2025-12-08_uttam_-_gabe_-_bf_x_contextual_partnersh_8defe23e

Meeting Title: Uttam - Gabe - BF x Contextual Partnership - Insurance Platform Date: 2025-12-08 Meeting participants: Gabriel Lam, Uttam Kumaran

---

WEBVTT

1 00:05:07.820 ⇒ 00:05:09.030 Uttam Kumaran: Hello, sir.

2 00:05:09.880 ⇒ 00:05:11.260 Gabriel Lam: Hello. How are you.

3 00:05:11.430 ⇒ 00:05:12.390 Uttam Kumaran: How are you?

4 00:05:12.720 ⇒ 00:05:15.600 Gabriel Lam: I’m good, thanks for looking it up,

5 00:05:17.130 ⇒ 00:05:25.869 Gabriel Lam: Yeah, it’s been good, had a good weekend, mostly spent today trying to do a little spike on…

6 00:05:26.320 ⇒ 00:05:27.540 Gabriel Lam: are…

7 00:05:28.850 ⇒ 00:05:34.430 Gabriel Lam: Basically, the whole, like, trying to figure out how to get people to adopt the platform a little more, and what.

8 00:05:34.430 ⇒ 00:05:34.750 Uttam Kumaran: Okay.

9 00:05:34.750 ⇒ 00:05:38.290 Gabriel Lam: points are, because I feel like that’s a… that’s a sort of…

10 00:05:38.690 ⇒ 00:05:45.259 Gabriel Lam: not like a KPI, but something that I feel like we can try to start measuring or trying to get people to adopt.

11 00:05:45.470 ⇒ 00:05:49.190 Gabriel Lam: Yeah, and then…

12 00:05:49.190 ⇒ 00:05:54.629 Uttam Kumaran: what do you… what do you think it is? Like, I still feel like it’s just because some of the tools are, like.

13 00:05:57.740 ⇒ 00:06:03.220 Uttam Kumaran: Yeah. From the edges, like, I feel like we haven’t…

14 00:06:03.610 ⇒ 00:06:11.760 Uttam Kumaran: I mean, there is also, like, a… you need feedback, and you need more people… To set your roadmap.

15 00:06:12.530 ⇒ 00:06:30.400 Uttam Kumaran: You know, there’s a couple things in product management that they teach about this, like, you need, like, early adopters, basically, you know? Yeah. You need some people on the team who are, like, gonna be okay with the issues. There will be some people on the team that are not okay with the issues, and they’ll be like, okay, this doesn’t work, I’m not gonna use it.

16 00:06:32.370 ⇒ 00:06:38.880 Uttam Kumaran: But, like, you do have some people that are like that. It’s, like, almost like creating a forum for those people to set your roadmap, you know?

17 00:06:39.530 ⇒ 00:06:44.930 Gabriel Lam: Yeah, yeah, I think the main thing really has been the, sort of, client hub.

18 00:06:45.470 ⇒ 00:06:54.249 Gabriel Lam: experience, and how… in some ways, linear. The linear app itself is able to…

19 00:06:54.850 ⇒ 00:07:07.140 Gabriel Lam: at least seamlessly transition to tickets a little better. I think something that I hope to see is to have summaries from meetings be slowly shifting into a sort of live summary. So…

20 00:07:07.140 ⇒ 00:07:07.600 Uttam Kumaran: You know.

21 00:07:07.600 ⇒ 00:07:09.350 Gabriel Lam: Each client, you’ll have.

22 00:07:09.610 ⇒ 00:07:20.179 Gabriel Lam: all the follow-ups, all the action items. I just… my personal gripe with linear is you can get, like, ticket fatigue, where it’s just, like, there’s too many tickets going on. I have no idea, and I have to groom it.

23 00:07:21.090 ⇒ 00:07:21.820 Uttam Kumaran: Oh, totally.

24 00:07:21.820 ⇒ 00:07:38.189 Gabriel Lam: early adopter thing, I think, is definitely… I mean, I see myself as also an early adopter, and the rest of the AI team is people that I’ve been messing off with. I’ve talked to, like, Hannah, I’ve talked to, like, Rico, and I think Rico has been a good person to be like, hey, you know, these are the prompts that I’m using.

25 00:07:38.540 ⇒ 00:07:45.600 Gabriel Lam: like, he’s like, this is the prompt that I’m using for this, and I’m like, okay, what are things that work for you, what are things that don’t work, and how do they fit into our roadmap?

26 00:07:45.600 ⇒ 00:07:59.540 Uttam Kumaran: I mean, part of it, dude, is, like, you’re dealing with tough circumstance, like, you’re deal… you’re… the engineers on your team are, like, not… like, and this is where I play both sides, like, I… the folks that are working on your stuff, it’s not their number… it’s, like, their stated not-number-one priority.

27 00:07:59.540 ⇒ 00:08:00.220 Gabriel Lam: Yeah, no.

28 00:08:00.220 ⇒ 00:08:02.510 Uttam Kumaran: It was very, very tough, yeah. Like, so…

29 00:08:02.510 ⇒ 00:08:03.280 Gabriel Lam: It’s fine.

30 00:08:03.280 ⇒ 00:08:15.700 Uttam Kumaran: Which is fine, but what it forces you to do is, like, you have to really find how to innovate, and, you know, one of the things I would think about is, like, there is a clear path towards more budget for that team.

31 00:08:15.700 ⇒ 00:08:25.279 Uttam Kumaran: But it really has to come at the, okay, people are able to actually get work done faster in a measurable way. Like… Right.

32 00:08:25.310 ⇒ 00:08:36.109 Uttam Kumaran: And that’s the thing, you know, I think you may have talked to Clarence, and I talked to Clarence today about, like, look, if Gabe can prove to me that X task that used to take

33 00:08:36.169 ⇒ 00:08:37.580 Uttam Kumaran: 3 hours.

34 00:08:37.940 ⇒ 00:08:43.519 Uttam Kumaran: Or, like, nobody ever did it, is now able to be done in 30 minutes.

35 00:08:43.940 ⇒ 00:08:49.440 Uttam Kumaran: okay, I… I can do… I can do the math as the business owner on…

36 00:08:49.560 ⇒ 00:08:54.260 Uttam Kumaran: Okay, great, like, it means, like, for every additional hour that goes into this.

37 00:08:54.370 ⇒ 00:09:02.160 Uttam Kumaran: this is what the output is. And I don’t really need even that much data to show that, but that is ultimately what gets you from…

38 00:09:02.360 ⇒ 00:09:15.800 Uttam Kumaran: okay, I’m gonna try my best to get you 20% of everybody on AI’s time. They’re like, okay, actually, maybe I can prove to Robert and to the rest of the business that we should actually be spending money here, because here’s the through line to margin.

39 00:09:15.810 ⇒ 00:09:17.900 Gabriel Lam: You know, that’s just something to…

40 00:09:18.670 ⇒ 00:09:20.789 Uttam Kumaran: to think about. I think…

41 00:09:21.510 ⇒ 00:09:40.690 Uttam Kumaran: I… it is… it is tough, because I think we’ve… the bets… like, part of this is, like, I have to… I pushed on this team because I knew… I had a vision that, like, okay, even without proving this, it can make money now, but now it is a serious, like, cost center. It’s not, like, a side project like it was before.

42 00:09:40.690 ⇒ 00:09:41.619 Gabriel Lam: You know? Yeah, yeah.

43 00:09:41.620 ⇒ 00:09:50.939 Uttam Kumaran: Like, if you consider your true expense, you consider Sam, Mustafa, Casey, the cost of distraction, like, it is… it is a non, like.

44 00:09:51.070 ⇒ 00:10:04.349 Uttam Kumaran: there is an argument to be made of, like, okay, should we push this money here, or should I give this money to marketing, right? Like, what is the ROI? And so, yeah, it is… it is interesting. I think…

45 00:10:04.520 ⇒ 00:10:11.610 Uttam Kumaran: But I also think that, look, there’s, there’s also other things. There’s, like, opportunistic things. For example, I’m using…

46 00:10:11.700 ⇒ 00:10:30.620 Uttam Kumaran: I can… and I… I’m using, a mix of Kirscher with Claude to actually write, like, proposals and write meeting notes and summaries in literally, like, one-tenth the time that it used to take me. And that is something that I…

47 00:10:30.800 ⇒ 00:10:43.990 Uttam Kumaran: like, we need to productize and scale to people, and so part of… that’s something that, like, I just… as a… as a matter of, like, I just needed this… I need… we had so many proposals to write, I… there’s no way I could have done it.

48 00:10:44.240 ⇒ 00:10:51.190 Uttam Kumaran: The traditional way, and so I was like, fuck it, I’m just gonna throw all my transcripts into Claude, I’m gonna take a couple of the…

49 00:10:51.430 ⇒ 00:11:00.329 Uttam Kumaran: the proposals that Robert has generated that I liked, and then I’m just gonna use AI to help me do it, and it works! And we’ve won some of those!

50 00:11:00.330 ⇒ 00:11:00.700 Gabriel Lam: Yeah.

51 00:11:00.870 ⇒ 00:11:13.689 Uttam Kumaran: You know, and so that is… but that is a direct representation of, okay, initial proposal writing used to take, like, 2 or 3 hours. I did it in, like, 15 minutes.

52 00:11:13.690 ⇒ 00:11:14.659 Gabriel Lam: Yup. You know?

53 00:11:14.770 ⇒ 00:11:33.320 Uttam Kumaran: Okay, fair, like, now we can scale proposal writing. But also, part of this is, like, okay, but for this to get into everybody else’s hands, it needs to end up in the platform. Like, deck creation is another one where, like, there are clients for whom we are not creating decks, because our talent does not…

54 00:11:33.950 ⇒ 00:11:36.359 Uttam Kumaran: Have the experience.

55 00:11:36.600 ⇒ 00:11:54.469 Uttam Kumaran: or expertise in deck creation, which is a talent, like, that is a skill to create great decks. They don’t possess that. So what am I to do? I have to either, one, now pay more for our average analyst to go recruit people that do decks, or I need to get just people that do decks. Instead, if I’m like, Gabe.

56 00:11:54.470 ⇒ 00:11:55.890 Gabriel Lam: I’m about to…

57 00:11:55.890 ⇒ 00:11:58.949 Uttam Kumaran: Probably spend another 5 grand on talent.

58 00:11:59.180 ⇒ 00:12:03.940 Uttam Kumaran: just because we don’t have this deck expertise? Well, if you can make deck creation easier for me.

59 00:12:04.970 ⇒ 00:12:07.060 Uttam Kumaran: Okay, here’s 5 grand, here’s 5 grand.

60 00:12:07.060 ⇒ 00:12:07.599 Gabriel Lam: For sure.

61 00:12:08.000 ⇒ 00:12:18.149 Uttam Kumaran: Right? But see, that’s, like, that is the narrative, because that’s the narrative that I need to hear. But also, like, it’s really tough, like, so kind of a couple things. One, everybody that joins.

62 00:12:18.600 ⇒ 00:12:19.840 Uttam Kumaran: the company.

63 00:12:19.980 ⇒ 00:12:35.689 Uttam Kumaran: you have… I would say you have 10-20% of their time, but you may not be able to use every single one, but when it comes to data-related automations, totally, like, you see Demolade, OH, everybody’s down to give that sort of time, so I will be literally paying them to…

64 00:12:35.690 ⇒ 00:12:45.809 Uttam Kumaran: to help you wherever you need. Right now, there’s only Sam, Mustafa, Casey on the AI team. We’re gonna be adding one more person, Pranav. You can have… you can have his time, totally.

65 00:12:45.840 ⇒ 00:12:58.749 Uttam Kumaran: So, then it’s gonna be up to, like, you to figure out, okay, like, Surf is another really good example of, like, he’s someone that can help. So, as you get… as more people join, you’re gonna get a little bit of everybody’s bonus time, and then at some point.

66 00:12:58.750 ⇒ 00:13:06.369 Uttam Kumaran: it’s up to you to be like, hey, actually, instead of, like, everybody’s bonus time, can I just get 100% of somebody? Okay, like, let’s have that conversation.

67 00:13:06.500 ⇒ 00:13:15.130 Uttam Kumaran: You know, and that’s where, truly, it’s like… I don’t know, it gets more fun to think about, like, what does it look like if you were to have a full-time engineer

68 00:13:15.490 ⇒ 00:13:22.810 Uttam Kumaran: Versus, like, a couple of people, who would that be? Like, what, you know? That’s the stuff that’s, like… I think… I think we’re getting to the point where now that you have…

69 00:13:22.980 ⇒ 00:13:29.010 Uttam Kumaran: 10-20% of, like, 4 people’s time, you can say, okay, actually, let me just see, like.

70 00:13:29.110 ⇒ 00:13:32.140 Uttam Kumaran: What would it be like for me to find like…

71 00:13:32.570 ⇒ 00:13:40.060 Uttam Kumaran: someone that’s a little bit more senior than Casey to just work with you on the platform, maybe gives 50% of your time, or something like that, you know, so…

72 00:13:40.580 ⇒ 00:13:41.330 Uttam Kumaran: Yeah.

73 00:13:41.510 ⇒ 00:13:44.699 Uttam Kumaran: Trying. All of this only is possible to get more classes.

74 00:13:44.700 ⇒ 00:13:46.249 Gabriel Lam: Yeah, for sure, for sure.

75 00:13:47.200 ⇒ 00:13:51.239 Uttam Kumaran: Pranav, I can only get because we’re gonna have him working on some client work.

76 00:13:51.590 ⇒ 00:13:53.969 Uttam Kumaran: But the side effect is really, really awesome, so…

77 00:13:54.370 ⇒ 00:14:06.989 Gabriel Lam: Yeah, I think back to the contextual stuff, I guess I, like, sort of overshot, because I was totally assuming you wanted some sort of productized demo instead of just using the contextual platform.

78 00:14:06.990 ⇒ 00:14:23.469 Uttam Kumaran: Yeah, yeah, I’m, like, I’m, like, two steps behind you on that. Yeah, exactly, like, I think it’s… we’re just… because the reason why I gave it to you is because I’m like, now that you are familiar with, like, prompting, once you get into contextual, it’s actually a sick platform. I actually don’t think you’re gonna need

79 00:14:23.980 ⇒ 00:14:28.810 Uttam Kumaran: much more than just you and, like, maybe some of, like, Mustafa or Casey’s time.

80 00:14:28.840 ⇒ 00:14:43.770 Uttam Kumaran: to, like, help tune a couple things in contextual, but I think the entire demo can live there. Not because I don’t think we can build one, I think it’s actually just, like, I want you to churn… I want you to learn how it is to do it for insurance, and then we’re gonna churn a couple more of these out.

81 00:14:43.770 ⇒ 00:14:50.699 Uttam Kumaran: For a few other industries, and the complete demo is gonna live in contextual, basically, I feel like.

