam Kumaran: so hard to to like.
109 00:15:01.230 ⇒ 00:15:03.759 Uttam Kumaran: They’re just not gonna get the there isn’t.
110 00:15:03.760 ⇒ 00:15:04.480 bencohen: To go online.
111 00:15:04.480 ⇒ 00:15:07.480 Uttam Kumaran: Sort of speaking like I’m talking about magic, you know. I don’t know.
112 00:15:07.820 ⇒ 00:15:13.749 bencohen: They need to be in, they need to go into it. That’s the only the only way on this is to go ahead first.st
113 00:15:14.520 ⇒ 00:15:15.170 Uttam Kumaran: Yeah.
114 00:15:15.380 ⇒ 00:15:16.060 Uttam Kumaran: Okay.
115 00:15:16.730 ⇒ 00:15:33.770 Uttam Kumaran: okay, cool. So yeah. So Miguel on, my team is on the call. He’s the one that worked on a lot of most of the stuff that we’ve currently developed. We do have a lot of, I think, different ways to go but maybe I’ll just share
116 00:15:34.250 ⇒ 00:15:40.392 Uttam Kumaran: my screen briefly. And yeah, would just love to get like a sense of what the overall
117 00:15:42.130 ⇒ 00:15:46.709 Uttam Kumaran: you know, plan is, and like what what we can try to iterate on here.
118 00:15:47.450 ⇒ 00:15:47.845 bencohen: Alright.
119 00:15:51.330 ⇒ 00:15:59.520 bencohen: So I think it for chemistry. I think it’s really good. Yeah, I guess the question is.
120 00:15:59.920 ⇒ 00:16:02.360 bencohen: when we talk about like the ingestion of
121 00:16:02.560 ⇒ 00:16:07.400 bencohen: all information, had, like somewhat of a of a big
122 00:16:08.040 ⇒ 00:16:14.400 bencohen: idea with it. Which is that you know.
123 00:16:15.320 ⇒ 00:16:19.100 bencohen: I put on, do not disturb, and text still come through. I don’t even know how that’s possible.
124 00:16:20.930 ⇒ 00:16:25.810 bencohen: Think it’s because it’s on my other Mac that’s next to it. Let me just shut that off. Okay, Nope, didn’t work.
125 00:16:26.460 ⇒ 00:16:31.510 bencohen: Hold on. This will kill me, anyway.
126 00:16:33.950 ⇒ 00:16:40.629 bencohen: the idea is, everybody that has a pool could be an opportunity. So I think
127 00:16:41.400 ⇒ 00:16:44.549 bencohen: this is where I think our
128 00:16:44.670 ⇒ 00:16:48.660 bencohen: our thoughts are different. But he basically wants to take.
129 00:16:49.710 ⇒ 00:16:54.080 bencohen: you know, anything that we find anywhere, whether it’s Facebook
130 00:16:54.330 ⇒ 00:17:05.739 bencohen: forums, pool Care pool care handbook, which I have a digital copy of, and I also have a paper copy and just upload all of it.
131 00:17:05.880 ⇒ 00:17:19.407 bencohen: so that this has like fucking everything, and that we also have without needing to wait for you guys. Our team has a way to just upload whatever the hell we want and the system, you know, digests.
132 00:17:20.300 ⇒ 00:17:24.071 bencohen: my thought, though, was a little more focused around
133 00:17:24.890 ⇒ 00:17:28.120 bencohen: like fixing problems, because I think
134 00:17:28.580 ⇒ 00:17:34.239 bencohen: it’s very my concern that what I see happen is people try to go through like
135 00:17:34.440 ⇒ 00:17:51.690 bencohen: Youtube, or they’ll go through like a few different videos to try to solve their own problem. The 1st video, they invest 20 min. It’s not quite right. The second video, they lose a little patience. After 5 min they bounce, and then finally they give up and get angry, or they’re right. Our customer service, I think we need to.
136 00:17:51.940 ⇒ 00:18:00.480 bencohen: And the reason why, by the way, why that is the case is because every single pool is different. Everything is different, you know, even if the pools are like
137 00:18:00.840 ⇒ 00:18:03.900 bencohen: designed and equipment the same.
