Meeting Title: Brainforge <> Contextual: Bi-Weekly Catchup Date: 2026-03-11 Meeting participants: Uttam Kumaran, Abhishek Varma, Mike Klaczynski
WEBVTT
1 00:00:08.230 ⇒ 00:00:09.660 Abhishek Varma: Hey, thumb.
2 00:00:09.860 ⇒ 00:00:11.019 Uttam Kumaran: Hey, how are you?
3 00:00:11.770 ⇒ 00:00:13.440 Abhishek Varma: Doing well, doing well.
4 00:00:14.950 ⇒ 00:00:16.190 Uttam Kumaran: What’s the latest?
5 00:00:19.070 ⇒ 00:00:26.000 Uttam Kumaran: Nothing much, I’m actually taking next week off, because I’m gonna go see… Nice! No.
6 00:00:26.640 ⇒ 00:00:27.629 Uttam Kumaran: Oh, really?
7 00:00:27.870 ⇒ 00:00:29.770 Abhishek Varma: Yeah, I managed to get tickets.
8 00:00:29.770 ⇒ 00:00:35.390 Uttam Kumaran: Dude, how many… how many Arsenal fans are in the UK versus not in the UK?
9 00:00:35.570 ⇒ 00:00:40.259 Uttam Kumaran: everybody I know, like, I have 5 Arsenal friends here in Austin, and I’m like.
10 00:00:41.540 ⇒ 00:00:46.120 Uttam Kumaran: Is it… is it… I wonder if the fan base is bigger outside than it is.
11 00:00:46.120 ⇒ 00:00:51.430 Abhishek Varma: No, Arsenal… Arsenal are massive, and I mean, you’re in your early 30s? How old are you?
12 00:00:51.430 ⇒ 00:00:52.660 Uttam Kumaran: I’m 29.
13 00:00:52.660 ⇒ 00:01:04.700 Abhishek Varma: 29, okay, yeah. So around… you’re a little younger than me, the… our age group around that time, Arsenal were very successful when we… when…
14 00:01:04.700 ⇒ 00:01:07.959 Uttam Kumaran: So, a lot of, like, gifts. Oh, okay, okay.
15 00:01:07.960 ⇒ 00:01:10.909 Abhishek Varma: who were, like, you know, maybe, like, 8, 9 years old in 2004.
16 00:01:10.910 ⇒ 00:01:11.830 Uttam Kumaran: Yeah, yeah, yeah.
17 00:01:11.830 ⇒ 00:01:18.169 Abhishek Varma: Yeah, yeah. That became, like, as they’re choosing a team, Arsenal were good.
18 00:01:18.520 ⇒ 00:01:30.830 Abhishek Varma: And so… That is why amongst millennials, younger millennials, the, the arsenal is very well supported.
19 00:01:31.470 ⇒ 00:01:40.140 Uttam Kumaran: Okay, okay, okay. Interesting. Yeah, I think a lot of my friends that are… there’s also a lot of Arsenal bars here in Austin. Like, there’s not many other…
20 00:01:40.140 ⇒ 00:01:42.529 Abhishek Varma: which ones… I went to school there, which.
21 00:01:42.530 ⇒ 00:01:43.320 Uttam Kumaran: Oh, really?
22 00:01:43.320 ⇒ 00:01:44.180 Abhishek Varma: Yeah, yeah, I mean.
23 00:01:44.180 ⇒ 00:01:48.509 Uttam Kumaran: There’s a, Irish bar here called Beatty Riley’s.
24 00:01:49.510 ⇒ 00:01:50.410 Uttam Kumaran: It’s a.
25 00:01:50.410 ⇒ 00:01:52.340 Abhishek Varma: It’s where?
26 00:01:52.490 ⇒ 00:01:54.450 Uttam Kumaran: In Mueller, it’s, like, in East Austin.
27 00:01:54.450 ⇒ 00:01:55.270 Abhishek Varma: Okay.
28 00:01:56.120 ⇒ 00:01:59.900 Abhishek Varma: I haven’t been here, I’ve been to… oh, looks nice.
29 00:02:00.110 ⇒ 00:02:01.270 Abhishek Varma: Looks like an Irish puppy.
30 00:02:01.270 ⇒ 00:02:03.490 Uttam Kumaran: Kind of like a little suburb Irish bar.
31 00:02:03.490 ⇒ 00:02:03.960 Abhishek Varma: Yum.