82 00:14:51.210 ⇒ 00:15:02.849 Gabriel Lam: Okay. Yeah, I think… I was testing it out, I just wasn’t sure where our existing agents live, and I don’t want to, you know, put a ton of PRs or…

83 00:15:02.980 ⇒ 00:15:03.809 Gabriel Lam: Push a bunch of stuff.

84 00:15:03.810 ⇒ 00:15:05.800 Uttam Kumaran: No, so it’s all within the UI.

85 00:15:07.010 ⇒ 00:15:07.700 Gabriel Lam: Okay.

86 00:15:08.470 ⇒ 00:15:23.470 Uttam Kumaran: Yeah, so it’s all with… so it’s all within the contextual UI. You’ll see there’s a place called Agents, but all Agents is, is, like, it’s a mix of, like, a prompt and some type of data store. So you’ll… as you’ll… you should go through, like, some of the contextual docs, and I don’t know if they have a demo.

87 00:15:23.580 ⇒ 00:15:28.069 Uttam Kumaran: But you’re basically just, like, creating, like, an agent in there.

88 00:15:28.300 ⇒ 00:15:39.070 Uttam Kumaran: And so part of the demo is for us to literally, like, narrate, okay, like, let’s say we have 10 minutes for a demo, what is, like, what is a script for that salesperson, you know, to go through?

89 00:15:40.740 ⇒ 00:15:42.069 Gabriel Lam: Okay. And…

90 00:15:42.070 ⇒ 00:15:43.220 Uttam Kumaran: s… yeah.

91 00:15:43.220 ⇒ 00:15:49.679 Gabriel Lam: Yeah, I think… I think the way I was imagining it through Contextual was, like, you know, Ian or anyone would…

92 00:15:50.390 ⇒ 00:15:57.930 Gabriel Lam: Basically upload, you know, all their documents, whatever it might be, and either record

93 00:15:58.500 ⇒ 00:16:04.579 Gabriel Lam: a conversation or plug in a transcript, right? And, like, that sort of is the demo of how do we get…

94 00:16:04.580 ⇒ 00:16:07.700 Uttam Kumaran: I think about two demos. I think, think about one demo in where…

95 00:16:08.070 ⇒ 00:16:14.220 Uttam Kumaran: it’s just, like, on the call. Let’s say I just call broker A. Hey, can I get you 30 minutes of your time? I want to show you the demo.

96 00:16:14.450 ⇒ 00:16:17.679 Uttam Kumaran: I don’t get anything in advance, so I have to create something that’s, like.

97 00:16:18.280 ⇒ 00:16:31.810 Uttam Kumaran: something a little bit more generic, where maybe we have some sample supplemental applications on our side, I pull up the demo and screen share, then think about, like, okay, what is maybe, like, a more involved demo where I get something from them, I say, hey.

98 00:16:31.930 ⇒ 00:16:36.910 Uttam Kumaran: what if I turned this around and made you a specific demo just for your company? Would that be interesting?

99 00:16:37.200 ⇒ 00:16:39.840 Uttam Kumaran: Think about… maybe split those two up, because…

100 00:16:40.340 ⇒ 00:16:46.690 Uttam Kumaran: like, first just start with, like, okay, this is something that you can just put in front of, like, if I was to just go meet someone in insurance tonight.

101 00:16:46.850 ⇒ 00:16:48.800 Uttam Kumaran: And try to put something in front of them.

102 00:16:48.930 ⇒ 00:16:50.710 Uttam Kumaran: That’s the first demo.

103 00:16:51.190 ⇒ 00:16:51.700 Gabriel Lam: Yep.

104 00:16:52.730 ⇒ 00:17:11.909 Uttam Kumaran: And then second could be, hey, like, why don’t you go ahead and send me some prospective documents, maybe send me a meeting transcript or two. How about I’ll turn around in a couple days, we can talk again, and I’ll show it to you how it would work on your documents. Better yet would be, like, we all do the whole thing live, but I don’t mind if even the first version of the demo is, like.

105 00:17:12.190 ⇒ 00:17:28.959 Uttam Kumaran: it’s sort of, like, scripted just to us, you know? And you’ll see the contextual platform is very visual. Like, you’ll see that… and, you know, what it’ll actually show… what I want you to see also is, like, what’s actually possible with RAG right now. Like, we are not using enough

106 00:17:28.960 ⇒ 00:17:41.150 Uttam Kumaran: document intelligence-related stuff, you’ll see that the product is actually so sick. It’ll literally draw a box on, like, where exactly it took, like, part of a diagram from. Like, it’s very, very powerful.

107 00:17:41.270 ⇒ 00:17:42.230 Uttam Kumaran: So…

108 00:17:42.510 ⇒ 00:17:51.620 Uttam Kumaran: I… part of the reason I want to keep it in contextual is because contextual, all it is, is a series of APIs, and some visual elements. It’s, like, entirely a platform.

109 00:17:51.670 ⇒ 00:18:05.709 Uttam Kumaran: They created a simple version of using it, but we could just go ahead and use it all there, because we’re going to think about one for legal, one for insurance, one for real estate, and it’ll all just live in contextual.

110 00:18:05.790 ⇒ 00:18:15.200 Uttam Kumaran: Which I think will be… will be nice. And they actually… this is also a thing, if we label the type of demo we want, they said they could dedicate some resources to help us build it, too.

111 00:18:15.200 ⇒ 00:18:17.470 Gabriel Lam: So that’s part of the reason I want…

112 00:18:17.470 ⇒ 00:18:19.730 Uttam Kumaran: it’s all living contextually, because I want…

113 00:18:19.860 ⇒ 00:18:23.530 Uttam Kumaran: To use some of their, you know, muscle to help us build this.

114 00:18:23.530 ⇒ 00:18:24.620 Gabriel Lam: Okay.

115 00:18:24.620 ⇒ 00:18:31.620 Uttam Kumaran: Because they’ll see it and be like, oh, we could totally do this, like, other thing you guys didn’t highlight, or whatever. I’m like, okay, okay, cool. You know, because they’re the experts.

116 00:18:31.620 ⇒ 00:18:32.230 Gabriel Lam: Yeah.

117 00:18:32.480 ⇒ 00:18:35.520 Gabriel Lam: I guess another question that I had would be.

118 00:18:35.630 ⇒ 00:18:40.979 Gabriel Lam: I think I reached out to ask if there’s any way to contact Ian, because I know he was on the call.

119 00:18:40.980 ⇒ 00:18:46.819 Uttam Kumaran: Yes, so I can send him… I can send him both his phone number, and I sent his email into Slack.

120 00:18:46.820 ⇒ 00:18:47.980 Gabriel Lam: Yeah.

121 00:18:47.980 ⇒ 00:19:00.849 Uttam Kumaran: Yeah, he’s, like, Ian’s, like, a really, really close friend of mine, actually, so… here in Austin, so… Awesome. But he’s super, super passionate about this. He’s the one that actually sent me that company outmarket, and sent me a demo of it. I think that’s a good.

122 00:19:00.850 ⇒ 00:19:01.580 Gabriel Lam: See that.

123 00:19:01.580 ⇒ 00:19:04.709 Uttam Kumaran: Yeah, I think that’s a good product to look at. It’s basically…

124 00:19:04.830 ⇒ 00:19:10.879 Uttam Kumaran: They had several modules. I almost… I’m hoping that you look at that, you’re like, oh shit, well, let’s just build…

125 00:19:11.030 ⇒ 00:19:17.719 Uttam Kumaran: let’s just build 3 of the most important modules and just copy what these guys did. And that’s, like, what it ends up. I’m like, perfect.

126 00:19:17.730 ⇒ 00:19:20.969 Gabriel Lam: But I’m pretty sure that that product demo.

127 00:19:20.970 ⇒ 00:19:29.940 Uttam Kumaran: And the documentation of that product will, like, help you accelerate, because you’ll see the product decisions they made. They basically built, like.

128 00:19:30.210 ⇒ 00:19:43.220 Uttam Kumaran: 20 or 30 of, like, these, like, piecemeal modules around insurance brokerages, like, filling out supplemental applications, like, translating a transcript to this. They, like, built a bunch of those. But they’re… but they’re selling, like, software.

129 00:19:43.570 ⇒ 00:19:54.810 Uttam Kumaran: Right. So our difference is that we are showing, like, what we’re capable of, powered by a platform like Contextual. All of our stuff, though, is built bespoke.

130 00:19:55.030 ⇒ 00:20:09.499 Uttam Kumaran: Like, at this moment, we’re not doing, like, a managed service, we’re also not selling software. So, Contextual is a tool that powers our ultimate solution. Our solution is tailored towards each broker, each brokerage, you know.

131 00:20:09.830 ⇒ 00:20:10.390 Gabriel Lam: Right.

132 00:20:11.680 ⇒ 00:20:17.599 Gabriel Lam: Okay. That sounds great. I’d love to take a look at the demos, and then I…

133 00:20:17.600 ⇒ 00:20:18.150 Uttam Kumaran: Yeah, can you…

134 00:20:18.150 ⇒ 00:20:20.679 Gabriel Lam: I tried to reach out to you. Did I send you that?

135 00:20:21.300 ⇒ 00:20:26.769 Gabriel Lam: I… I think you sent the homepage?

136 00:20:27.600 ⇒ 00:20:33.059 Gabriel Lam: And I scoured through it. They’re a little secretive about what exactly it looks like, and I think they really.

137 00:20:33.060 ⇒ 00:20:38.040 Uttam Kumaran: Oh, okay, here, let me, he, he, he emailed it to me, let me get it for you, hold on one second.

138 00:20:38.350 ⇒ 00:20:38.850 Gabriel Lam: Yep.

139 00:20:42.970 ⇒ 00:20:44.699 Uttam Kumaran: Oh, I, emailed it to you.

140 00:20:45.160 ⇒ 00:20:48.649 Gabriel Lam: Oh, okay, I might have lost it in the…

141 00:20:49.050 ⇒ 00:20:53.630 Uttam Kumaran: Check it out here, I’ll send it to Kelsey in the thread here. But, Ian is on that email.

142 00:20:56.630 ⇒ 00:20:58.999 Gabriel Lam: Oh, got it, yep, I see it.

143 00:21:00.380 ⇒ 00:21:05.180 Gabriel Lam: Oh, I see it. On, on grain, okay. I missed this, on Saturday.

144 00:21:07.560 ⇒ 00:21:09.160 Gabriel Lam: Okay, awesome.

145 00:21:10.510 ⇒ 00:21:12.060 Gabriel Lam: Let me just start.

146 00:21:13.570 ⇒ 00:21:15.470 Uttam Kumaran: Yeah, and let me put it in the,

147 00:21:17.880 ⇒ 00:21:23.770 Uttam Kumaran: The other thing, dude, is, like, how, okay, so actually, I’ll… sort of, like, go on a tangent. Let’s.

148 00:21:23.770 ⇒ 00:21:24.990 Gabriel Lam: Yeah, you’re good.

149 00:21:24.990 ⇒ 00:21:34.259 Uttam Kumaran: So… sorry, I was gonna completely change the subject, actually. Okay, so I just sent the recording in the thread. I left some comments in the Notion. Yeah, basically, I was like.

150 00:21:34.440 ⇒ 00:21:39.100 Uttam Kumaran: Focus on just a couple of, like, end-to-end workflows. Ideally.

151 00:21:39.450 ⇒ 00:21:46.569 Uttam Kumaran: a couple of them we can demo just in the contextual platform in, like, less than 10 minutes. I think that’s a fair constraint.

152 00:21:47.910 ⇒ 00:21:56.709 Uttam Kumaran: And… I didn’t get a chance to look at the PRD, but I feel like… the magic patterns, but I feel like… like, don’t worry too much about the UI.

153 00:21:56.870 ⇒ 00:22:04.129 Uttam Kumaran: If you develop something, Ian will totally hop on and help you, kind of, like, configure it.

154 00:22:06.030 ⇒ 00:22:10.549 Uttam Kumaran: Yeah, I just sent you the video, so I’ll also probably… I haven’t watched it yet, I’ll probably watch it later tonight as well.

155 00:22:10.550 ⇒ 00:22:12.690 Gabriel Lam: Okay. Sounds great.

156 00:22:17.160 ⇒ 00:22:22.880 Uttam Kumaran: So, basically, the comments I left was, like, I think there’s a couple of different workflows. There’s one, there’s, like.

157 00:22:23.070 ⇒ 00:22:27.130 Uttam Kumaran: transcript to… Like, filling a document out?

158 00:22:28.120 ⇒ 00:22:31.780 Uttam Kumaran: there’s, like, probably something around…

159 00:22:32.440 ⇒ 00:22:40.770 Uttam Kumaran: A new document comes in, like, a new supplemental application comes in, and you need to fill that out, and you just have, like, all these existing documents you’ve already worked on?

160 00:22:40.770 ⇒ 00:22:43.199 Gabriel Lam: Your pre-existing contacts just automatically fills.

161 00:22:43.200 ⇒ 00:22:58.379 Uttam Kumaran: Yeah, so, like, can you automatically fill in, or if you… if you… if you can’t, can you at least give me the interview? Can you draft me an email of what other information I need to get, right? So, there’s, like, that. There’s also, like, chat over my client’s context.

162 00:22:58.440 ⇒ 00:23:04.609 Uttam Kumaran: Right? Like, this is something that I think Ian could probably help you with, is, like, what are the common questions that you ask your

163 00:23:04.820 ⇒ 00:23:08.170 Uttam Kumaran: Like, they have, like, people underneath them, like, their sales…

164 00:23:08.530 ⇒ 00:23:15.200 Uttam Kumaran: assistants or whatever, who’s like, hey, go check, like, this thing for this client, and that person combs through PDFs and shit.

165 00:23:15.560 ⇒ 00:23:16.020 Gabriel Lam: That’s something…

166 00:23:16.020 ⇒ 00:23:29.910 Uttam Kumaran: what are some of those questions? And, like, are there any questions that, like, we can just throw into contextual, where it’s like, tell me… I mean, some of these questions could be, like, tell me what the liability risk is for the last few years across, like.

167 00:23:29.910 ⇒ 00:23:37.490 Uttam Kumaran: this company’s policies. Tell me, like, any open, like, hey, this company is going up for sale.

168 00:23:37.670 ⇒ 00:23:55.830 Uttam Kumaran: anything I should flag in their policies as a preparation. You know, like, those types of meaty questions that someone will be like, damn, yeah, that’ll probably take, like, 10 hours for my guy to do, and what you’ll see is Contextual will pull out the parts of the document and reference back to it. It’s, like, really, really cool.

169 00:23:55.880 ⇒ 00:24:02.720 Uttam Kumaran: I think the biggest challenge for you is gonna be working with Ian to find out, like, what are these, like, dummy documents that you may need? Right. Yeah.