138 00:18:04.060 ⇒ 00:18:08.969 bencohen: if they’re in a different location. Even within the same town they can behave differently because of the
139 00:18:09.150 ⇒ 00:18:32.199 bencohen: trees, whatever amount of sun it gets, so the answer can always be different. And then you have other variables, like the technician that’s installing your pump might have like a preference on how they like angle the pipe, or, you know, like real brainy, really like micro stuff. So my thought with all this is, how do you coach somebody
140 00:18:32.750 ⇒ 00:18:34.460 bencohen: into doing these things?
141 00:18:35.070 ⇒ 00:18:51.620 bencohen: That’s my thought is, that’s the biggest money saver we can provide, because the majority of people we sell to either are Diy or they want to be, and the biggest barrier to that is knowledge and and like guidance. So
142 00:18:52.040 ⇒ 00:18:55.620 bencohen: I have found myself thinking about that more than anything which is like.
143 00:18:55.840 ⇒ 00:18:57.579 bencohen: No matter what I buy
144 00:18:58.440 ⇒ 00:19:05.950 bencohen: that needs installation like. I’m always 1st like I don’t feel like spending $300 on that handyman to come over. Let me see if I can do it, and then
145 00:19:06.270 ⇒ 00:19:10.999 bencohen: inevitably, I can’t do it because I bought the thing. And from, you know, a Chinese.
146 00:19:11.280 ⇒ 00:19:16.610 bencohen: you know factory in China that has Fba, and it’s productions are not great, and
147 00:19:16.900 ⇒ 00:19:20.649 bencohen: the video sucks, and I just give up and pay for it. And
148 00:19:20.840 ⇒ 00:19:40.778 bencohen: I think that we can conquer that for the pool I really do. And I think we could even go out of our products because some of these things are agnostic to the, to the brand and help people do that. And then I think if we help people install other brands pools that we actually provide them like a real service.
149 00:19:41.840 ⇒ 00:19:47.090 bencohen: we’ve got them, you know, and we have a chance now at selling them something or whatever. So
150 00:19:47.290 ⇒ 00:20:00.379 bencohen: I I think in in, I think in both areas, we need to continue one with the ingestion and training, and 2 with like actual as as much as we can. Kind of
151 00:20:00.780 ⇒ 00:20:04.650 bencohen: problem, solving advice and guidance.
152 00:20:06.000 ⇒ 00:20:21.281 Uttam Kumaran: So I think you’re right in that. The more specific of the use case that we can have the agent do the better outcomes you’re gonna have right now to give you a sense like our original sort of thing that we looked at was looking at.
153 00:20:21.870 ⇒ 00:20:34.780 Uttam Kumaran: sorry, looking at this, which was, here’s the pop questions that we wanted to answer. And a lot of it was about about pool related problems, but not necessarily about
154 00:20:35.000 ⇒ 00:20:47.049 Uttam Kumaran: like products to buy to solve them, or how to implement those products right? Like, there is some on equipment. But a lot of them are just generic questions right? Even we went and scraped.
155 00:20:47.240 ⇒ 00:20:54.130 Uttam Kumaran: We’ve sort of built a scraper for for reddit threads. And so we have a lot of we can pull as many of those as we want
156 00:20:54.538 ⇒ 00:20:59.630 Uttam Kumaran: same thing with with Facebook groups. We ended up getting a ton of and
157 00:20:59.630 ⇒ 00:21:05.679 Uttam Kumaran: a ton of a ton of data, and Facebook gave us like a lot of stuff. But we, we have the actual like
158 00:21:05.890 ⇒ 00:21:13.070 Uttam Kumaran: question, answer threads. And the group the group that came from, and bike
159 00:21:14.460 ⇒ 00:21:18.310 Uttam Kumaran: replacing liner in our 22 round pool. Right? So all of these
160 00:21:18.830 ⇒ 00:21:33.420 Uttam Kumaran: and we can get as many of these as we want but I do think it’s like it’s it’s helpful to think about either having a how they’re having the use case based agents and then having a sort of different chat experience based on that
161 00:21:35.190 ⇒ 00:21:49.365 Uttam Kumaran: right cause right now. It’s pretty simple, just text based chat. But if you’re pulling something from a manual you might want. It’s actually pulled the document up and sort of put a box over where it’s pulling from.