32 00:02:03.960 ⇒ 00:02:05.880 Uttam Kumaran: Which one are you… which one are you talking about?
33 00:02:06.210 ⇒ 00:02:09.580 Abhishek Varma: I forget which one I used to go to. I didn’t go.
34 00:02:09.580 ⇒ 00:02:11.070 Uttam Kumaran: Was it near UT?
35 00:02:12.770 ⇒ 00:02:15.590 Uttam Kumaran: There’s a place called A Tavern, that probably is it.
36 00:02:15.590 ⇒ 00:02:17.000 Abhishek Varma: That sounds familiar.
37 00:02:17.000 ⇒ 00:02:25.210 Uttam Kumaran: that’s closer to UT, and I don’t only tell it because my other Arsenal friend, we usually go to one of those two. I mean, most of the time, I don’t wake up
38 00:02:25.660 ⇒ 00:02:30.580 Uttam Kumaran: To go with him. But he usually goes to one of these to go watch.
39 00:02:30.580 ⇒ 00:02:31.310 Abhishek Varma: Yeah.
40 00:02:31.510 ⇒ 00:02:34.709 Uttam Kumaran: The tavern is probably it, because that’s, like, a classic, like…
41 00:02:35.360 ⇒ 00:02:36.040 Abhishek Varma: Yeah.
42 00:02:36.040 ⇒ 00:02:36.950 Uttam Kumaran: bar thing.
43 00:02:36.950 ⇒ 00:02:45.689 Abhishek Varma: I used to just watch at home or, like, at a friend’s place or something, because it was always so early, and especially in college, I’m not… like, I’m hungover on the weekends.
44 00:02:45.690 ⇒ 00:02:49.850 Uttam Kumaran: Yeah, yeah, that’s why I’m like, I don’t know how I’m gonna wake up to see this.
45 00:02:49.850 ⇒ 00:02:50.740 Abhishek Varma: But, like…
46 00:02:50.740 ⇒ 00:02:56.770 Uttam Kumaran: For me, I’m, like, a big… I don’t follow much football, but I follow a lot of… I’m, like, a big Lakers fan.
47 00:02:56.880 ⇒ 00:03:04.539 Uttam Kumaran: Oh, okay, nice. So, I watch a lot of… during the playoffs more these days, I watch, like, almost… I’ll watch every playoff game.
48 00:03:04.540 ⇒ 00:03:05.200 Abhishek Varma: Yeah.
49 00:03:05.200 ⇒ 00:03:09.760 Uttam Kumaran: But… So, like, U.S. soccer sucks.
50 00:03:10.430 ⇒ 00:03:15.350 Uttam Kumaran: You know, and it’s like, I think… I mean, are you going to any World Cup?
51 00:03:15.350 ⇒ 00:03:19.720 Abhishek Varma: Yes, I’m going to 3 World Cup games. No way! Yeah. Oh, yeah.
52 00:03:20.010 ⇒ 00:03:24.709 Abhishek Varma: I managed to get tickets, I even got tickets for, like, contextual. I got tickets.
53 00:03:24.710 ⇒ 00:03:27.170 Uttam Kumaran: Oh, let’s go!
54 00:03:27.830 ⇒ 00:03:28.630 Uttam Kumaran: Hell yeah.
55 00:03:28.630 ⇒ 00:03:29.520 Abhishek Varma: Yeah.
56 00:03:29.520 ⇒ 00:03:30.670 Mike Klaczynski: team, right?
57 00:03:31.370 ⇒ 00:03:34.030 Abhishek Varma: No, bro, do you even know what sport we’re talking about?
58 00:03:34.030 ⇒ 00:03:38.220 Mike Klaczynski: Soccer?
59 00:03:39.350 ⇒ 00:03:40.270 Abhishek Varma: Yes.
60 00:03:40.480 ⇒ 00:03:42.460 Uttam Kumaran: Yes.
61 00:03:42.460 ⇒ 00:03:45.060 Mike Klaczynski: I don’t know. Not American football.
62 00:03:45.060 ⇒ 00:03:48.109 Uttam Kumaran: The World Cup, the only cup.
63 00:03:48.640 ⇒ 00:03:50.270 Uttam Kumaran: Dude, I’m Polish.
64 00:03:50.270 ⇒ 00:03:54.320 Mike Klaczynski: Like, I’ve watched a lot of UEFA and FIFA and all that, I know what it is.