170 00:24:03.020 ⇒ 00:24:14.329 Uttam Kumaran: And I would honestly suggest using AI to create some of them, or using AI to, like, create a fictitious… I mean, you could use Brainforge, I don’t care. Creating a fictitious company.

171 00:24:14.550 ⇒ 00:24:19.080 Uttam Kumaran: That, like, needs insurance, or here’s… Brainforge AI is a client of yours.

172 00:24:19.430 ⇒ 00:24:30.970 Uttam Kumaran: loaded into this area, we have 3 years of policies, we have our last 20 meetings with them, like, and you can use AI to generate synthetic versions of all these things if you want.

173 00:24:31.030 ⇒ 00:24:43.690 Uttam Kumaran: I think in particular, I’m trying to see… like, if you saw that meeting with Contextual, I have not seen their, like, document filling thing before. So I wonder, like, if that’s something that they can help you build, or if that comes out of the box?

174 00:24:43.860 ⇒ 00:25:00.309 Uttam Kumaran: Contextual also has a shit ton of parameters. Like, part of the reason I wanted to use it is because building RAG systems, to get, like, a demo going is easy, but to, like, nail it every single time is really hard, because there’s tons and tons of parameters to retrieval.

175 00:25:00.400 ⇒ 00:25:16.619 Uttam Kumaran: like, to rag. And so Contextual makes that… a lot of that, like, programmatic, where you don’t have to, like… they just give you, like, 50 different things you can configure. So this is where I think leveraging the help of their team is gonna help you build the best demo.

176 00:25:16.750 ⇒ 00:25:25.309 Uttam Kumaran: But where their… but I don’t think their team is gonna bring the insurance context. Right. So, Raj on their team is gonna be the best person.

177 00:25:25.520 ⇒ 00:25:31.539 Uttam Kumaran: Honestly, find a way to see how much of their time you can get for this, to be honest.

178 00:25:31.540 ⇒ 00:25:32.220 Gabriel Lam: For sure.

179 00:25:32.870 ⇒ 00:25:38.849 Uttam Kumaran: And then keep in the back of your head, also, that we’re gonna be scaling this to insurance, and to…

180 00:25:38.970 ⇒ 00:25:41.009 Uttam Kumaran: Real estate as well.

181 00:25:41.540 ⇒ 00:25:41.910 Gabriel Lam: Yes.

182 00:25:41.910 ⇒ 00:25:45.170 Uttam Kumaran: So ideally, like, keep an understanding of, like, hey.

183 00:25:45.360 ⇒ 00:25:56.119 Uttam Kumaran: what does Raj need to help you put together a great demo? Like, and yeah, like, kind of when you… if you give me the call, like, I have a couple of other demo ideas that we’ll put together for those groups.

184 00:25:56.470 ⇒ 00:26:11.660 Uttam Kumaran: Basically, what I’m hoping for is two things. One, we’re gonna go to net new people that don’t know Brave Forge. I’m also gonna circle back with some people, of which I may already have discovery documents, or maybe I ask, like, hey, go ahead and provide me with these documents, we’ll put together a demo.

185 00:26:11.760 ⇒ 00:26:29.320 Uttam Kumaran: So those are some of the ways that this can get more tailored towards a sales opportunity, but dude, even something just that’s visual, that we can put behind a loom, or I can show in a meeting, is, like, 10 steps ahead of what I’ve been doing, which is just talking out of my ass in most of these meetings. So, like…

186 00:26:29.320 ⇒ 00:26:29.640 Gabriel Lam: Okay.

187 00:26:29.640 ⇒ 00:26:31.090 Uttam Kumaran: It’s… yeah.

188 00:26:31.090 ⇒ 00:26:37.330 Gabriel Lam: Sounds great. Yeah, I’ll… I’ll dive right in. I think… I think that’s super interesting. I think it’s also, like.

189 00:26:39.200 ⇒ 00:26:44.860 Gabriel Lam: you know, in some ways, Ian and insurance is a placeholder, right? It’s like, once you have that.

190 00:26:45.160 ⇒ 00:26:45.800 Uttam Kumaran: Yes.

191 00:26:45.800 ⇒ 00:26:46.370 Gabriel Lam: Done.

192 00:26:46.370 ⇒ 00:26:55.709 Uttam Kumaran: The more I want you to understand the primitives of, like, contextual, and everything around document intelligence. Yeah. So, what the… this is where, like.

193 00:26:56.030 ⇒ 00:26:58.639 Uttam Kumaran: from… now I’m breaking the sales mold.

194 00:26:58.730 ⇒ 00:27:17.730 Uttam Kumaran: there’s a lot of industries who… they spend a lot of money on situations that sound like taking some information from something and putting it into a document, taking stuff out of PDFs, right? Even those two use cases, there’s a shit ton of use… there’s a shit ton of industries where that is a majority of someone’s job.

195 00:27:17.730 ⇒ 00:27:18.550 Gabriel Lam: It is, yeah.

196 00:27:18.570 ⇒ 00:27:21.159 Uttam Kumaran: Our role is to find out

197 00:27:21.680 ⇒ 00:27:32.320 Uttam Kumaran: those ones that exist, find out who the decision maker is, if there’s budget, and if there’s urgency. And that’s how we qualify leads. I find out one is, like, is this a real problem?

198 00:27:32.410 ⇒ 00:27:40.299 Uttam Kumaran: Is there a budget, right? And is there urgency, right? And if those three are true, there’s a good reason that you should be working for Brainforge.

199 00:27:40.300 ⇒ 00:27:54.319 Uttam Kumaran: The other thing that… then, okay, you may say, okay, cool, there’s probably, like, 100 industries. The reason why I picked insurance is because it’s really hard to get into insurance without an Ian on your side, and we’re gonna co-sell a solution with Ian.

200 00:27:54.580 ⇒ 00:28:02.099 Uttam Kumaran: Like, I’m gonna… Ian is gonna open up his Rolodex and say, hey guys, I just started working with Brainforge, they have this dope solution.

201 00:28:02.160 ⇒ 00:28:17.489 Uttam Kumaran: like, can we grab dinner? Like, I’d love to show it with you. And he’s gonna open up his Rolodex. Yeah. And legal as well, it’s like a real buddy-buddy thing. Real estate as well, it’s a real buddy-buddy thing. So I want to go into industries where the competition is low, because without that sort of sponsor, you can’t get in.

202 00:28:17.490 ⇒ 00:28:19.690 Gabriel Lam: Where are people gonna go first?

203 00:28:19.690 ⇒ 00:28:28.740 Uttam Kumaran: Some people are gonna try to get into those industries, but they’ll never get in, because you can’t cold email your way into those. It has to be, like, a boy, Ian’s boy from some other deal or something like that.

204 00:28:28.740 ⇒ 00:28:30.410 Gabriel Lam: Or, like, you’ve done… yeah, yeah.

205 00:28:30.410 ⇒ 00:28:39.959 Uttam Kumaran: Yeah, it’s gonna be a lot of that. Like, the use cases where you can… like, for example, you may say, oh, like, healthcare has a lot of… yeah, but healthcare’s hella regulation.

206 00:28:39.960 ⇒ 00:28:51.380 Uttam Kumaran: And companies like Epic and those guys are already, like, gonna start doing this other stuff. Like, in real estate, you have… in real estate and legal and insurance brokers, you have a lot of money at stake.

207 00:28:51.400 ⇒ 00:28:58.349 Uttam Kumaran: You also have a buyer that hates operations. Like, they hate how much money they’re spending on back-of-office stuff.

208 00:28:58.350 ⇒ 00:28:58.700 Gabriel Lam: Right.

209 00:28:58.700 ⇒ 00:28:59.230 Uttam Kumaran: All they want.

210 00:28:59.230 ⇒ 00:29:00.160 Gabriel Lam: I know every day.

211 00:29:00.160 ⇒ 00:29:13.040 Uttam Kumaran: Yes, these guys are pure sellers. They make money on the transaction, meaning the number of transactions and how fast you can do it is the thing that matters. They don’t care whether they have, like, anybody back office.

212 00:29:13.040 ⇒ 00:29:16.349 Gabriel Lam: Yeah. And so for me, when I come in, I focus on…

213 00:29:16.350 ⇒ 00:29:28.110 Uttam Kumaran: Automating back office to increase revenue. And we still think about, okay, our goal is to speed up the transaction, so that’s why you saw me talking to Ian, about how long does the average transaction take.

214 00:29:28.600 ⇒ 00:29:47.140 Uttam Kumaran: how many back and forths are there? It allows us to easily size what a reduction in that could be, and then I’m able to price towards that reduction, right? Towards that, okay, hey, we’re able to now do 20% more deals, right? You’re able to now, like, cut your time by X.

215 00:29:47.140 ⇒ 00:29:53.220 Uttam Kumaran: I want a piece of that, right? And, like, that’s how… that’s how I want to talk about our solution in the market.

216 00:29:53.350 ⇒ 00:29:55.880 Uttam Kumaran: Because if we go after just cost.

217 00:29:56.420 ⇒ 00:30:04.330 Uttam Kumaran: they’re not gonna pay us a lot, but if we’re saying, hey, this solution unlocks this amount of revenue, there’s money… there’s a lot more money there. And so.

218 00:30:04.430 ⇒ 00:30:08.400 Uttam Kumaran: That’s the only reason we’re going after these three, is because I happen to have some…

219 00:30:08.460 ⇒ 00:30:22.989 Uttam Kumaran: ends into these industries, where also the buyer is not being hit by a lot of solutions, because it’s very hard to get in. There’s not, like, a lot… like, for example, Outmarket is one of the few people in InsurTech that have, like, done something

220 00:30:22.990 ⇒ 00:30:29.059 Uttam Kumaran: That both me and Ian saw, and we’re like, okay, this is actually impressive. And we’ve been looking at this industry for, like, 2 years.

221 00:30:29.520 ⇒ 00:30:30.330 Uttam Kumaran: So…

222 00:30:30.720 ⇒ 00:30:38.480 Uttam Kumaran: That’s the thing, like, wherever we can get in where there’s, like, not a lot of great stuff already gives us a huge leg up. Yeah. Yeah.

223 00:30:39.670 ⇒ 00:30:40.520 Gabriel Lam: Okay.

224 00:30:41.080 ⇒ 00:30:44.760 Gabriel Lam: Yeah, I think… I think that clears up a lot about the whole…

225 00:30:45.060 ⇒ 00:30:50.459 Gabriel Lam: vision of the demo. I… I’ll reach out to Ian and try to schedule a time to go through.

226 00:30:50.460 ⇒ 00:30:54.989 Uttam Kumaran: Yeah, reach out to Ian, try to get time with Raj, he’s in the contextual channel.

227 00:30:54.990 ⇒ 00:30:55.650 Gabriel Lam: Yeah.

228 00:30:56.920 ⇒ 00:31:01.490 Uttam Kumaran: You know, those are good guys to know, like, I think we’re gonna try to develop a few demos with them.

229 00:31:01.630 ⇒ 00:31:14.580 Uttam Kumaran: they basically were like, dude, if you can develop a demo, we’ll put it on our, like, on our public-facing… we’ll start to scale the demo with you. And so I think we have a good chance, like, right now they do a lot in semiconductor, they do a lot in, like, biochem, and, like.

230 00:31:14.710 ⇒ 00:31:21.689 Uttam Kumaran: they don’t do a lot in some of these industries because, like, they’re not, like, they’re here to sell contextual the platform.

231 00:31:21.960 ⇒ 00:31:26.319 Uttam Kumaran: Like, they can’t go to an insurance company and sell contextual because there’s no one there to implement.

232 00:31:26.320 ⇒ 00:31:26.870 Gabriel Lam: Yep.

233 00:31:27.100 ⇒ 00:31:43.969 Uttam Kumaran: That’s also the reason I want to… I like these guys, because they’re like, dude, if you get us there, then yeah. Or if any time we get an insurance company that calls us, we’ll say, hey, you can buy us, but you have to go through these guys to get it implemented. So that’s the kind of, like, mutual, sort of win-win thing that we’re going for.

234 00:31:44.990 ⇒ 00:31:45.860 Gabriel Lam: Awesome.

235 00:31:46.600 ⇒ 00:31:49.100 Gabriel Lam: Okay, that sounds great.

236 00:31:49.100 ⇒ 00:31:53.250 Uttam Kumaran: Cool, dude. My other question was…

237 00:31:53.380 ⇒ 00:31:56.220 Uttam Kumaran: I think one thing to think about is…

238 00:31:57.150 ⇒ 00:32:01.690 Uttam Kumaran: I’m starting, like, like, let’s take that use case that I mentioned, which is, like.

239 00:32:01.920 ⇒ 00:32:05.510 Uttam Kumaran: I’m… I’m starting to write… SOWs.

240 00:32:05.590 ⇒ 00:32:13.570 Uttam Kumaran: using Claude a lot. Like, what’s the best way you can take someone like me that is, like, a super user of AI,

241 00:32:13.600 ⇒ 00:32:27.160 Uttam Kumaran: And I want to scale this to the platform, because I want Robert to… because Robert’s not gonna do the shit I’m doing in Cursor, but it’s just, like, it’s just kind of complicated. But I also want, like… I want Rico to be able to write the first SOW, so, like.

242 00:32:27.640 ⇒ 00:32:30.939 Uttam Kumaran: How… what… even just, like, riff with me, like, how do you think…

243 00:32:31.260 ⇒ 00:32:37.749 Uttam Kumaran: I could take some type of workflow like that and maybe, like, get it to the platform. Like, what would be some of the first things to consider?

244 00:32:39.640 ⇒ 00:32:46.560 Gabriel Lam: I think… Off the top of my head, the first thing is, like, there’s…

245 00:32:46.730 ⇒ 00:32:57.579 Gabriel Lam: there’s the, I need to know this possibility exists, and I want to see how someone else does it. So I can be like, oh, you know, I see how Utam, like, takes

246 00:32:57.810 ⇒ 00:33:06.320 Gabriel Lam: this piece of information, and then brings it into Cloud Code and Cursor, and then it spits that out, and then I take it from there. I think just for…

247 00:33:06.320 ⇒ 00:33:06.990 Uttam Kumaran: Same balloon.

248 00:33:06.990 ⇒ 00:33:18.399 Gabriel Lam: to, like, see it happen, and be like, oh, if I take, you know, whatever ops meeting I have, and then I bring it and basically, like, copy-paste his workflow, what do I get from that? I think that’s one.

249 00:33:18.400 ⇒ 00:33:19.040 Uttam Kumaran: Yeah, yeah.