162 00:21:50.020 ⇒ 00:22:03.279 Uttam Kumaran: you also may want, you know, 1 1 common ui sort of pattern that we do is we also do have suggestions, right? So based on the question, we could say, here’s what the next question you could ask could be.
163 00:22:05.170 ⇒ 00:22:14.830 Uttam Kumaran: you know, it’s also, as soon as you come in here we should have examples of questions that you can ask. But if you do think that it’s gonna be more about fixing, then? Then we sort of need to.
164 00:22:15.010 ⇒ 00:22:20.789 Uttam Kumaran: We probably need to walk through with Cody or whomever on like, how do they typically
165 00:22:21.319 ⇒ 00:22:31.569 Uttam Kumaran: issue these instructions like, like, maybe it’s step by step. And then you say, Okay, let me know when you’ve done that. Let me know when you’ve done that right. Those types of the chat base interface
166 00:22:32.097 ⇒ 00:22:42.589 Uttam Kumaran: and we can certainly upload any document like we. We sort of have this knowledge base thing. Now, you can actually go up. You can upload. We can enable that and
167 00:22:44.320 ⇒ 00:22:47.964 Uttam Kumaran: it’ll get loaded, indexed and then made available.
168 00:22:48.740 ⇒ 00:22:52.419 Uttam Kumaran: but I do think we should have a very clear
169 00:22:53.120 ⇒ 00:22:59.769 Uttam Kumaran: it it the more, broader set of questions, the less accurate. It’s going to be about one of them like we just can’t go deep on one of them. Yeah.
170 00:22:59.770 ⇒ 00:23:01.429 bencohen: I think it’s like, you know.
171 00:23:01.650 ⇒ 00:23:07.159 bencohen: you start in a simple use case, and Cody will be better. But use case number one.
172 00:23:07.350 ⇒ 00:23:15.180 bencohen: You’re making a pool from scratch, and you want to install your equipment. So like you’re starting from absolute 0. And then you.
173 00:23:15.370 ⇒ 00:23:22.169 bencohen: you know that that’s the easiest in some ways, because you can just tell them what to do. And there’s you’re not like navigating around.
174 00:23:22.390 ⇒ 00:23:28.310 bencohen: you know, weird weirdly or incorrectly installed stuff that’s already there, that doesn’t.
175 00:23:28.630 ⇒ 00:23:31.890 bencohen: And then you then you go from you know, how do you replace
176 00:23:32.250 ⇒ 00:23:39.410 bencohen: that same scenario? But 3 years later, which is, there’s already a you know, a pad full of equipment.
177 00:23:39.530 ⇒ 00:23:44.100 bencohen: the the pump or the heater broke. We’re going to replace it with another one.
178 00:23:44.410 ⇒ 00:23:53.839 bencohen: How do you do that? And there’s some replumbing and whatever. And I think that that would I mean, I know what Dan’s thinking. It’s just like, let’s be Google for the pool.
179 00:23:54.139 ⇒ 00:23:54.440 Uttam Kumaran: But.
180 00:23:54.732 ⇒ 00:24:06.140 bencohen: Which I think is a nice idea, but it’s just too open, and I don’t think that we can do it as well as like chat, gpt like who can keep up with them. You know what I mean.
181 00:24:06.140 ⇒ 00:24:06.730 Uttam Kumaran: Yeah, yeah.
182 00:24:06.730 ⇒ 00:24:25.619 bencohen: But I do think that there’s a huge gap that Cody knows a lot about, which is, how do you actually empower people to save money and be in control of their pool. And I think that that’s kind of the magic here. I think with chemicals it’s a big one, because a lot of people.
183 00:24:25.830 ⇒ 00:24:32.649 bencohen: you know, would like to do that. And then the next one is equipment, you know, diagnosing what you know, like the pool
184 00:24:33.180 ⇒ 00:24:43.919 bencohen: could be green, for example, which is something that that’s a that’s a it could straddle both sides. It could be a chemical problem. It could also be an equipment problem. How do we help them figure out that
185 00:24:44.040 ⇒ 00:24:49.329 bencohen: their filter sucks or that they haven’t done something correctly, you know, that’s
186 00:24:49.670 ⇒ 00:24:56.853 bencohen: that’s something like the diagnosing Cody knows all about that, like, I think, set a call with Cody, and we can get into all of this.