65 00:03:54.320 ⇒ 00:03:54.709 Uttam Kumaran: team of…
66 00:03:54.710 ⇒ 00:03:55.790 Abhishek Varma: Polish player.
67 00:03:57.560 ⇒ 00:03:59.200 Abhishek Varma: Name a Polish player.
68 00:03:59.370 ⇒ 00:04:00.429 Mike Klaczynski: Ali Sade Abbe.
69 00:04:01.050 ⇒ 00:04:02.399 Abhishek Varma: Who is that?
70 00:04:02.590 ⇒ 00:04:06.400 Abhishek Varma: There you go, dude! Alisa? How do you spell?
71 00:04:06.400 ⇒ 00:04:07.640 Mike Klaczynski: Elisa Daba.
72 00:04:08.250 ⇒ 00:04:08.919 Abhishek Varma: How do you spell it?
73 00:04:08.920 ⇒ 00:04:10.929 Mike Klaczynski: He’s like… it’s like 10, 15 years ago, probably.
74 00:04:10.930 ⇒ 00:04:12.930 Abhishek Varma: Oh, bro, I’m talking about current!
75 00:04:15.190 ⇒ 00:04:17.859 Abhishek Varma: One of the best players in the world is Polish.
76 00:04:17.860 ⇒ 00:04:20.570 Uttam Kumaran: Yeah, what’s his name? I know, I forgot his name, bro.
77 00:04:20.570 ⇒ 00:04:22.809 Abhishek Varma: Robert Lewandowski. Oh, yeah.
78 00:04:22.810 ⇒ 00:04:24.130 Mike Klaczynski: Of course. Yeah, yeah, live on the.
79 00:04:24.130 ⇒ 00:04:24.520 Uttam Kumaran: Of course.
80 00:04:24.870 ⇒ 00:04:34.180 Uttam Kumaran: Yeah. Yeah, you can hear the Polish coming out when you… when you… my other… my… one of my close friends is Polish, too. Yeah. And, yeah, that’s awesome.
81 00:04:34.910 ⇒ 00:04:37.600 Abhishek Varma: You’re on Fraud Watch if you don’t know who Lewandowski is, by the way.
82 00:04:40.090 ⇒ 00:04:41.100 Uttam Kumaran: Bowling card.
83 00:04:41.470 ⇒ 00:04:43.539 Abhishek Varma: One of the best players in the world was Polish.
84 00:04:45.530 ⇒ 00:04:52.260 Mike Klaczynski: I… yeah, I… not off the top of my head. My son knows a bunch more. He’s the soccer player in the house.
85 00:04:52.430 ⇒ 00:04:53.310 Abhishek Varma: Nice.
86 00:04:53.650 ⇒ 00:05:12.300 Abhishek Varma: Cool! I was a little concerned that Farzad, we have a, we have a sales, in our sales team, Farzad. We were, pitching BMW, and at the end, he’s like, oh, I’m a big Bayern Munich fan. I really love Peter Neuer, when his name is Manuel Neuer.
87 00:05:12.300 ⇒ 00:05:13.699 Uttam Kumaran: I was like, oh, no.
88 00:05:13.700 ⇒ 00:05:15.299 Abhishek Varma: Oh, keep up on!
89 00:05:21.110 ⇒ 00:05:21.530 Uttam Kumaran: Brutal.
90 00:05:21.950 ⇒ 00:05:32.749 Uttam Kumaran: I’d be like, dude, you’re really… we’re at the end of the line here. In the beginning of the meeting, and honestly, just say the last name.
91 00:05:32.750 ⇒ 00:05:33.150 Abhishek Varma: Yeah.
92 00:05:33.150 ⇒ 00:05:34.859 Uttam Kumaran: What are the last names?
93 00:05:34.860 ⇒ 00:05:35.210 Abhishek Varma: Yeah.
94 00:05:39.480 ⇒ 00:05:42.729 Abhishek Varma: Cool, so, how are you guys doing with Contextual?
95 00:05:43.080 ⇒ 00:05:52.950 Uttam Kumaran: Good, yeah, I mean, I think, so one, we’ve… we’ve been pitching a lot of the demos that we built on the insurance. A couple of other things that we’re starting to do on the marketing side.
96 00:05:53.030 ⇒ 00:06:06.640 Uttam Kumaran: We’re starting, like, a kind of a webinar series where we’re pitching, sort of, several vendors, so we have one planned, around, like, demonstrating contextual capabilities when it comes to, like, document search.