250 00:33:19.040 ⇒ 00:33:21.409 Gabriel Lam: The other thing might be…

251 00:33:21.630 ⇒ 00:33:32.239 Gabriel Lam: whether it’s, like, a loom, whether it’s a guideline, or, like, a one-pager, I think the setup is gonna be more annoying, in the sense that

252 00:33:32.700 ⇒ 00:33:41.200 Gabriel Lam: like… If you’re not technical, or you’ve never really dealt in… in code land, you’re like, oh, you know.

253 00:33:41.310 ⇒ 00:33:54.719 Gabriel Lam: what’s this repository? What do I… what does it mean by cloning? What’s the jargon that I… that… that, you know, we’re expecting to use? How do I push? You know, even, like, little things that… that once you have done it before, you’re like, oh, of course I know what that means, I think, to someone who

254 00:33:55.160 ⇒ 00:33:55.710 Gabriel Lam: is…

255 00:33:55.710 ⇒ 00:33:59.970 Uttam Kumaran: So, like, so, like, give me a sense of, like, in this situation.

256 00:34:00.120 ⇒ 00:34:05.050 Uttam Kumaran: About a week ago, I was like, I’m fucked, I need to write, like, 5 proposals.

257 00:34:05.220 ⇒ 00:34:14.409 Uttam Kumaran: I’m gonna just, like, figure out how to do this in cursor. And let’s say I figured it out. What would you have rather… like, at that point, what should I have done? Should I just, like, record a loom and, like.

258 00:34:14.610 ⇒ 00:34:17.369 Uttam Kumaran: Here’s how I’m doing this, step-by-step, basically.

259 00:34:17.880 ⇒ 00:34:23.880 Gabriel Lam: I feel like that’s the best first step of, like, hey, I wrote the last 5 proposals, this is what I did.

260 00:34:23.889 ⇒ 00:34:24.229 Uttam Kumaran: Okay.

261 00:34:24.230 ⇒ 00:34:24.870 Gabriel Lam: it.

262 00:34:25.440 ⇒ 00:34:28.020 Uttam Kumaran: And I should just send it in, like, the Brainforce team channel?

263 00:34:28.020 ⇒ 00:34:28.340 Gabriel Lam: Sorry.

264 00:34:28.340 ⇒ 00:34:32.150 Uttam Kumaran: I’m being, like, very, very dumb here, but, like, because I do this shit every day, like.

265 00:34:32.530 ⇒ 00:34:37.039 Uttam Kumaran: like, I’m really honest when I tell you, like, the only reason this business exists

266 00:34:37.370 ⇒ 00:34:54.650 Uttam Kumaran: like, am I doing it for this much money is because we used AI everywhere. So, like, I find… like, I just used to be a really good cheater at school. Like, I just find a way through it, and here it is, like, helping me, because I get very… I’m really allergic to doing the same thing over and over again. Like, it’s… it’s getting harder and harder for me. For example, like.

267 00:34:54.650 ⇒ 00:34:55.020 Gabriel Lam: Yeah.

268 00:34:55.020 ⇒ 00:34:59.350 Uttam Kumaran: A lot of our client work is stuff I used to do when I first graduated, dude, and so…

269 00:34:59.690 ⇒ 00:35:05.990 Uttam Kumaran: I can, like, do it with my eyes closed, but, like, I can’t… it’s so hard for me to focus, like, and so I just find…

270 00:35:06.170 ⇒ 00:35:16.970 Uttam Kumaran: my way to use AI to do it, but so you’re saying, okay, just record it end-to-end at that moment, send it to the team channel, being like, hey, and just explain why I did the thing I did.

271 00:35:17.300 ⇒ 00:35:25.200 Gabriel Lam: Yeah, or like, let’s say you say you write, like, you know, you wrote 5 yesterday, like, maybe on the fourth one, you’re like, I think I got it down pretty good, like, this is what I do.

272 00:35:25.360 ⇒ 00:35:32.760 Gabriel Lam: And, you know, like, Rico, I know you have, like, 3 proposals that you need to get out, like, you can do this too, and save yourself.

273 00:35:32.760 ⇒ 00:35:33.210 Uttam Kumaran: Okay.

274 00:35:33.210 ⇒ 00:35:34.340 Gabriel Lam: You know, two hours.

275 00:35:34.340 ⇒ 00:35:39.419 Uttam Kumaran: But so… so let me give you… so, but the… let’s say the GitHub thing is a little bit complicated, because you have to, like.

276 00:35:39.670 ⇒ 00:35:41.510 Uttam Kumaran: Oh, you have to know cursor?

277 00:35:41.660 ⇒ 00:35:46.159 Uttam Kumaran: You kind of have to, like… Install some extensions, like…

278 00:35:46.820 ⇒ 00:35:53.660 Uttam Kumaran: at that point, let’s say it’s like, hey, I just… this needs to get into, like, the platform somehow.

279 00:35:53.830 ⇒ 00:35:55.360 Uttam Kumaran: By, like, the end of the week.

280 00:35:55.610 ⇒ 00:35:56.440 Uttam Kumaran: like…

281 00:35:56.850 ⇒ 00:36:03.269 Uttam Kumaran: what do you think… what do you… what do you think you need from me at that point? If I was to be like, hey, this is the end-to-end loom, but, like.

282 00:36:03.690 ⇒ 00:36:16.899 Uttam Kumaran: it’s… it’s gonna be complicated for anyone who’s non-technical to, like, do this. Like, Robert’s gonna look at it and be like, yo, that’s sick, and, like, I could use that, but, like, unless you set it up for me, I’m not gonna be able to do it. So, like, what do you… what do you think in that moment?

283 00:36:22.110 ⇒ 00:36:23.750 Gabriel Lam: Well, I think it… hmm…

284 00:36:24.530 ⇒ 00:36:25.400 Uttam Kumaran: Yeah.

285 00:36:25.570 ⇒ 00:36:32.529 Gabriel Lam: I wonder if there’s, like, a, you know, explain, like, on 5, step by step, which is sometimes what I use. Thank you.

286 00:36:32.740 ⇒ 00:36:39.620 Gabriel Lam: ChatGPT4, I’m just like, hey, like, just list out every step in its dumbest form.

287 00:36:39.620 ⇒ 00:36:40.120 Uttam Kumaran: Yeah.

288 00:36:40.130 ⇒ 00:36:42.220 Gabriel Lam: And, like, this is how you set it up.

289 00:36:42.220 ⇒ 00:36:44.179 Uttam Kumaran: But let’s say I gave you that loom.

290 00:36:44.320 ⇒ 00:36:55.209 Uttam Kumaran: do you think you could be like, okay, I… like, one thing… I was hoping you’re like, okay, I’ll put that into AI and say, like, what is the literally, like, minimal viable product?

291 00:36:55.560 ⇒ 00:36:58.529 Uttam Kumaran: that given our AI team’s stack.

292 00:36:58.650 ⇒ 00:37:01.160 Uttam Kumaran: We could ship, like, a version of this by Friday.

293 00:37:01.420 ⇒ 00:37:10.739 Uttam Kumaran: Because, dude, like, to tell you the truth, that… all that is, is, like, you have to build some type of simple… I mean, you don’t even need to build the upload functionality, frankly. All you need to do is say, like.

294 00:37:11.040 ⇒ 00:37:14.840 Uttam Kumaran: Okay, if you take all your transcripts and upload it into this Google Drive.

295 00:37:15.320 ⇒ 00:37:19.670 Uttam Kumaran: and click on these couple buttons, then we can enable Sonnet over…

296 00:37:20.240 ⇒ 00:37:34.470 Uttam Kumaran: that document set. And the reason why I use Sana is because it’s just good at writing SOWs. I literally put in the SOW prompt, I write it. Like, that’s… I’m kind of like, okay, what… what’s the next step between, like… because I just don’t… dude, I just really don’t think…

297 00:37:34.680 ⇒ 00:37:37.280 Uttam Kumaran: Unless we make it… and this is just, like, a personal…

298 00:37:38.130 ⇒ 00:37:44.479 Uttam Kumaran: opinion is that I don’t think unless we make it so easy, that people are going to do what I’m doing, which is, like.

299 00:37:44.880 ⇒ 00:37:50.009 Uttam Kumaran: trying to innovate in cursor, but, like, I think there’s maybe, like, 2 or 3 people are gonna do that.

300 00:37:50.160 ⇒ 00:37:51.090 Uttam Kumaran: You know?

301 00:37:51.320 ⇒ 00:38:10.519 Uttam Kumaran: Like, I think the usage you are seeing is because the platform is actually, in many ways, very, very easy, and you can already see that even though it’s that easy, people aren’t fucking using it. Yeah, yeah. So it’s like, I don’t think that… I think people are gonna give me, like, a nice emoji, and be like, oh yeah, that’s, like, so cool.

302 00:38:10.630 ⇒ 00:38:14.970 Uttam Kumaran: But the odds that at that moment, someone who doesn’t actively code

303 00:38:15.530 ⇒ 00:38:19.550 Uttam Kumaran: downloads cursor. Like, it took me a month to learn Cursor.

304 00:38:20.620 ⇒ 00:38:32.319 Gabriel Lam: So that’s kind of, like, my… what I would kind of, like, push back on. Like, I don’t know whether it happens, even if I were to do that, you know? Yeah. I think my… my only feedback on bringing things to the platform is, like.

305 00:38:32.500 ⇒ 00:38:36.049 Gabriel Lam: We… It needs to be good enough.

306 00:38:37.830 ⇒ 00:38:42.709 Gabriel Lam: I think, like, literally what you said, it needs to be good enough or easy enough for people to be like, oh.

307 00:38:43.890 ⇒ 00:38:48.619 Gabriel Lam: that… makes sense, I’ll install cursor, I’ll… or…

308 00:38:48.780 ⇒ 00:38:52.779 Gabriel Lam: Maybe for us, it’s like, okay, these are all… this is on the roadmap for us to add.

309 00:38:52.780 ⇒ 00:38:55.410 Uttam Kumaran: I mean, the alternative thing is I could…

310 00:38:55.520 ⇒ 00:39:08.359 Uttam Kumaran: just be like, hey, everybody, if you’re driving proposals, I’m gonna do, like, a little, like, how to… I’m just gonna walk you through how to install Cursor, and, like, I’m gonna walk you through, like, how to do the… literally, like, the end-to-end.

311 00:39:08.750 ⇒ 00:39:09.580 Uttam Kumaran: You know?

312 00:39:09.830 ⇒ 00:39:11.119 Uttam Kumaran: Maybe that’s it.

313 00:39:12.370 ⇒ 00:39:22.299 Uttam Kumaran: And then for the people who… it’s tough… it’s tougher, they’ll just figure… they’ll just… they’ll just do it. Like, literally walk… I’m gonna uninstall Cursor on my machine and walk you through installing it.

314 00:39:22.420 ⇒ 00:39:41.789 Uttam Kumaran: opening… setting up GitHub, like, I guess that’s what I have to do, right? Because I don’t know. Because the other thing, I’m like, okay, maybe… this is where I’m like, okay, if Gabe can think about the platform as just, like, a series of endpoints, for example, there’s some use cases that are chat over documents. So we need… we need to enable some type of, like.

315 00:39:41.860 ⇒ 00:39:49.720 Uttam Kumaran: document… document storage… endpoint, where people can store stuff, and you can chat over it. Okay, seems like…

316 00:39:49.910 ⇒ 00:39:50.950 Uttam Kumaran: that’s, like.

317 00:39:51.670 ⇒ 00:40:08.339 Uttam Kumaran: like, some type of primitive, like, so that’s, like, one module. The second thing, okay, text. Or, like, chat window. Okay, we’re just gonna make sure that our chat window has all the necessary options to chat over something. Like, are you thinking about the platform and that, or are you thinking about it, like.

318 00:40:09.070 ⇒ 00:40:11.009 Uttam Kumaran: Use case by use case.

319 00:40:11.160 ⇒ 00:40:12.050 Uttam Kumaran: You know?

320 00:40:13.980 ⇒ 00:40:17.200 Uttam Kumaran: I mean, I don’t think it’s one or the other, right? But I think…

321 00:40:18.210 ⇒ 00:40:19.970 Gabriel Lam: I think people go…

322 00:40:20.030 ⇒ 00:40:39.660 Gabriel Lam: primarily… as a user, you go primarily for a use case. And so, like, a platform is like, I’m going there, like, because I need to write an SOP, and I’m not really thinking about, like, oh, I want AI to, like, look over, like, this set of documents and then that set of documents. I think on the technical side, like, yes, of course we have to think about that, because

323 00:40:40.770 ⇒ 00:40:43.660 Gabriel Lam: That’s scalable, and, like, when you do…

324 00:40:43.850 ⇒ 00:40:55.680 Gabriel Lam: you know, text over documents on this side, then you’re gonna be like, oh, how do we make documents? Or, like, chat over… chat window over text, and I’m like, okay, where… what kind of text? Is it, like, transcripts? Is it documentation? Is it…

325 00:40:55.810 ⇒ 00:40:58.149 Uttam Kumaran: I think that’s where it’s scalable.

326 00:40:58.480 ⇒ 00:41:03.179 Gabriel Lam: I think when you talk about MVPs, I think the has…

327 00:41:04.450 ⇒ 00:41:08.249 Gabriel Lam: I… I think of it coming from the use case side, mostly because.

328 00:41:08.450 ⇒ 00:41:08.999 Uttam Kumaran: That’s how you’re gonna.

329 00:41:09.000 ⇒ 00:41:15.969 Gabriel Lam: adoption. It’s just, like, something… like, for the first things, I just needed to do SOPs, for example. Yeah.

330 00:41:16.710 ⇒ 00:41:22.260 Gabriel Lam: I think when you talk about, like, doing a demo, I mean, maybe that might be helpful,

331 00:41:23.810 ⇒ 00:41:24.450 Gabriel Lam: I mean…

332 00:41:24.450 ⇒ 00:41:27.729 Uttam Kumaran: No, no, no, I hear you. I just think it’s like, okay, the…

333 00:41:27.870 ⇒ 00:41:31.999 Uttam Kumaran: the writing SOWs piece sounds like a mix of

334 00:41:32.350 ⇒ 00:41:41.019 Uttam Kumaran: I need to be able to choose, like, Sonnet, right? So you’d have to have some type of multi… you had to support multiple models. I need an ability to, like.

335 00:41:41.260 ⇒ 00:41:54.319 Uttam Kumaran: for example, what I did is I literally took, like, every granola from, like, an every, like, platform transcript from all my meetings with, like, a prospect, put it into Markdown files in a folder.

336 00:41:54.500 ⇒ 00:42:00.979 Uttam Kumaran: And then I… then I just put, like, a bunch of example… SOWs that I liked.