187 00:24:57.460 ⇒ 00:25:05.679 bencohen: But yeah, I think that that’s like an area that my goal for this was to was basically to save money for
188 00:25:05.810 ⇒ 00:25:13.590 bencohen: our customers and future customers. Because I think that that’s our differentiated area, versus our competition.
189 00:25:14.280 ⇒ 00:25:17.000 Uttam Kumaran: I guess my my other question is, gonna be like, how does
190 00:25:17.200 ⇒ 00:25:20.129 Uttam Kumaran: lead generation? Or like selling products
191 00:25:20.390 ⇒ 00:25:31.100 Uttam Kumaran: come into play like, is that top of mind for this like, do we want to consider inserting links to products or suggesting products?
192 00:25:32.710 ⇒ 00:25:34.259 Uttam Kumaran: Just maybe we can start there.
193 00:25:36.030 ⇒ 00:25:39.020 bencohen: I mean, obviously, we want to sell stuff. Yeah.
194 00:25:39.430 ⇒ 00:25:45.969 bencohen: so definitely, if it if it seems like the customer, is that an area where
195 00:25:46.350 ⇒ 00:25:56.260 bencohen: you know, they’re at an impasse, and like the next step is to buy a product to install. We would like to be in that. We, you know, we want. We wanna we want to sell something for sure.
196 00:25:57.460 ⇒ 00:26:06.210 Uttam Kumaran: Sounds okay, that makes sense. So similarly, with like lead generation, you know, or like escalations,
197 00:26:08.460 ⇒ 00:26:17.759 Uttam Kumaran: is there an opportunity to say like, well, if you have any other questions, feel free to enter your email like some sort of capture.
198 00:26:17.760 ⇒ 00:26:19.517 bencohen: Yeah, I think I think we
199 00:26:20.300 ⇒ 00:26:27.320 bencohen: yeah, I think we we, I guess there’s 2 ways you can. We can reach out to them, or we can move them into like
200 00:26:28.440 ⇒ 00:26:33.079 bencohen: our Zendesk chat and talk to the team, live, or a combo of both.
201 00:26:33.080 ⇒ 00:26:39.579 Uttam Kumaran: It could also be like a Hey, we can. Well, it’s we’re happy to email you a summary of this conversation for your email.
202 00:26:39.680 ⇒ 00:26:43.650 Uttam Kumaran: Or if you have any questions, you would give your phone number. We can call you back.
203 00:26:44.590 ⇒ 00:26:46.590 Uttam Kumaran: Right? Something like that.
204 00:26:46.980 ⇒ 00:26:56.019 bencohen: Okay, yeah, I think I think any of that can all work. I think with those kinds of things, it’s all about momentum. So like, if you have somebody that’s kinda hot, you don’t wanna
205 00:26:56.610 ⇒ 00:27:01.880 bencohen: like a 6 a 6 h lapse in email. It might not even be relevant anymore.
206 00:27:01.880 ⇒ 00:27:02.500 Uttam Kumaran: Yeah.
207 00:27:02.730 ⇒ 00:27:06.500 bencohen: Like. If it’s me, I would just give up, and just like I would just call them.
208 00:27:07.070 ⇒ 00:27:32.849 Uttam Kumaran: You know the other thing that we can do. And actually, Miguel, who’s on the call? This is actually why why I originally brought him on is, we’ve done a lot of work with AI phone calling. And so one option we can do. And again, I think this is probably dependent on just thinking about the audience. That may be older folks like you could have the AI say, like, Are you having trouble like I’m happy to call you and walk you through it.
209 00:27:33.464 ⇒ 00:27:37.019 Uttam Kumaran: Like voice related. AI has gotten extremely good
210 00:27:37.439 ⇒ 00:27:42.469 Uttam Kumaran: in that. There’s like a very low latency like, even for folks like us. And for people that
211 00:27:42.960 ⇒ 00:27:58.409 Uttam Kumaran: are used to call center calling, it’s like, this is a lot better. And it will just walk through basically the exact same text that you’ll get like we could modify it to be like more or less text heavy because we want to image. But that could be a great option as well to offer like.
212 00:27:58.960 ⇒ 00:28:03.060 Uttam Kumaran: If if you wanted to handle this over a call, we can give you a call and walk you through it too.