97 00:06:06.870 ⇒ 00:06:09.890 Uttam Kumaran: So that’s been another way where we’re, like.
98 00:06:10.050 ⇒ 00:06:22.559 Uttam Kumaran: And I feel like we… we’ve done some events, but, like, we’re… just to continue to have something weekly, we’re doing, like, kind of a weekly Pulse webinar where we’re showing different solutions. We’re also starting to go to market with…
99 00:06:22.700 ⇒ 00:06:26.269 Uttam Kumaran: More solutions built on top of, like, open work.
100 00:06:26.370 ⇒ 00:06:28.259 Uttam Kumaran: Because… What is the open world?
101 00:06:28.390 ⇒ 00:06:32.890 Uttam Kumaran: Open work is, like, an OSS version of co-work, Claude Co-work.
102 00:06:32.890 ⇒ 00:06:33.430 Abhishek Varma: Okay.
103 00:06:34.220 ⇒ 00:06:43.120 Uttam Kumaran: And internally, like, at Brainforge, we built a little bit of, like, you know, a context layer, you know, on top of, like.
104 00:06:43.230 ⇒ 00:06:56.400 Uttam Kumaran: log files on top of transcripts with, like, MCPs and CLIs and the different things. And so we have some clients that we’re actually engaging with that are interested in,
105 00:06:56.860 ⇒ 00:07:05.950 Uttam Kumaran: basically part of… mostly interested in, like, that layer getting established, but also interested in, like, almost like a CloudBot-style environment on top of that.
106 00:07:06.250 ⇒ 00:07:19.959 Uttam Kumaran: Where you have, like, skills or, like, things that can get… can run. And so, we kind of have, like, a stack ready to go that we’re pitching to people, that involves Contextual and a few other vendors that sort of power that experience.
107 00:07:20.330 ⇒ 00:07:25.199 Uttam Kumaran: So that’s another thing that we’re, like, we’re basically going after right now.
108 00:07:26.740 ⇒ 00:07:36.899 Uttam Kumaran: I think we’re still, like, hitting some… somewhat of a wall with some folks, where they’re… they’re just now, like, figuring out that, like, Glean or… is not working for them.
109 00:07:36.900 ⇒ 00:07:47.059 Uttam Kumaran: And as we’re also going larger and larger businesses, I think, like, they’re still even further behind. So, like, I feel like some of our challenge is still the same, where
110 00:07:47.230 ⇒ 00:07:49.250 Uttam Kumaran: We have, like, use case-specific
111 00:07:49.460 ⇒ 00:07:58.320 Uttam Kumaran: things, but ultimately, the people are still, like, working on the price point. Like, how do they even value some of these things that we’re pitching?
112 00:07:58.510 ⇒ 00:07:59.380 Uttam Kumaran: But…
113 00:07:59.380 ⇒ 00:08:01.920 Abhishek Varma: Our… what is the main… sorry, what is the.
114 00:08:01.920 ⇒ 00:08:02.290 Uttam Kumaran: Yeah, yeah.
115 00:08:02.290 ⇒ 00:08:08.259 Abhishek Varma: What is the main, issue that people are finding with green?
116 00:08:09.940 ⇒ 00:08:11.529 Uttam Kumaran: It’s expensive.
117 00:08:11.690 ⇒ 00:08:17.400 Uttam Kumaran: Like, this… it’s, like, a pretty expensive tool, and, like, the performance isn’t good.
118 00:08:17.660 ⇒ 00:08:27.139 Uttam Kumaran: And, like, because it’s, like, a software platform, the ability to, like, tune and change things and have control from the engineering perspective isn’t as high.
119 00:08:27.270 ⇒ 00:08:34.510 Uttam Kumaran: But I think they were first market movers, so a lot of enterprises just, like, turn them on. It’s, like, really easy to just turn on.
120 00:08:34.539 ⇒ 00:08:35.900 Abhishek Varma: Like, put in Slack.
121 00:08:36.070 ⇒ 00:08:46.309 Uttam Kumaran: But, like, for example, like, one of the big things I wanted to do is actually try to displace Glean at one of the clients that we’re trying to pitch Contextual into.
122 00:08:46.420 ⇒ 00:08:55.119 Uttam Kumaran: they’re just slow, and, like, it’s a legacy, it’s, like, an old IT business, so they’re, like, it’s an old IT team, so they’re just, like, one thing at a time.