337 00:42:01.220 ⇒ 00:42:02.420 Uttam Kumaran: And then…

338 00:42:03.460 ⇒ 00:42:14.700 Uttam Kumaran: I was like, okay, just go ahead and write, like, the first version of this. And then I looked at the first version, I’m like, okay, pretty good, go ahead and change these aspects, like, for example, don’t put pricing in, because

339 00:42:15.410 ⇒ 00:42:29.779 Uttam Kumaran: I don’t know, like, we have to have them review the first version of this, or don’t put pricing in. Second, like, no emojis, like, you know, so what am I doing? I’m basically kind of like, that should all end up as, like, a fixed prompt, so it gets it, like, 90% right the first time, right?

340 00:42:31.630 ⇒ 00:42:39.260 Uttam Kumaran: So there’s, like, those types of things where I’m like, okay, well, ultimately, Would this have been, like…

341 00:42:46.940 ⇒ 00:42:48.229 Uttam Kumaran: So this has been some…

342 00:42:48.350 ⇒ 00:42:54.970 Uttam Kumaran: thing where, hey, I should have, like, went to the create uploads. I don’t know, I’m trying to, you know, I’m kind of thinking about, like.

343 00:42:55.250 ⇒ 00:43:00.890 Uttam Kumaran: How could we have made it easier for me to build that into… for, like, after my second…

344 00:43:02.060 ⇒ 00:43:05.449 Uttam Kumaran: 10 or third go-around, because I’m… I’m telling you, sir.

345 00:43:05.950 ⇒ 00:43:13.489 Uttam Kumaran: I’m just gonna go… I’m gonna go home, and I have to write a couple things, I’m gonna just keep using, because it’s sick. It works so, so well. It’s better than…

346 00:43:15.550 ⇒ 00:43:17.050 Uttam Kumaran: What eye guster?

347 00:43:19.890 ⇒ 00:43:34.849 Uttam Kumaran: Like, and we’re… we are winning this, like, nobody has pushed back, holding back. Every single client has said, wow, you’re… I can’t… thank you so… you put this together so fast, like, it means a lot. I can tell that you’re working on our behalf.

348 00:43:35.240 ⇒ 00:43:38.969 Uttam Kumaran: like… Come on, I’m just gonna keep doing that, right?

349 00:43:39.270 ⇒ 00:43:40.350 Gabriel Lam: Yeah.

350 00:43:40.350 ⇒ 00:43:40.950 Uttam Kumaran: Right?

351 00:43:46.460 ⇒ 00:43:47.150 Gabriel Lam: Yeah.

352 00:43:47.970 ⇒ 00:43:49.190 Uttam Kumaran: a form…

353 00:43:57.210 ⇒ 00:43:59.580 Gabriel Lam: Sorry, you cut off a little bit there.

354 00:43:59.820 ⇒ 00:44:01.590 Uttam Kumaran: I think we’re still maybe, like…

355 00:44:01.750 ⇒ 00:44:03.160 Uttam Kumaran: Can you… can you hear me?

356 00:44:03.460 ⇒ 00:44:04.629 Gabriel Lam: Yes, I hear you now.

357 00:44:07.200 ⇒ 00:44:11.290 Uttam Kumaran: I said that maybe it would have taken a month or two months, right, to have built

358 00:44:11.490 ⇒ 00:44:13.429 Uttam Kumaran: The feature-for-feature match.

359 00:44:13.590 ⇒ 00:44:14.210 Gabriel Lam: Yeah.

360 00:44:14.560 ⇒ 00:44:15.260 Uttam Kumaran: Right.

361 00:44:18.630 ⇒ 00:44:19.280 Gabriel Lam: Hmm.

362 00:44:19.280 ⇒ 00:44:19.910 Uttam Kumaran: Yeah.

363 00:44:23.260 ⇒ 00:44:26.139 Uttam Kumaran: Which is tough, which is like, okay, at that point, should I…

364 00:44:26.370 ⇒ 00:44:28.300 Uttam Kumaran: As a user, should I wait?

365 00:44:28.430 ⇒ 00:44:32.749 Uttam Kumaran: Or should I just, like, fuck it, I’m gonna rip cursor? I don’t know, I don’t know, I don’t know what the right answer is.

366 00:44:38.620 ⇒ 00:44:43.790 Gabriel Lam: I think if you have it, if you…

367 00:44:44.260 ⇒ 00:44:50.010 Gabriel Lam: I am of the camp that If you’ve got something working-ish.

368 00:44:50.320 ⇒ 00:44:52.550 Gabriel Lam: Like, don’t wait for the feature match.

369 00:44:52.820 ⇒ 00:44:56.650 Gabriel Lam: And then once it comes out, you’re like, great, I can just transition this whole workflow.

370 00:44:56.960 ⇒ 00:45:10.949 Gabriel Lam: But I… it comes from maybe a more malleable mindset of, like, hey, I… I’ve seen it work for me, I’m willing to put in the effort to, like, then migrate instead of the one and done.

371 00:45:11.960 ⇒ 00:45:13.429 Gabriel Lam: But I, I feel like…

372 00:45:14.230 ⇒ 00:45:17.090 Uttam Kumaran: I guess my… yeah. Yeah, go ahead, go, go ahead.

373 00:45:17.630 ⇒ 00:45:20.760 Gabriel Lam: I like the loom, just because… I mean…

374 00:45:20.930 ⇒ 00:45:25.710 Gabriel Lam: how do you get people… it might be on the Friday retros, where it’s like, hey, like…

375 00:45:25.860 ⇒ 00:45:28.290 Gabriel Lam: I use Cursor, this is how I use it.

376 00:45:28.680 ⇒ 00:45:35.139 Gabriel Lam: Anyone who writes proposals, like, you’re gonna wanna do this. And do a step-by-step there.

377 00:45:35.140 ⇒ 00:45:40.590 Uttam Kumaran: Well, I guess here… I guess here’s my… yeah, here’s my question. I, like, do you… do you think, like.

378 00:45:40.720 ⇒ 00:45:42.970 Uttam Kumaran: My use case, you should support.

379 00:45:43.270 ⇒ 00:45:53.309 Uttam Kumaran: Or should you say, hey, our platform is not enough of a platform right now for me to dedicate the limited resources? You should go ahead and keep doing that in cursor.

380 00:45:54.550 ⇒ 00:45:58.519 Gabriel Lam: I think… yeah, for sure. I… I don’t think the platform…

381 00:45:58.840 ⇒ 00:46:04.470 Gabriel Lam: Is at a point in which that…

382 00:46:07.640 ⇒ 00:46:08.660 Uttam Kumaran: And, like, we’re…

383 00:46:14.080 ⇒ 00:46:32.950 Gabriel Lam: I’ll have to think about it, because in some ways, a lot of the information is there, right? Like, you have all the tables, you have all the Zoom meeting transcripts, you have all the Slack messages, like, that all exists already. And so you have agents that can already go through it and are extracting information, extracting summaries, linear tickets, like, that’s all there.

384 00:46:33.250 ⇒ 00:46:35.500 Gabriel Lam: so…

385 00:46:37.930 ⇒ 00:46:44.089 Gabriel Lam: I think it’s… I think it’s possible. I think I am just trying to figure out what the…

386 00:46:45.430 ⇒ 00:46:51.570 Gabriel Lam: lift on that will look like, and is that going to be an easier flow

387 00:46:52.160 ⇒ 00:47:01.549 Gabriel Lam: than you doing in cursor. Like, how long it takes you to do it in cursor versus, like, hey, I’m gonna use this sort of janky thing in the platform that we haven’t fully fleshed out.

388 00:47:11.730 ⇒ 00:47:15.349 Uttam Kumaran: using ChatGPT custom prompts.

389 00:47:17.250 ⇒ 00:47:29.179 Uttam Kumaran: are those good use cases for me to be like, okay, I shouldn’t be using custom ChatGPT prompts, I should only be using… So, what I’m saying is, like, what things should I prioritize

390 00:47:29.320 ⇒ 00:47:35.720 Uttam Kumaran: Using the platform, or asking you to make this easy for me to do in the platform versus others.

391 00:47:36.330 ⇒ 00:47:47.390 Uttam Kumaran: Because that allows… at least allows me to start to segment what my AI workflows are, and be like, okay, these are the ones that I can reasonably go to Gabe and be like, you should… maybe we can support these, because I’m using them all the time.

392 00:47:49.260 ⇒ 00:47:50.050 Gabriel Lam: Hmm.

393 00:47:51.750 ⇒ 00:47:52.520 Uttam Kumaran: You know?

394 00:47:52.520 ⇒ 00:47:53.160 Gabriel Lam: Yeah.

395 00:47:58.140 ⇒ 00:48:01.119 Gabriel Lam: Because I actually think it’s okay for you to say…

396 00:48:01.380 ⇒ 00:48:11.379 Uttam Kumaran: Okay, this cursor thing, I know it’s great. Frankly, you should go ahead and use a cursor, but actually, here’s a couple things. One, how about you let our team build you the cursor rules?

397 00:48:11.500 ⇒ 00:48:20.460 Uttam Kumaran: You know, how about maybe you help, like… there’s some cursor configuration, maybe, that can scale, and I consider platform as not just, like.

398 00:48:21.260 ⇒ 00:48:24.140 Uttam Kumaran: anything with the UI, like, platform is actually, like.

399 00:48:24.370 ⇒ 00:48:37.590 Uttam Kumaran: tools, processes, know-how, like, it’s everything. So even as the platform, you know, the platform team can support me in cursor by saying, oh, well, given that use case, here’s, like, a couple of cursor tips and tricks, like, you should…

400 00:48:37.830 ⇒ 00:48:47.550 Uttam Kumaran: you should use. We’re like, actually, hey, let’s go ahead and take your prompt and make it something that you could do a slash command with in cursor. So there’s actually ways to still help me out.

401 00:48:50.730 ⇒ 00:48:52.300 Gabriel Lam: Yeah, I like that. But I…

402 00:48:55.410 ⇒ 00:48:56.209 Uttam Kumaran: I think it’s actually…

403 00:49:02.970 ⇒ 00:49:03.840 Uttam Kumaran: be fair.

404 00:49:06.020 ⇒ 00:49:07.770 Uttam Kumaran: Trevely, look, we…

405 00:49:11.100 ⇒ 00:49:11.930 Uttam Kumaran: Cheers.

406 00:49:15.240 ⇒ 00:49:20.000 Gabriel Lam: Sorry, you cut out the last 10 seconds. I don’t know if it’s me or if it’s you.

407 00:49:27.760 ⇒ 00:49:29.029 Gabriel Lam: Hello, can you hear me?

408 00:50:44.160 ⇒ 00:50:45.250 Gabriel Lam: Hello.

409 00:50:50.320 ⇒ 00:50:51.230 Gabriel Lam: Hello?

410 00:50:52.840 ⇒ 00:50:53.899 Gabriel Lam: Oh, dear.

411 00:51:16.040 ⇒ 00:51:17.879 Uttam Kumaran: Does this work?

412 00:51:18.350 ⇒ 00:51:20.460 Gabriel Lam: A little better.

413 00:51:20.690 ⇒ 00:51:24.800 Gabriel Lam: Yeah, I think you got cut off after you were talking about, like.

414 00:51:25.130 ⇒ 00:51:28.410 Gabriel Lam: Platform as, like, including tools.

415 00:51:28.410 ⇒ 00:51:29.260 Uttam Kumaran: Yeah. Processes.

416 00:51:29.260 ⇒ 00:51:30.010 Gabriel Lam: and workflows.

417 00:51:30.010 ⇒ 00:51:31.179 Uttam Kumaran: I guess it’s, like, outside of

418 00:51:31.510 ⇒ 00:51:42.129 Uttam Kumaran: That’s something as a user, yeah, I’d be happy if you were like, okay, look, we can’t support you there, but, like, if you are using Cursor, like, here’s a bunch of stuff you should try to do. Or, like.

419 00:51:42.380 ⇒ 00:51:48.830 Uttam Kumaran: for example, I think one… more of my point is, like, how does the platform team help me scale

420 00:51:48.970 ⇒ 00:51:58.110 Uttam Kumaran: any AI solution, like, even if it’s like, hey, everybody should be using Granola, okay, like, here we’re gonna put together a little granola tips and tricks or something. I don’t know, those are the kind of the things…

421 00:51:58.230 ⇒ 00:52:01.769 Uttam Kumaran: that I’m like, okay, we can’t… if we can’t do these things now.

422 00:52:01.930 ⇒ 00:52:12.359 Uttam Kumaran: maybe we’re… maybe we’re, like, okay, maybe you’re like, hey, I need, like, $20,000 and, like, two engineers to do this. Fine. Okay, you don’t have it right now. But, what I do… what I can help you is, like.

423 00:52:12.560 ⇒ 00:52:26.219 Uttam Kumaran: I can… we can walk people through creating… using Cursor, I’ll help you create a repo where you can store your, like, cursor prompts. Maybe, like, there’s a bunch of workflows that actually are better served in Cursor. Like, I actually feel like a lot of writing

424 00:52:26.250 ⇒ 00:52:36.759 Uttam Kumaran: should just be done in Cursor. I’ll send you this video you’ll like about this guy, he’s a PM at… I forgot what company, but he does all of his PRD writing in Cursor, and…

425 00:52:37.380 ⇒ 00:52:44.939 Uttam Kumaran: I don’t know, it was, like, great, because I’m like, oh shit, we should not be writing anything in, like, Google Docs, because it’s way nicer to use those same features.

426 00:52:45.090 ⇒ 00:52:48.060 Uttam Kumaran: And maybe that’s just it, that we’re like, okay, actually.

427 00:52:48.280 ⇒ 00:52:58.959 Uttam Kumaran: Cursor is where we should support Cursor for writing, and some of these workflows, actually, we shouldn’t go ahead and reinvent this in our platform, because it would be a waste of time. Cursor’s way better, they’re always gonna be better, you know?

428 00:52:59.310 ⇒ 00:53:07.880 Gabriel Lam: Yeah. Well, I think I’m with you on that, actually. I feel that… I think that the… current…

429 00:53:08.890 ⇒ 00:53:16.219 Gabriel Lam: or the use cases that we’ve used the platform for so far are sort of so niche that they kind of have to be built up, but I think I am personally a fan of

430 00:53:16.390 ⇒ 00:53:18.959 Gabriel Lam: Like, is there an existing tool?

431 00:53:19.210 ⇒ 00:53:19.800 Gabriel Lam: That works.

432 00:53:19.800 ⇒ 00:53:20.400 Uttam Kumaran: Yeah.

433 00:53:20.400 ⇒ 00:53:20.980 Gabriel Lam: It’s.

434 00:53:20.980 ⇒ 00:53:23.270 Uttam Kumaran: our UX, that’s already good.

435 00:53:23.270 ⇒ 00:53:26.169 Gabriel Lam: Yeah. Because they’re gonna invest, like, a bajillion dollars.