213 00:28:03.460 ⇒ 00:28:10.379 bencohen: Yeah, no, I think we. I think we might as well walk down that I mean, I use. I actually use
214 00:28:11.080 ⇒ 00:28:14.830 bencohen: chat, gpt, like the audio, and
215 00:28:15.220 ⇒ 00:28:18.029 bencohen: the intonation has gotten so much better. By the way.
216 00:28:18.030 ⇒ 00:28:22.379 Uttam Kumaran: Yeah, it’s way better the last like month and a half or something. They did a big update. It’s super.
217 00:28:23.070 ⇒ 00:28:27.166 bencohen: Huge. I mean, I remember, for, like the nft stuff when we were making videos,
218 00:28:28.240 ⇒ 00:28:31.840 bencohen: voice like a tone was just so difficult.
219 00:28:32.480 ⇒ 00:28:41.829 bencohen: and part of it was on whatever that oceans 11 or whatever the hell website it was. I haven’t used it in a while, but we would just kind of paste, a script.
220 00:28:42.150 ⇒ 00:28:42.880 Uttam Kumaran: Yes.
221 00:28:42.880 ⇒ 00:28:45.740 bencohen: It wasn’t smart enough to understand
222 00:28:45.900 ⇒ 00:28:49.109 bencohen: how it the tone that it ought to have.
223 00:28:49.300 ⇒ 00:28:56.749 bencohen: So it was just tough. We ended up hiring voice actors for some stuff. It just wasn’t good enough. But I think now it’s probably yeah.
224 00:28:56.750 ⇒ 00:29:03.499 Uttam Kumaran: I think you’ll be surprised at how insane like the chat Gpt has a has quite a bit of latency, just because.
225 00:29:03.730 ⇒ 00:29:19.140 Uttam Kumaran: of course they’re like searching through their entire corpus of stuff in a very directed like voice agent. You can get very good latency, and they have tons of options for voice. You can add background noise. You can have it like stutter, or like
226 00:29:19.530 ⇒ 00:29:23.910 Uttam Kumaran: sort of like act. Very like it’s just. It’s like quite insane.
227 00:29:24.347 ⇒ 00:29:36.169 Uttam Kumaran: And the latency is really it’s it’s not that bad because you have an agent. So it’s like a very simple, prompt. And then it’s pulling from a knowledge base. So and frankly, it actually builds on a lot of the work. It’s not like
228 00:29:36.760 ⇒ 00:29:38.110 Uttam Kumaran: an additional
229 00:29:38.350 ⇒ 00:29:45.240 Uttam Kumaran: tons of work to set that up, and we have. We have demo voice agents that we can even share with you to try
230 00:29:46.160 ⇒ 00:29:48.680 bencohen: It’d be interesting like, I wonder.
231 00:29:48.840 ⇒ 00:29:53.270 bencohen: even like, by the way, just like even the the part that you’ve done with chemistry.
232 00:29:53.270 ⇒ 00:29:53.750 Uttam Kumaran: Yeah.
233 00:29:53.750 ⇒ 00:30:00.180 bencohen: There’s some parts of it. Somebody might prefer to just talk and not like my dad, for example, does not type like.
234 00:30:00.180 ⇒ 00:30:06.170 Uttam Kumaran: That’s who I’m thinking of. That’s really who I’m thinking about. Like, yeah, my mom would definitely call versus.
235 00:30:06.330 ⇒ 00:30:11.098 Uttam Kumaran: She’d be on her reading glasses and like trying to type. And her font is like 900. So
236 00:30:11.350 ⇒ 00:30:19.169 Uttam Kumaran: oh, it’s a whole mess, or just like, if we ever turn this thing into a mobile app just hitting the you know the wave at the bottom, and just
237 00:30:19.170 ⇒ 00:30:19.670 Uttam Kumaran: yes.
238 00:30:19.670 ⇒ 00:30:24.339 bencohen: That’s kind of the same as a call. It’s just in your own pace, I suppose.
239 00:30:25.530 ⇒ 00:30:49.900 Uttam Kumaran: And then my other question is sort of about the ui. Right now, you know where it’s very basic, like, it’s just like text text, we can do other like ui elements, like different types of components. Or, for example, really, what I’m thinking is like, if we propose a product, we should have the product images pop up with like the summary and like where to buy it. So we can get more creative than just
240 00:30:50.100 ⇒ 00:30:52.140 Uttam Kumaran: tax tax back and forth.