123 00:08:55.390 ⇒ 00:09:00.719 Uttam Kumaran: You know, so I feel like the larger companies we go after.
124 00:09:00.830 ⇒ 00:09:06.559 Uttam Kumaran: Sometimes, unless we have, like, a really good internal engineering champion, they’re still really new.
125 00:09:06.710 ⇒ 00:09:12.060 Uttam Kumaran: But we’re trying to focus primarily on, like, the outcomes, and then…
126 00:09:12.190 ⇒ 00:09:23.779 Uttam Kumaran: typically, we have a ton of governance over, like, what the stack we use is. We use… so, we oftentimes just need to get, like, a deal where we can actually build, like, a robust platform.
127 00:09:23.870 ⇒ 00:09:27.069 Uttam Kumaran: But oftentimes, we’re coming in just to build, like, one piece.
128 00:09:27.090 ⇒ 00:09:41.769 Uttam Kumaran: like, we’re about to sign a deal with a client where we’re building, should actually find it, but we’re building sort of a little bit of an, like, an internal context layer for their marketing team, where they can use Slack and a few other things to… to, like.
129 00:09:41.770 ⇒ 00:09:57.640 Uttam Kumaran: pull data from a bunch of systems, we’re landing in a data warehouse. But I’m telling that team that, like, the moment that company wants to emerge, or go after more files, or, like, deeper context, like, we should plug Contextual in there. So part of it is just, like.
130 00:09:58.070 ⇒ 00:10:05.509 Uttam Kumaran: like, I don’t know, we’re just… we’re still just trying. I think some of the hurdles have been the same, where a lot of the people we’ve pitched the insurance thing to.
131 00:10:05.720 ⇒ 00:10:11.689 Uttam Kumaran: they’re just, like, not ready, and especially some of the legacy industries, you know?
132 00:10:13.200 ⇒ 00:10:15.749 Abhishek Varma: Yeah, no, I get it.
133 00:10:15.750 ⇒ 00:10:19.670 Mike Klaczynski: Are they saying anything about, like, their data estate being…
134 00:10:20.390 ⇒ 00:10:23.640 Mike Klaczynski: Needing to get fixed or cleaned up before they’re ready?
135 00:10:24.210 ⇒ 00:10:31.720 Uttam Kumaran: Yeah, I mean, and then that’s actually, like, what we’re trying to hinge on as, like, our competitive advantage, because we do a ton of data work.
136 00:10:31.780 ⇒ 00:10:47.389 Uttam Kumaran: And so, in, like, two ways, we’re taking advantage of that. One, for the most part, we talk about this problem as, like, primarily, like, a data engineering problem, and, like, using a tool like Contextual, allows you to just, like, quickly ramp up on that problem and not have to, like.
137 00:10:47.500 ⇒ 00:10:58.489 Uttam Kumaran: Especially, like, especially in certain use cases. And then for us, also, we have existing clients who are starting to pitch our internal, like, again, like, everyone at…
138 00:10:58.670 ⇒ 00:11:12.579 Uttam Kumaran: I’m sure even at Contextual, like, a lot of people are using Kurser or Cloud Code, right? But for the business side, it’s not as easy, because there’s not, like, a GitHub repo-style thing for, like, knowledge work.
139 00:11:12.800 ⇒ 00:11:15.839 Uttam Kumaran: And so we’re trying to pitch that as, like.
140 00:11:16.140 ⇒ 00:11:23.270 Uttam Kumaran: We’re gonna build some system that connects to all those pieces, and then you can actually do retrieval and context over that.
141 00:11:25.650 ⇒ 00:11:33.220 Uttam Kumaran: Yeah, that’s, like, sort of, like, what our… our angle is right now. And so there’s some people in the enterprise that are, like, trying to push for it.
142 00:11:33.450 ⇒ 00:11:36.460 Uttam Kumaran: But I still think that, like, for some… like, there…
143 00:11:36.610 ⇒ 00:11:44.390 Uttam Kumaran: It’s expensive to build these things, and expensive for us to build them, and so whether it’s price… a lot of the people I should just say, like, it’s not the right time.
144 00:11:45.310 ⇒ 00:11:51.210 Uttam Kumaran: Which is actually, like, much more promising, but versus a year ago, it was like, I don’t even get what, like, what this is.
145 00:11:53.070 ⇒ 00:11:54.619 Mike Klaczynski: So they know they have the need.