436 00:53:26.210 ⇒ 00:53:35.490 Uttam Kumaran: to just make sure great at coding, and being great at coding is not so dissimilar from being good at writing. Like, you could use a lot of the same things. In fact.

437 00:53:36.220 ⇒ 00:53:46.900 Uttam Kumaran: when I realized that Cursor was actually so good at writing, I’m like, holy shit. Like, all these things are so helpful for when you’re writing, like, looking up other documents, things like that, so…

438 00:53:47.210 ⇒ 00:54:00.049 Gabriel Lam: Yeah, I think it’s, like, even… I think until you hit a point in which you’re like, hey, I think this, A, we have the capability to do it in-house, or this thing is taking up so much money in which it is more…

439 00:54:00.410 ⇒ 00:54:06.159 Gabriel Lam: financially sensible to then move it, I think that’s the conversation that happens where…

440 00:54:06.570 ⇒ 00:54:11.320 Gabriel Lam: now maybe we actually take the time to invest and do it. It’s like, you know, we talk about

441 00:54:11.910 ⇒ 00:54:15.049 Gabriel Lam: even things like, I don’t know, Clockify, where it’s like, okay, it’s a time.

442 00:54:15.050 ⇒ 00:54:15.660 Uttam Kumaran: Yeah.

443 00:54:15.660 ⇒ 00:54:19.250 Gabriel Lam: How… like, are we paying for things we don’t need, versus, like…

444 00:54:19.250 ⇒ 00:54:20.080 Uttam Kumaran: Cursor…

445 00:54:20.080 ⇒ 00:54:24.700 Gabriel Lam: Like, okay, this is something that’s really good that… It’ll be…

446 00:54:24.700 ⇒ 00:54:28.209 Uttam Kumaran: 20 bucks a month, dude, like, maybe I should just, like…

447 00:54:28.520 ⇒ 00:54:31.320 Uttam Kumaran: We should just onboard everybody who writes.

448 00:54:31.430 ⇒ 00:54:34.200 Uttam Kumaran: onto cursor, like, maybe we should take the time to teach

449 00:54:34.310 ⇒ 00:54:40.369 Uttam Kumaran: everybody on sales how to use Cursor, and like, you’re like, hey, that costs way less than $20,000 in 3 months.

450 00:54:40.630 ⇒ 00:54:41.290 Gabriel Lam: That’s fair.

451 00:54:41.290 ⇒ 00:54:41.910 Uttam Kumaran: You know?

452 00:54:42.140 ⇒ 00:54:44.230 Gabriel Lam: Yeah, so…

453 00:54:47.600 ⇒ 00:54:51.320 Gabriel Lam: I think it’s just like any other onboarding, like.

454 00:54:52.150 ⇒ 00:55:07.119 Gabriel Lam: the… I think people just need to have… either see it happen, or have someone walk them through it, and where, like, hey, you know, I’m installing Cursor, I’m trying to clone a repository, or, like, install these packages and dependencies, like, what does that mean? How do I do it? And I think…

455 00:55:08.130 ⇒ 00:55:09.290 Gabriel Lam: You know, it’s like…

456 00:55:09.290 ⇒ 00:55:23.800 Uttam Kumaran: It’s actually helpful just for me to even get the pushback in that way, actually, because without this conversation, I would have been like, damn it, our platform sucks, we can’t support this. Instead, it’s like, okay, actually, maybe we shouldn’t support this. Actually, maybe…

457 00:55:23.800 ⇒ 00:55:32.440 Uttam Kumaran: I should just start teaching… I should just start treating Cursor as an extension of the platform, and it’s like, what is Brainforge’s unique advantage?

458 00:55:32.500 ⇒ 00:55:35.860 Uttam Kumaran: when using or adopting Furser, you know, like.

459 00:55:35.930 ⇒ 00:55:54.499 Uttam Kumaran: And maybe the answer is, hey, actually, just go train, like, maybe I’ll call Rico tomorrow and be like, hey, I want to walk you through how to use Cursor, and the time it’ll take me to teach him, which may be just, like, a week or two, is better than, like, if we were to go build this, and then we’d still probably be behind Cursor in terms of UX and stuff, you know, like…

460 00:55:54.500 ⇒ 00:55:55.090 Gabriel Lam: Yeah.

461 00:55:55.090 ⇒ 00:55:55.680 Uttam Kumaran: Yeah.

462 00:55:56.750 ⇒ 00:55:58.429 Gabriel Lam: Yeah, I think, I think that…

463 00:55:58.640 ⇒ 00:56:03.120 Gabriel Lam: The processes that we are building in-house work at the moment, because

464 00:56:04.970 ⇒ 00:56:13.320 Gabriel Lam: like, the architecture that exists with, like, Supabase, and all the transcripts, and all the messages being linked up the way it is, like, I think it makes sense to bring that in, but…

465 00:56:14.430 ⇒ 00:56:21.630 Gabriel Lam: In some ways, if… like, you could argue Cursor has its own… rag, right?

466 00:56:21.830 ⇒ 00:56:32.100 Uttam Kumaran: Yeah, no, it basically does, and it’s like… it’s actually so easy for me to be like, oh, I just created a repo, just gonna throw all these things in there, and then I now can basically generate something and, like.

467 00:56:32.820 ⇒ 00:56:34.780 Uttam Kumaran: You know, it was, like, 30 seconds.

468 00:56:35.540 ⇒ 00:56:42.649 Uttam Kumaran: I mean, and honestly, more of my point is, like, I’m kind of like, why don’t we write everything? I’m kind of trying to think about…

469 00:56:43.160 ⇒ 00:56:46.579 Uttam Kumaran: like, why don’t I do more writing in…

470 00:56:47.560 ⇒ 00:56:50.849 Uttam Kumaran: in cursor. Like, I feel like we sh- we don’t do enough

471 00:56:50.980 ⇒ 00:56:54.280 Uttam Kumaran: like, some people in our company are still just raw-dogging Google Docs.

472 00:56:54.510 ⇒ 00:56:56.390 Uttam Kumaran: And I’m like, what? Like, what… what do we.

473 00:56:56.390 ⇒ 00:56:59.140 Gabriel Lam: I think it’s only because they’re used to it, right? Like…

474 00:56:59.140 ⇒ 00:57:01.200 Uttam Kumaran: Totally, my heart blaming me was just, like.

475 00:57:01.200 ⇒ 00:57:11.459 Gabriel Lam: what does a Markdown format look like? How do I read Markdown when it’s wrong? Yeah, yeah. Like, that part is confusing for people, but I think, like, you know, give them

476 00:57:13.040 ⇒ 00:57:14.060 Gabriel Lam: Like, a week’s plan.

477 00:57:14.060 ⇒ 00:57:14.970 Uttam Kumaran: the benefits.

478 00:57:14.970 ⇒ 00:57:16.449 Gabriel Lam: And be like, okay.

479 00:57:16.610 ⇒ 00:57:23.060 Gabriel Lam: Instead of bolding, like, this is what it might look like, or, like, this is what headers, like, how you organize headers.

480 00:57:23.800 ⇒ 00:57:25.209 Gabriel Lam: That’s the new…

481 00:57:25.340 ⇒ 00:57:30.660 Gabriel Lam: terminology, and sure, it’s gonna look clanky the first time, couple times you do it, because you don’t know how to do it.

482 00:57:32.660 ⇒ 00:57:34.619 Gabriel Lam: The other thing might just be, like, hey.

483 00:57:37.070 ⇒ 00:57:38.619 Gabriel Lam: Something that I’ve done is, like.

484 00:57:38.920 ⇒ 00:57:42.890 Gabriel Lam: you know, like, ChatGPT, can you turn this into a Markdown raw format?

485 00:57:45.550 ⇒ 00:57:46.170 Uttam Kumaran: Yeah.

486 00:57:46.170 ⇒ 00:57:50.440 Gabriel Lam: And if there’s, like, notes that I’m writing on the side, then I’m like, hey, I have some ideas.

487 00:57:50.780 ⇒ 00:57:57.310 Gabriel Lam: give it to me in that as well. And I don’t think it has to be like, oh, I have to learn exactly, you know, how to write word for word.

488 00:57:57.310 ⇒ 00:58:01.389 Uttam Kumaran: But dude, this is also the case, like, maybe instead of doing agents…

489 00:58:01.510 ⇒ 00:58:14.909 Uttam Kumaran: in the platform, like, any agent that’s, like, a writing agent, we should say, we’re not gonna support this use case in the platform. This use case is supported in Cursor, and we have a repo with all of our core prompts.

490 00:58:15.540 ⇒ 00:58:29.749 Uttam Kumaran: you know, you can pull the repo down, it has all of our, like, most important prompts there, and you can choose any model, you can do a bunch of things, and, like, that’s just gonna be where you do things. Like, there’s gonna be no opportunity for you to do

491 00:58:30.250 ⇒ 00:58:33.980 Uttam Kumaran: chat over Markdown exclusively things, and…

492 00:58:34.180 ⇒ 00:58:45.319 Uttam Kumaran: in the platform instead. We’re gonna simplify the platform to, like, just do things that are more complicated, like chatting over all your slacks and things like that.

493 00:58:45.480 ⇒ 00:58:49.159 Uttam Kumaran: You know? And then maybe instead, I just, like, teach everybody how to use cursor.

494 00:58:49.370 ⇒ 00:58:53.979 Uttam Kumaran: So I think the list on… I think the list on Teaching Cursor is gonna be a lot less than the list on…

495 00:58:55.160 ⇒ 00:59:04.910 Uttam Kumaran: you know, building, like, every single type of agent. I don’t know. Because in Cursor, you could literally be like, hey, go ahead and use this prompt from the prompts folder, and then do this thing.

496 00:59:05.130 ⇒ 00:59:05.920 Gabriel Lam: Yep.

497 00:59:05.920 ⇒ 00:59:10.820 Uttam Kumaran: It’s like, oh shit, I should’ve just… it’s like a better… maybe it’s a better UI.

498 00:59:10.820 ⇒ 00:59:11.510 Gabriel Lam: Yeah.

499 00:59:13.910 ⇒ 00:59:18.300 Gabriel Lam: Yeah, like, you could even, you know, have… I don’t know.

500 00:59:18.650 ⇒ 00:59:22.860 Gabriel Lam: like… Langth views be the most updated version?

501 00:59:23.140 ⇒ 00:59:28.679 Gabriel Lam: People will take it and edit it as they wish. And, you know, if anyone has a good idea, be like, hey, I tried this.

502 00:59:29.010 ⇒ 00:59:32.759 Gabriel Lam: like, it gets me this kind of result. They can update it themselves.

503 00:59:33.560 ⇒ 00:59:44.329 Uttam Kumaran: Yeah. And then, because you can also generate, like, we can generate our own MCPs and things like that. So yeah, I think this is helpful for me to think about, like, okay, maybe it is… maybe what I’ll do is I’ll ask Rico, I’ll ask

504 00:59:44.510 ⇒ 00:59:47.110 Uttam Kumaran: I’ll see, because he does a lot of stuff where he’s,

505 00:59:47.300 ⇒ 00:59:54.669 Uttam Kumaran: he’s writing a ton of emails and recruiting stuff, like, I think he’s just using ChatGPT a lot, but he can probably save a lot of those prompts.

506 00:59:54.970 ⇒ 00:59:55.890 Uttam Kumaran: And, like.

507 00:59:56.020 ⇒ 01:00:02.329 Uttam Kumaran: probably speed up a lot of his workflows, so maybe that’s… maybe I’ll try with him to see if I can get him to…

508 01:00:02.510 ⇒ 01:00:08.560 Uttam Kumaran: So, because generating, sales SOW is not, like, an easy thing to do.

509 01:00:08.880 ⇒ 01:00:24.689 Uttam Kumaran: You know, but if I can teach Rico to do it, that’s a great use case of, like, oh, we’ve taken someone who has, like, limited skill set in, like, drafting sales scopes of work, and now is able to do it, which is great, because I would have had to

510 01:00:24.840 ⇒ 01:00:27.660 Uttam Kumaran: Hire someone who could do that, you know?

511 01:00:28.150 ⇒ 01:00:28.810 Gabriel Lam: Yeah.

512 01:00:29.050 ⇒ 01:00:32.580 Gabriel Lam: I mean, I think, like, in that case, it might even be, like.

513 01:00:34.260 ⇒ 01:00:36.960 Gabriel Lam: like, when Cursor first opens, you have this…

514 01:00:37.510 ⇒ 01:00:49.009 Gabriel Lam: you know, it’s, like, open a file a little bit. It’s, like, new agent, new terminal, like, research files, and it’s, like, very… I mean, it’s VS Code, right? And so it’s, like… but I think you can actually even toggle the view into.

515 01:00:49.010 ⇒ 01:00:51.000 Uttam Kumaran: Yeah, you can turn a lot of that off, yeah.

516 01:00:51.000 ⇒ 01:00:52.979 Gabriel Lam: You can turn all the… you can turn the terminals in the.

517 01:00:52.980 ⇒ 01:00:59.790 Uttam Kumaran: Oh, maybe it’s, like, as part of onboarding, like, everybody, during their onboarding, like, we just set up… you set up your cursor.

518 01:00:59.990 ⇒ 01:01:02.280 Gabriel Lam: Yeah, yeah, and be like, hey, Rico, like.

519 01:01:02.940 ⇒ 01:01:08.270 Gabriel Lam: press the toggle, this is gonna look exactly like ChatGPT, so you know what to do, you know what to expect.

520 01:01:08.270 ⇒ 01:01:16.690 Uttam Kumaran: And then later on, you can preload a settings file. Yeah. So we can preload a settings file for, like, a couple of the roles, like, hey, if you’re in sales.

521 01:01:16.810 ⇒ 01:01:28.410 Uttam Kumaran: use this settings file. It comes preloaded with, like, a bunch of the most common prompts, places for you to, like, do the most common things. This is, like, generate follow-up email, generate SOWs.

522 01:01:28.570 ⇒ 01:01:30.140 Uttam Kumaran: You know, and maybe that’s it.

523 01:01:30.830 ⇒ 01:01:31.370 Gabriel Lam: Yeah.

524 01:01:31.370 ⇒ 01:01:39.599 Uttam Kumaran: Versus, dude, doing things like summarizing Slack messages and the unique UI things that you’re trying to do, it’s not possible in Cursor, like…

525 01:01:39.920 ⇒ 01:01:51.979 Uttam Kumaran: Those… there’s just significantly different use cases, you know? Yeah. Like, single… like, just, like, agent discussions and things, because Kershaw’s also nice, you can select multiple models, it looks through, really, files in a great way.

526 01:01:52.220 ⇒ 01:01:58.419 Uttam Kumaran: Like, for me, as an engineer, like, I’m going from writing docs to writing code and back, it’s, like, perfect.