241 00:30:52.690 ⇒ 00:30:55.630 Uttam Kumaran: So one way is like, if we
242 00:30:56.050 ⇒ 00:31:09.699 Uttam Kumaran: if like, something has like step by step, we if like, we’re doing a lot of. For example, if we’re doing a lot of step by step work, we should build a little bit of a thing that like helps instead of just a block of text. It’s like step one step 2. We can carve it out right.
243 00:31:11.430 ⇒ 00:31:19.959 Uttam Kumaran: The things where the more specific our replies can be, the the more tailored of of a ui experience. We can have
244 00:31:20.621 ⇒ 00:31:42.640 Uttam Kumaran: otherwise we have ton of our supporting, like so many replies that but we could kind of still do it. But it’s it’s a lot like if if we’re like, if if Cody’s like, Hey, we always kind of go step by step, too. And the user needs to click like a confirm. I did this or reply with what they did before moving on. Then there’s a really like simple
245 00:31:42.910 ⇒ 00:31:52.569 Uttam Kumaran: Ui, we can just build right into this. That’s that would like really facilitate that versus like, here’s a huge wall of text, or like it’s so much tech.
246 00:31:52.570 ⇒ 00:31:57.850 bencohen: Let’s do it. Let’s do it. And I think Dan even tried to. I don’t think he communicated it.
247 00:31:57.850 ⇒ 00:32:03.849 Uttam Kumaran: To communicate, because it’s I don’t. Only after building these several times can I even like, walk through that. But.
248 00:32:03.850 ⇒ 00:32:10.410 bencohen: Well he was. He was trying to show you like, you know, where’s the nearest place to buy chemicals? How the chat showed like a map.
249 00:32:10.410 ⇒ 00:32:13.010 Uttam Kumaran: That showed like the picture and stuff like that. Yeah.
250 00:32:13.321 ⇒ 00:32:18.929 bencohen: Think what he was trying to say is like, let’s use like like bold and like chunking and.
251 00:32:18.930 ⇒ 00:32:19.700 Uttam Kumaran: Yeah, yeah.
252 00:32:19.700 ⇒ 00:32:26.080 bencohen: Let’s do that, do it. Do as much as you can on that. So it’s really looks easy on the eyes. And then
253 00:32:26.530 ⇒ 00:32:28.580 bencohen: I could have my designer
254 00:32:28.870 ⇒ 00:32:34.380 bencohen: like help a little bit with style like if we think that maybe there’s a font, or you know, whatever we can.
255 00:32:34.380 ⇒ 00:32:41.880 Uttam Kumaran: Take the 1st crack and then modern, like, yeah, we can hand it off to sort of make it like sharper. I just want to sort of give you guys
256 00:32:42.340 ⇒ 00:32:51.790 Uttam Kumaran: what we’re seeing in chat interfaces like kind of on the edge, like, I feel like a perplexity is a good example of sort of like the Max of what’s possible when you like.
257 00:32:52.170 ⇒ 00:33:01.059 Uttam Kumaran: Go deep on the ui like. If you go on perplex and you type in a stock ticker. It’s great. You get a ton of stuff similarly, with like news and other things like they really.
258 00:33:01.310 ⇒ 00:33:06.810 Uttam Kumaran: they’re probably the deepest on, like, what is chat based Ui like what it can be, and so.
259 00:33:06.810 ⇒ 00:33:09.039 bencohen: I haven’t used. I actually stopped using it.
260 00:33:09.440 ⇒ 00:33:22.820 Uttam Kumaran: I don’t. I don’t use it. I use it only for news that’s like, for example, there’s like a new part of the new Trump Bill. I wanted to look something up. I was like, let me just ask perplexity, because maybe it has the knowledge, so I don’t. I don’t use it as often.
261 00:33:23.422 ⇒ 00:33:32.329 Uttam Kumaran: But sometimes I want to get the article like, I want the articles behind it. We’re writing a blog or something. And I’m like, okay, I want to just get like 5 references.