527 01:01:58.530 ⇒ 01:02:03.619 Uttam Kumaran: You know? In fact, what sucks is that I actually have to move stuff to Google Docs.

528 01:02:04.350 ⇒ 01:02:10.340 Uttam Kumaran: You know, when I just wish I could just do all the writing in GitHub and not have to leave.

529 01:02:10.900 ⇒ 01:02:12.130 Uttam Kumaran: Yeah.

530 01:02:16.940 ⇒ 01:02:24.160 Gabriel Lam: Yeah, I… yeah. I… I… I am a fan, because I… I like the optionality of choosing between

531 01:02:24.400 ⇒ 01:02:30.289 Gabriel Lam: models… like, I think Sonnet easily is the most accurate with the least hallucinations at the moment.

532 01:02:30.290 ⇒ 01:02:30.689 Uttam Kumaran: Yeah. It’s like.

533 01:02:30.740 ⇒ 01:02:35.200 Gabriel Lam: Yeah. Even GPT-5.1, I think, struggles in that regard.

534 01:02:37.620 ⇒ 01:02:57.100 Gabriel Lam: Yeah, I think, like, if different teams are able to see Cursor as not just a coding platform, be like, oh, if I want to use it the way, like, the way I use chat, which I’ve come to learn, then the mental load of that as well is a lot smaller, where you’re like, okay, I can just, you know…

535 01:02:57.530 ⇒ 01:03:07.829 Gabriel Lam: take custom prompts in ChatGPT that I know how to use, bring it as an agent in cursor, I know what to expect, I don’t have to relearn everything that I thought I did.

536 01:03:08.100 ⇒ 01:03:11.900 Gabriel Lam: It’s just… a different tool.

537 01:03:14.790 ⇒ 01:03:21.750 Uttam Kumaran: So here’s one thing that I’ll do. So tonight, I’ll… I’m gonna move a lot of the writing stuff that I’ve been doing to, like, a new…

538 01:03:22.210 ⇒ 01:03:24.539 Uttam Kumaran: I’m just gonna say, like, Brain Forge.

539 01:03:24.760 ⇒ 01:03:28.640 Uttam Kumaran: Cursor, or like… I don’t know, just a repo.

540 01:03:28.950 ⇒ 01:03:39.009 Uttam Kumaran: But I think what… so I will do that for you, in that I’m just gonna move some of my workloads, I’m gonna record me doing an SOW. What I need your help with is I can’t…

541 01:03:39.440 ⇒ 01:03:58.850 Uttam Kumaran: do the adoption. Like, I can’t put the pressure on everybody and, like, follow up and, like, find out. I just can’t do that. So that’s maybe where I can… what I can do, and what you know I can do, is come up with these, like, novel use cases, because I’m just trying anyway to get back… to reclaim my time.

542 01:03:58.850 ⇒ 01:03:59.200 Gabriel Lam: Hmm.

543 01:03:59.200 ⇒ 01:04:13.810 Uttam Kumaran: So when I find these, let me just send them over, but I think my ask for you is, like, help me get them adopted. Like, think about ways that we can make it interesting, or even if you’re, like, incentivized, like, even if you say, hey.

544 01:04:14.060 ⇒ 01:04:23.349 Uttam Kumaran: hey, can you put a $500 budget to the side for the person that is not on engineering that uses Cursor the most this month? Okay, I’m down to do that.

545 01:04:24.070 ⇒ 01:04:31.020 Uttam Kumaran: But I see, I don’t care. For me, the money is actually, like, we’re losing money by people spending time doing it manually.

546 01:04:31.020 ⇒ 01:04:31.360 Gabriel Lam: Yeah.

547 01:04:31.490 ⇒ 01:04:35.310 Uttam Kumaran: I’m not… can I give you, really, like, almost, like, 180?

548 01:04:35.520 ⇒ 01:04:47.960 Uttam Kumaran: there are engineers on our team that are writing. They’re writing docs and things like that. The writing is really poor. They’re misspelling things, the grammar is wrong, they’re not running it through simple

549 01:04:48.260 ⇒ 01:04:58.610 Uttam Kumaran: like, grammar check prompts. They’re not running it through, hey, a client feedback prompt, like, hey, put yourself in the shoes of the client. Is this document clear?

550 01:04:58.610 ⇒ 01:05:13.380 Uttam Kumaran: Like, is there a clear header of, like, what the purpose is? They… today, we had two documents that went out that I got client feedback from that said, this is not clear enough, you didn’t land a plane on this document. I can’t use this.

551 01:05:14.470 ⇒ 01:05:17.650 Uttam Kumaran: Simply, had they ran it through, like, a…

552 01:05:18.090 ⇒ 01:05:31.409 Uttam Kumaran: you are a client for BrainForge. Your job is to take this document and give feedback. Please highlight the top points of feedback for me to make this better. If they had done that, I would not have gotten yelled at, like, 2 hours ago.

553 01:05:31.410 ⇒ 01:05:32.030 Gabriel Lam: Yeah.

554 01:05:32.460 ⇒ 01:05:39.240 Uttam Kumaran: So that’s, like, an opposite case where the engineers are not good at writing, so they’re not using cursor for writing.

555 01:05:39.520 ⇒ 01:05:40.210 Uttam Kumaran: You know, they’re.

556 01:05:40.210 ⇒ 01:05:45.509 Gabriel Lam: I think that actually would be easier adopters, where it’s like, hey, you already know what cursor’s like.

557 01:05:45.510 ⇒ 01:05:59.439 Uttam Kumaran: Or, dude, what they do is they just take it, they just take what I said, and they put it into ChatGPT, they copy-paste, and there’s emojis, there’s all this shit. I’m like, dude, who are you fooling? Like, our customers are very smart. What are you gonna do, put this AI slop in front of them? They’re paying us, like.

558 01:05:59.700 ⇒ 01:06:13.680 Uttam Kumaran: $20,000,$30,000 for this AI, like, nonsense? I almost, like, I’m, like, feeling bad that you think that this is… this is what is gonna fly around here. Like, at least use Sonnet and, like, do a prompt, and, like, come on, like…

559 01:06:13.680 ⇒ 01:06:14.520 Gabriel Lam: Yo.

560 01:06:14.520 ⇒ 01:06:20.579 Uttam Kumaran: you know, just don’t… I don’t… I’m not saying do it manually, I’m not… I’m also not saying cheat, like, 100%, like…

561 01:06:20.990 ⇒ 01:06:26.199 Uttam Kumaran: you know, use your brain. But it’s also, like, tough, because I didn’t teach people

562 01:06:26.350 ⇒ 01:06:38.570 Uttam Kumaran: how to write. I’m obviously telling everybody to use AI, like, non-stop, and so I can’t fault them for that, but, like, what can I do? Okay, like, maybe we do figure out, how do I get engineers to do their writing.

563 01:06:39.010 ⇒ 01:06:45.170 Uttam Kumaran: using AI in a reasonable manner that actually speeds it up but doesn’t sacrifice quality, right? Like… But…

564 01:06:45.170 ⇒ 01:06:45.800 Gabriel Lam: Yeah.

565 01:06:45.800 ⇒ 01:06:51.280 Uttam Kumaran: Yeah, that’s… maybe that’s just ways for us to do… to think about incentivizing people and trying to get people to…

566 01:06:51.410 ⇒ 01:06:57.659 Uttam Kumaran: to do more, but yeah, I mean, think outside the box, like, I’m happy to sponsor, because these are real tasks that are, like.

567 01:06:57.770 ⇒ 01:07:11.910 Uttam Kumaran: you can tell and clockify how many hours we’re spending. Like, we’re spending real volume of hours on tasks that can probably take 25… 10-25% of the time they’re taking today, because people are not using simple AI

568 01:07:12.040 ⇒ 01:07:22.270 Uttam Kumaran: advancements that me and you, if we were in their spot, we would… we would do faster. So there is… there is money sitting there. So I don’t mind incentivizing, if we can incentivize the correct behavior.

569 01:07:22.440 ⇒ 01:07:32.829 Uttam Kumaran: And, like, you know, so I’m down for it. I think for me, it’s like, I just don’t have the creativity on, like, how to get the adoption. I just know…

570 01:07:33.080 ⇒ 01:07:37.220 Uttam Kumaran: I’ve done every job in this business, so I just know every single way.

571 01:07:37.320 ⇒ 01:07:39.760 Uttam Kumaran: to use AI to kind of do it faster.

572 01:07:40.150 ⇒ 01:07:40.830 Gabriel Lam: Yes.

573 01:07:43.100 ⇒ 01:07:45.170 Gabriel Lam: I… I… I’m thinking about it. I…

574 01:07:45.170 ⇒ 01:07:46.830 Uttam Kumaran: Okay.

575 01:07:46.830 ⇒ 01:07:54.260 Gabriel Lam: It’s like, the sponsorship thing is kind of interesting, right? Because it’s like, then they have… they have some sort of additional skin of, like, hey.

576 01:07:54.630 ⇒ 01:07:58.360 Gabriel Lam: I mean, then there’s, like, how do you track it, but… .

577 01:07:58.710 ⇒ 01:08:08.789 Uttam Kumaran: I mean, dude, yeah, I think… I think that’s exactly it. Maybe it is, like, hey, for every… for all of our… for… you could do it for both sides of the house, the engineers and the non-engineers. You know, like…

578 01:08:09.530 ⇒ 01:08:11.399 Uttam Kumaran: I don’t know.

579 01:08:11.400 ⇒ 01:08:13.379 Gabriel Lam: Or it’s like, even if it’s like, hey, like.

580 01:08:14.320 ⇒ 01:08:18.460 Gabriel Lam: If you come up with a new use case, or, like, any new use cases get…

581 01:08:19.060 ⇒ 01:08:25.469 Gabriel Lam: some sort of reward in some way, where you’re like, hey, I’m gonna go out of my way to try to automate something, and…

582 01:08:25.680 ⇒ 01:08:28.790 Gabriel Lam: Write something, and get it reviewed, and…

583 01:08:29.180 ⇒ 01:08:36.399 Gabriel Lam: like, it’ll be more than just, like, I get to save time, but it’s like, hey, I have…

584 01:08:39.090 ⇒ 01:08:42.049 Gabriel Lam: like, there’s something that I can look forward to as well.

585 01:08:42.520 ⇒ 01:08:44.590 Uttam Kumaran: Yeah, like, well, the,

586 01:08:44.740 ⇒ 01:08:49.899 Uttam Kumaran: I mean, or you can say, like, hey, for… Yeah, for every, like.

587 01:08:50.060 ⇒ 01:08:57.039 Uttam Kumaran: article for every, like, piece of writing that you submit that passes this, like… So, one of the things… have I told you about Vixel?

588 01:08:57.700 ⇒ 01:09:02.560 Gabriel Lam: Cool. I’ve heard of it, I keep hearing about it, I don’t actually know…

589 01:09:02.569 ⇒ 01:09:13.989 Uttam Kumaran: So Vixel is a… yeah, Vixel is an accelerator that Robert and I joined with Brainforge, and one of the things that’s very interesting is they had us write, like, OKRs and stuff early on, but, like.

590 01:09:14.359 ⇒ 01:09:33.299 Uttam Kumaran: typically, they would… you would have to send it to somebody and get a review, but they actually built an agent that gives you a review. So you write your OKRs, and you literally send it into the agent in Slack, and then it gives you feedback, and it scores it. And only when you score a 5 out of 5 can you go check the button in the platform, that little Airtable.

591 01:09:33.429 ⇒ 01:09:42.899 Uttam Kumaran: So you just keep iterating until you get to 5 out of 5, and what is that? It’s probably a simple prompt that they wrote. So maybe one way to do it is you’re like, hey, every time you write with

592 01:09:43.189 ⇒ 01:09:48.800 Uttam Kumaran: every time you write with whatever… Like, write a document, it has to be… Submit it in this channel.

593 01:09:48.800 ⇒ 01:09:49.430 Gabriel Lam: Yeah.

594 01:09:49.430 ⇒ 01:09:54.949 Uttam Kumaran: And the… and then, like, the more… if you… as many 5… whoever gets the most 5 out of 5s.

595 01:09:55.310 ⇒ 01:09:58.619 Uttam Kumaran: like… Gets $1,000.

596 01:10:00.030 ⇒ 01:10:04.369 Uttam Kumaran: At the end of the month, and you just can just do these challenges. Like, I think that’s what’s really…

597 01:10:04.560 ⇒ 01:10:09.690 Uttam Kumaran: Yeah. That’s what really makes, like, sort of this interesting sort of growth product management is, like.

598 01:10:09.920 ⇒ 01:10:19.140 Uttam Kumaran: okay, how do we actually get people to incentivize, right? Cash is a great way of doing it. I think it’s up to you to think about creatively how you use the cash.

599 01:10:19.470 ⇒ 01:10:21.359 Uttam Kumaran: To drive the right behavior.

600 01:10:21.370 ⇒ 01:10:26.009 Gabriel Lam: Right. Because the lovely thing about a business like ours is, like, the cash is so…

601 01:10:26.010 ⇒ 01:10:30.070 Uttam Kumaran: So small, given the behavioral change it could impact.

602 01:10:30.430 ⇒ 01:10:35.629 Uttam Kumaran: Right? If now on, everybody just starts using, like, everybody’s grammar just gets…

603 01:10:35.760 ⇒ 01:10:39.839 Uttam Kumaran: 10 times better, okay, like, perfect, you know?

604 01:10:40.040 ⇒ 01:10:46.959 Uttam Kumaran: Or if any time now, now, no matter what, people are gonna be like, oh, like, it’d be nice to just get the cash, I’m gonna do it anyway, so I might as well just do it right.

605 01:10:47.230 ⇒ 01:10:50.819 Uttam Kumaran: So, I don’t know, there’s, like, interesting things you could do there. I think, immediately.

606 01:10:51.060 ⇒ 01:10:58.230 Uttam Kumaran: Giving some amount of money to, like, the first and second place person, non… Non,

607 01:10:58.560 ⇒ 01:11:03.150 Uttam Kumaran: engineering to use Cursor in a month? Yeah, I mean, that’d be awesome.

608 01:11:03.150 ⇒ 01:11:09.900 Gabriel Lam: I’m… yeah, I mean… it, like, gamifies, which, like, I think is a good thing.

609 01:11:10.160 ⇒ 01:11:11.450 Uttam Kumaran: Yeah.

610 01:11:15.200 ⇒ 01:11:17.810 Gabriel Lam: Okay. That’s what our company needs.

611 01:11:18.060 ⇒ 01:11:18.420 Gabriel Lam: Yeah.

612 01:11:18.420 ⇒ 01:11:22.539 Uttam Kumaran: more help with, and, you know, I think we’re thinking about getting

613 01:11:22.880 ⇒ 01:11:31.880 Uttam Kumaran: Another operations person, Lauren, who may join. And so, really, part of the directive for, I think, all of leadership is going to be on thinking about how to

614 01:11:32.240 ⇒ 01:11:36.810 Uttam Kumaran: do more with AI, but… but, like, I want to do it in a way where it helps the team, like.

615 01:11:36.940 ⇒ 01:11:40.030 Uttam Kumaran: They’re actually seeing the positive benefits.

616 01:11:40.170 ⇒ 01:11:42.440 Uttam Kumaran: Of it in their, like, day-to-day, like…

617 01:11:42.540 ⇒ 01:11:47.190 Uttam Kumaran: A lot of the work we do on the engineering side is a lot of maintenance, and it’s a lot of the same thing.

618 01:11:47.310 ⇒ 01:11:50.759 Uttam Kumaran: And I think people are doing the lazy thing with AI, which is just, like.

619 01:11:50.970 ⇒ 01:11:53.810 Uttam Kumaran: Okay, I have to do this thing, go for it, you know?

620 01:11:54.510 ⇒ 01:11:59.890 Uttam Kumaran: We can be smarter, you know, and help them do the right thing, still just as fast.

621 01:12:00.010 ⇒ 01:12:08.330 Uttam Kumaran: Versus if we give everyone ChatGPT, right? Like, I’m almost at the point where at some point, dude, you tell me, you’re like, turn off ChatGPT, I’ll turn it off.

622 01:12:08.880 ⇒ 01:12:14.589 Uttam Kumaran: And you can say the only way you can start doing this stuff is through the platform or through our pre-built cursor.

623 01:12:15.070 ⇒ 01:12:20.000 Uttam Kumaran: But, so those are the kind of interesting things that we can do. Like, even granola, if…

624 01:12:20.400 ⇒ 01:12:27.050 Uttam Kumaran: if the platform got good enough, I would just keep Granola just for the people that are doing sales, because otherwise, I…

625 01:12:27.170 ⇒ 01:12:29.780 Uttam Kumaran: But people should just be using the platform recordings.

626 01:12:29.780 ⇒ 01:12:30.190 Gabriel Lam: Yeah.

627 01:12:30.190 ⇒ 01:12:32.480 Uttam Kumaran: So those are, like, interesting things we could do.

628 01:12:32.820 ⇒ 01:12:37.489 Uttam Kumaran: That I think you can think about, you know, how do we… Change the behavior.

629 01:12:37.750 ⇒ 01:12:38.520 Gabriel Lam: Oh, yeah.

630 01:12:38.520 ⇒ 01:12:39.140 Uttam Kumaran: Yeah.

631 01:12:44.300 ⇒ 01:12:48.089 Gabriel Lam: Yeah, it’s like… It’s like a…

632 01:12:48.560 ⇒ 01:12:50.140 Gabriel Lam: Stick and carrot kind of thing.

633 01:12:50.140 ⇒ 01:12:50.680 Uttam Kumaran: Yes.

634 01:12:50.680 ⇒ 01:12:52.560 Gabriel Lam: The carrot has to be big enough.

635 01:12:53.290 ⇒ 01:13:00.450 Gabriel Lam: That pushes, like, it pulls people forward, and also, like, at some point, if it’s like, hey, you know, we tried it, people are still…

636 01:13:01.500 ⇒ 01:13:06.529 Gabriel Lam: not upgrading their prompts, or, like, don’t have their chat At some point.

637 01:13:06.530 ⇒ 01:13:06.990 Uttam Kumaran: Literally be like.

638 01:13:06.990 ⇒ 01:13:08.580 Gabriel Lam: Figured, it’s like maybe…

639 01:13:08.580 ⇒ 01:13:14.240 Uttam Kumaran: Yeah, if we find that 6 months… that, like, 2 months in, someone’s not using the platform.

640 01:13:14.340 ⇒ 01:13:27.859 Uttam Kumaran: I’m gonna go to them and be like, I want to have a conference, go to them and be like, hey, you are actively sabotaging the company by not using this. Like, you are wasting company resources. So I better have… you better have some great reasons as to why, because.

641 01:13:27.860 ⇒ 01:13:28.450 Gabriel Lam: Yeah.

642 01:13:28.630 ⇒ 01:13:33.350 Uttam Kumaran: You know, otherwise, like… If you just don’t like AI, then there’s no… there’s no home for you here.

643 01:13:33.670 ⇒ 01:13:38.569 Gabriel Lam: Yeah, and I think on… it’s like, for me, it’s like, okay, what are the resources that people need, and like, how.

644 01:13:38.570 ⇒ 01:13:39.170 Uttam Kumaran: Yes.

645 01:13:39.170 ⇒ 01:13:39.859 Gabriel Lam: set up.

646 01:13:40.050 ⇒ 01:13:46.410 Uttam Kumaran: I would rather the carrots. I don’t like steak at all. I don’t like building, like, a fear-based environment.

647 01:13:46.410 ⇒ 01:13:46.960 Gabriel Lam: Right.

648 01:13:46.960 ⇒ 01:14:05.120 Uttam Kumaran: And for the most part, the people I’ve… the people you have now, I’ve gotten… I’ve gotten a lot of those people who were sort of, like, anti out of the company. That’s why it’s a great place right now, because everybody who’s here is actually, like, pretty good, and is very… but it’s not gonna always be the case. Right. You know?

649 01:14:05.120 ⇒ 01:14:13.829 Uttam Kumaran: And so we kind of have to have both. We have to have positive incentive, and then we have to have, like, accountability. But another thing, maybe, like, look, we’re going to think about, like, sort of leaders for each

650 01:14:13.920 ⇒ 01:14:27.600 Uttam Kumaran: like, an operations leader, or a sales leader. Maybe part of their OKRs is, like, your team needs to be… make… needs to be… make… get, like, X amount of AI points, and you can get AI points by, like, these types of activities, and…

651 01:14:27.720 ⇒ 01:14:30.460 Uttam Kumaran: You know, you kind of think about it in an interesting way like that.

652 01:14:30.840 ⇒ 01:14:33.050 Gabriel Lam: Yeah. Yeah, yeah, yeah, yeah.

653 01:14:38.680 ⇒ 01:14:40.839 Uttam Kumaran: Just some thoughts. Just some thoughts.

654 01:14:40.840 ⇒ 01:14:46.259 Gabriel Lam: Yeah, like, Like, the sort of more pessimistic side of me is reminded of the whole, like.

655 01:14:46.660 ⇒ 01:14:52.009 Gabriel Lam: I forget if it was the X or Tesla email, where it’s like, you know, you have to report to me.

656 01:14:52.010 ⇒ 01:14:52.559 Uttam Kumaran: I know.

657 01:14:52.560 ⇒ 01:15:00.929 Gabriel Lam: And I’m like, okay, I get it, right? Like, that’s the sort of dystopian way to look at it. But I think the main thing is, okay.

658 01:15:03.080 ⇒ 01:15:08.460 Uttam Kumaran: We’re small enough where it’s, like, there’s not… it’s not like there’s, like, 100,000 people on payroll that we have no idea what they.

659 01:15:08.460 ⇒ 01:15:12.729 Gabriel Lam: No, nor… I think the main thing is just knowing…

660 01:15:13.020 ⇒ 01:15:18.619 Gabriel Lam: that you can’t take everything AI says at face value, you’re always gonna have to do a review for it, but.

661 01:15:18.620 ⇒ 01:15:19.250 Uttam Kumaran: Yeah.

662 01:15:19.750 ⇒ 01:15:27.639 Gabriel Lam: running things through reviews, and, like, it could be with yourself, it could be with agents as well, but just to know that, like, hey, like, you can.

663 01:15:27.640 ⇒ 01:15:31.370 Uttam Kumaran: But that’s a couple of rules, you can just say that, hey, every time you use AI,

664 01:15:31.470 ⇒ 01:15:40.709 Uttam Kumaran: try to do a review, and maybe there’s some way for us to measure that. Like, I don’t know what data you get out of ChatGBT Enterprise, but I can give it to you, or whatever, and you can see, like.

665 01:15:40.910 ⇒ 01:15:46.449 Uttam Kumaran: maybe you can proactively identify people not using AI properly and help them out, or like.

666 01:15:46.990 ⇒ 01:15:52.710 Uttam Kumaran: somehow measure that, but because, yeah, dude, that’s, like, a big thing. People don’t take a second pass at stuff.

667 01:15:52.710 ⇒ 01:15:57.379 Gabriel Lam: They just copy and paste it raw, and then it comes to me, and I’m like, what the fuck is this? Yeah.

668 01:15:57.380 ⇒ 01:15:58.200 Uttam Kumaran: And oftentimes…

669 01:15:58.200 ⇒ 01:16:00.120 Gabriel Lam: It’s not much, it’s like, you know…

670 01:16:00.120 ⇒ 01:16:02.000 Uttam Kumaran: No, it’s not!

671 01:16:02.000 ⇒ 01:16:06.710 Gabriel Lam: Copy, paste an additional, like, custom instruction, and that takes, like, what, 10 seconds?

672 01:16:06.710 ⇒ 01:16:12.589 Uttam Kumaran: Exactly, just say, like, hey, now… actually, like, now… but also, some people are, like, they don’t read it.

673 01:16:12.850 ⇒ 01:16:18.100 Uttam Kumaran: I’m like, yo, you just got handed, like, a goldmine, just read it once.

674 01:16:18.370 ⇒ 01:16:21.910 Uttam Kumaran: Like, do… just, like, don’t be so lazy.

675 01:16:22.520 ⇒ 01:16:24.599 Uttam Kumaran: I also, I don’t blame people, like…

676 01:16:25.370 ⇒ 01:16:28.670 Uttam Kumaran: Ultimately, like, look, I’m on the hook for it being right.

677 01:16:28.870 ⇒ 01:16:33.340 Uttam Kumaran: not fast, so I was doing all this stuff without AI, so now I know, like.

678 01:16:33.540 ⇒ 01:16:42.759 Uttam Kumaran: how to sort of get the same output, or 90% of the same output. And in SOWs, dude, it’s not about getting it right, it’s actually about showing the depths.

679 01:16:42.900 ⇒ 01:16:56.979 Uttam Kumaran: Because it’s for sale… it’s a sales thing, so no one’s trying to look at all the specifics, but when we’re… when we’re sending a client a document that’s like, here’s the reason you should pick this tool, and there’s emojis on it, I’m, like, gonna break my laptop in half.

680 01:16:56.980 ⇒ 01:16:57.650 Gabriel Lam: Yeah.

681 01:16:57.650 ⇒ 01:17:05.970 Uttam Kumaran: I’m like, dude, nobody’s paying, like, you think Bain Consulting would put an emoji here? Like, what? You know what I mean? And so that’s the sort of thing where I’m like.

682 01:17:06.320 ⇒ 01:17:12.289 Uttam Kumaran: Okay, we tried to use AI, so at least we have people using AI. Many of the companies we go to don’t have that.

683 01:17:12.890 ⇒ 01:17:16.009 Uttam Kumaran: But there’s a second layer now, we have to be smart. And for us, that’s…

684 01:17:16.130 ⇒ 01:17:20.959 Uttam Kumaran: that’s what takes a company that’s so small like us and puts us into, like, a different league, you know? So…

685 01:17:21.100 ⇒ 01:17:26.460 Uttam Kumaran: Part of it is just, like, It’s the opportunity cost, like… What could we have gotten?

686 01:17:26.680 ⇒ 01:17:37.800 Uttam Kumaran: How do we, you know, and also, a lot of people who come to our company, you know, I feel like are seeing how much we’re using here, like, damn, okay, like, this is a cool place to work. Like, when we recruit, a lot of people are like.

687 01:17:38.140 ⇒ 01:17:52.350 Uttam Kumaran: honestly want to work somewhere where they treat AI, like, very seriously, versus a place where it’s just, like, committee or something like that. So that’s why I think even a lot of this is, like, recruiting, retention, like, imagine going from our company to a place that doesn’t allow, like, any AI tools.

688 01:17:53.000 ⇒ 01:17:54.779 Gabriel Lam: Yeah, you’ll feel sorry.

689 01:17:54.780 ⇒ 01:17:55.420 Uttam Kumaran: Crazy.

690 01:17:55.420 ⇒ 01:17:56.279 Gabriel Lam: And you’re all like, you’re.

691 01:17:56.280 ⇒ 01:18:04.040 Uttam Kumaran: I don’t even know what planet I’d be on. I don’t even know what I’d do. Yeah. I don’t think I could go back.

692 01:18:07.560 ⇒ 01:18:14.989 Gabriel Lam: Okay, I think, yeah, for sure. I… that… I’m down. That… Okay. That sounds like a game plan.

693 01:18:14.990 ⇒ 01:18:20.209 Uttam Kumaran: Okay, I know the contextual thing and everything, but yeah, I don’t know, this is really fun, I just like chatting about this stuff.

694 01:18:20.210 ⇒ 01:18:24.319 Gabriel Lam: I’m glad, yeah, I think that was fun. I think it also helps me to think about

695 01:18:25.240 ⇒ 01:18:28.789 Gabriel Lam: because I come from a more… like, my initial…

696 01:18:29.460 ⇒ 01:18:34.799 Gabriel Lam: Vision is, like, okay, how do you get things to their ideal state, and, like, what does it look like?

697 01:18:34.960 ⇒ 01:18:36.230 Gabriel Lam: Versus…

698 01:18:36.730 ⇒ 01:18:54.859 Gabriel Lam: you know, what’s the minimum thing that people need? And sometimes it’s… it’s a mix, right? Sometimes it’s like, okay, you know, we hope that everything is on this platform in this very beautiful, like, system that takes the best of everyone else’s worlds, but sometimes it’s like, hey, like, what… what works for people in…

699 01:18:55.320 ⇒ 01:19:00.580 Gabriel Lam: Like, on the… on the ground, yeah.

700 01:19:01.240 ⇒ 01:19:01.690 Uttam Kumaran: Yeah.

701 01:19:01.690 ⇒ 01:19:02.980 Gabriel Lam: Yeah, so…

702 01:19:06.430 ⇒ 01:19:10.010 Uttam Kumaran: Okay, alright, dude, I’m gonna get some food, and then I gotta log back in, so…

703 01:19:10.010 ⇒ 01:19:13.260 Gabriel Lam: Alright, sounds good. I’ll catch you tomorrow, I guess.

704 01:19:13.260 ⇒ 01:19:14.620 Uttam Kumaran: Okay, thank you, Gary.

705 01:19:14.620 ⇒ 01:19:15.620 Gabriel Lam: Have a good night.

706 01:19:15.620 ⇒ 01:19:16.670 Uttam Kumaran: Okay, bye.

707 01:19:16.670 ⇒ 01:19:17.250 Gabriel Lam: Bye.