262 00:33:32.650 ⇒ 00:33:38.390 Uttam Kumaran: It’s helpful. Although Chatgpt has search mode. So yeah, yeah.
263 00:33:38.390 ⇒ 00:33:44.240 bencohen: The coinbase ticker. It does. I see what you’re saying. It does look nice. You can easily understand what’s going on.
264 00:33:44.240 ⇒ 00:33:50.159 Uttam Kumaran: Yeah. So they have several that, and also for shopping. So perplexity is the only one that is really going deep on shopping.
265 00:33:50.990 ⇒ 00:33:55.130 Uttam Kumaran: Where, if you type in, I mean, I don’t know. I can even see it. I don’t know what it looks like for.
266 00:33:56.110 ⇒ 00:33:56.710 bencohen: Let’s see.
267 00:33:56.940 ⇒ 00:33:58.139 Uttam Kumaran: Cool products.
268 00:34:01.100 ⇒ 00:34:02.060 bencohen: I’m checking.
269 00:34:02.530 ⇒ 00:34:09.260 bencohen: I spelled Pump wrong intentionally to see if they’d catch it, which they did jeez they’re so fucking impressive.
270 00:34:12.730 ⇒ 00:34:14.779 Uttam Kumaran: Yeah, so this is like what I’m seeing.
271 00:34:20.330 ⇒ 00:34:22.159 bencohen: Yeah, they’re doing a killer job.
272 00:34:22.530 ⇒ 00:34:27.790 bencohen: Samsung just wrote them a monster. Vc. Check. My friend, that works with Samsung was telling me.
273 00:34:27.790 ⇒ 00:34:28.730 Uttam Kumaran: Oh, wow!
274 00:34:28.730 ⇒ 00:34:32.279 bencohen: Monster like hundreds of millions of dollars.
275 00:34:32.770 ⇒ 00:34:36.879 Uttam Kumaran: And see. You can buy directly within perplexity, too. So I just think like.
276 00:34:37.380 ⇒ 00:34:41.590 Uttam Kumaran: see. But see, this is a unique shopping like experience, right where
277 00:34:42.190 ⇒ 00:34:46.690 Uttam Kumaran: they sort of just give you the highlights. They give you the most popular products, and then they’re like.
278 00:34:48.179 ⇒ 00:34:52.410 Uttam Kumaran: you can see some pictures, although I mean it could evolve because Google shopping was
279 00:34:52.949 ⇒ 00:35:00.400 Uttam Kumaran: kind of a dud like I think people use it, but like not like to the extent I mean. Amazon is where I think. Still, people start their shopping experience.
280 00:35:00.400 ⇒ 00:35:05.720 bencohen: Yeah, yeah, I think, yeah, Google shopping, we have data that it works. But.
281 00:35:05.720 ⇒ 00:35:10.109 Uttam Kumaran: You have. I think you guys are really out of everyone I’ve met. You guys are the only folks that are like
282 00:35:10.790 ⇒ 00:35:12.679 Uttam Kumaran: really doing well, there.
283 00:35:12.680 ⇒ 00:35:14.588 bencohen: Yeah, we’re doing fine there
284 00:35:15.900 ⇒ 00:35:20.370 Uttam Kumaran: Perplexity. You know the thing with perplexity is like it might not even exist in 2 years.
285 00:35:20.370 ⇒ 00:35:21.969 Uttam Kumaran: Totally. Yeah, no, totally. I mean.
286 00:35:23.540 ⇒ 00:35:29.409 bencohen: I actually wonder what’s going to happen with this whole space? Because it’s all going to consolidate. Obviously.
287 00:35:29.720 ⇒ 00:35:31.019 bencohen: I don’t even know
288 00:35:31.420 ⇒ 00:35:36.020 bencohen: it’s hard to bet against Elon Musk, I think. I wonder who’s going to start buying like.
289 00:35:36.250 ⇒ 00:35:39.669 Uttam Kumaran: That’s what you’re seeing. I think they’re what they’re doing right now is they’re buying the talent
290 00:35:40.710 ⇒ 00:35:41.310 bencohen: Yeah.
291 00:35:41.310 ⇒ 00:35:57.739 Uttam Kumaran: You’re seeing like Facebook. And these guys really give a huge packages because there’s not many people that can do a lot of the foundational work for this and really they want to own the infrastructure versus like owning the plot. They want to own the platform. They don’t really care about building these experiences.
292 00:35:57.880 ⇒ 00:36:03.430 Uttam Kumaran: Chat, Gpt has to do both because they don’t have any user base. Right? So Facebook.
293 00:36:03.560 ⇒ 00:36:08.769 Uttam Kumaran: whatever they build, they just release to the billions of people, and they immediately trying to get away.
294 00:36:08.770 ⇒ 00:36:11.290 bencohen: The the distribution is pretty convenient.
295 00:36:11.290 ⇒ 00:36:17.879 Uttam Kumaran: The distribution is very convenient, but their product isn’t that good? Where chat Gbt sort of like.
296 00:36:18.570 ⇒ 00:36:25.460 Uttam Kumaran: I mean, it’s just. It’s just great, and like I’m I sort of don’t use, Claude. I just all that’s all I use.
297 00:36:25.580 ⇒ 00:36:31.800 Uttam Kumaran: and internally we use gpt behind the hood, and they’re all sort of converging on the same performance. So I don’t really care
298 00:36:32.000 ⇒ 00:36:38.149 Uttam Kumaran: like we have a bit of a deal with azure, because we’re like an azure startup. So I was like, put it all there, like, you know.
299 00:36:38.150 ⇒ 00:36:40.789 bencohen: Yeah, yeah, yeah, no, it’s
300 00:36:42.600 ⇒ 00:36:47.900 bencohen: it’s interesting. I was talking to my friend about it that I’m making an app with with AI, and
301 00:36:48.150 ⇒ 00:36:55.199 bencohen: he’s he. He’s like, you know, Vibe coding. He’s doing that whole and
302 00:36:56.160 ⇒ 00:36:58.790 bencohen: yeah, I he kind of agreed. He was.
303 00:36:59.030 ⇒ 00:37:00.900 bencohen: He uses cursor.
304 00:37:00.900 ⇒ 00:37:05.430 Uttam Kumaran: Yeah, we use cursor, too, but cursor under the hood. You’re using Claude or.
305 00:37:05.790 ⇒ 00:37:06.310 bencohen: Yeah.
306 00:37:07.140 ⇒ 00:37:08.120 Uttam Kumaran: One of them.
307 00:37:08.310 ⇒ 00:37:10.300 bencohen: He likes Claude a lot. He.
308 00:37:10.300 ⇒ 00:37:17.250 Uttam Kumaran: All the all, the all the coding folks like Claude, like Miguel. I know uses Claude and some other, but I don’t know. I don’t like.
309 00:37:17.950 ⇒ 00:37:20.730 Uttam Kumaran: I would be hard pressed if anyone said it was like.
310 00:37:21.200 ⇒ 00:37:24.140 Uttam Kumaran: Have to use Cloud, or it’s not gonna work, you know.
311 00:37:24.140 ⇒ 00:37:25.450 bencohen: Right, right.
312 00:37:26.930 ⇒ 00:37:29.370 Uttam Kumaran: And so yeah, I don’t know. It just sort of
313 00:37:29.970 ⇒ 00:37:33.500 Uttam Kumaran: yeah, all of our team. I got everyone clawed to use for stuff.
314 00:37:35.010 ⇒ 00:37:39.990 bencohen: That’s very nice I need. I actually had an 1130 that I didn’t realize I had. So
315 00:37:40.610 ⇒ 00:37:42.419 bencohen: so. But I think that we’ve got enough to.
316 00:37:42.420 ⇒ 00:37:46.709 Uttam Kumaran: I think we have enough. Yeah. So let me let me write some of this out. And then.
317 00:37:46.960 ⇒ 00:37:48.549 Uttam Kumaran: yeah, we should have some stuff
318 00:37:48.730 ⇒ 00:37:50.660 Uttam Kumaran: for you probably like early next week.
319 00:37:50.660 ⇒ 00:37:52.609 bencohen: Sweet, all right. Feel better, Otam.
320 00:37:52.610 ⇒ 00:37:54.399 Uttam Kumaran: Thank you so much. Appreciate it.
321 00:37:54.400 ⇒ 00:37:54.850 bencohen: Bye.
322 00:37:54.850 ⇒ 00:37:55.830 Uttam Kumaran: Yes, bye.