146 00:11:54.730 ⇒ 00:12:04.020 Mike Klaczynski: Yes. They want to do it, it’s not right now, right? So it’s like, why do anything? Why contextual? Why now? And it’s that final, why now? It’s just not there yet.
147 00:12:04.360 ⇒ 00:12:06.620 Uttam Kumaran: Yeah, because they trust… I mean, ultimately.
148 00:12:07.150 ⇒ 00:12:14.660 Uttam Kumaran: we’re gonna pitch the best tools, and so they… they trust us, and we’re… we’re coming in and saying, like, we’re gonna bring the best stack
149 00:12:14.860 ⇒ 00:12:29.669 Uttam Kumaran: Whether it’s data warehousing, or any piece of that, you know? But ultimately, I think the ROI and the outcome story, like, hasn’t been super, super clear to folks, especially as you go into industries
150 00:12:29.840 ⇒ 00:12:32.690 Uttam Kumaran: where I actually feel it’s the highest impact.
151 00:12:32.870 ⇒ 00:12:38.909 Uttam Kumaran: Like, for these tools, they have the least understanding of, like, what it takes to build them, or, like.
152 00:12:39.360 ⇒ 00:12:49.099 Uttam Kumaran: they don’t… they don’t have an IT or, like, a technical model that says, like, we’re gonna dedicate this budget to unlock this amount of revenue. So when we… when we go to a client like that.
153 00:12:49.520 ⇒ 00:12:56.859 Uttam Kumaran: We’ve oftentimes tried to tie the budget towards, like, what does this unlock in terms of revenue?
154 00:12:56.930 ⇒ 00:13:11.100 Uttam Kumaran: like, there is some part of our work that is fixed, right? Like, if we have to stand up certain tooling, but there is some of our work where we’re almost trying to pitch, like, performance-based, which is, like, if we’re able to help you reduce this thing down, this is how long it takes.
155 00:13:11.190 ⇒ 00:13:27.320 Uttam Kumaran: this is the ROI story, and that, I think, has been really, really positive for folks, versus… I think, typically, when they’re turning on other AI solutions, they… they’re just turning it on because it’s a shiny thing, or they’re, like, turning on Copilot for their team, and they have no path towards measuring.
156 00:13:27.570 ⇒ 00:13:30.279 Uttam Kumaran: And so that’s, like, what we’re trying right now.
157 00:13:30.700 ⇒ 00:13:36.969 Uttam Kumaran: But again, for all the industries, we’re trying to really go after the… still the difficult piece, which is it’s changing.
158 00:13:37.240 ⇒ 00:13:38.890 Uttam Kumaran: is,
159 00:13:40.040 ⇒ 00:13:48.669 Uttam Kumaran: for them to relinquish… like, some of the companies where we’re going for, they’re… I don’t think they’re ever gonna really take time to understand these core LLM concepts.
160 00:13:48.790 ⇒ 00:13:59.719 Uttam Kumaran: But ultimately, they do know that it’s working. They’ve all tried ChatGPT to write something, so they understand that there’s a need, and then for us, we’re like, hey, we want to stitch it together, we want to govern it.
161 00:13:59.950 ⇒ 00:14:03.729 Uttam Kumaran: And then we want to, like, enable it for your team and drive adoption.
162 00:14:04.000 ⇒ 00:14:10.499 Uttam Kumaran: And here are the time… it’s more of a time savings, multiply that time by the hour, and here’s the, like, revenue outcome.
163 00:14:10.700 ⇒ 00:14:15.559 Uttam Kumaran: Versus, typically, it’s like, oh, it’s… this thing costs this thing, and…
164 00:14:15.710 ⇒ 00:14:17.270 Uttam Kumaran: That’s just the way it is.
165 00:14:17.460 ⇒ 00:14:19.099 Uttam Kumaran: Right. You know.
166 00:14:22.050 ⇒ 00:14:24.539 Abhishek Varma: That makes sense. How are you,
167 00:14:24.890 ⇒ 00:14:29.120 Abhishek Varma: How are you actually understanding the…
168 00:14:29.400 ⇒ 00:14:43.910 Abhishek Varma: cost, so that you can drive an ROI argument, because something even we are not very good at when we evaluate a use case is trying to guess how many tokens it will take.
169 00:14:43.910 ⇒ 00:14:44.440 Uttam Kumaran: Yeah.
170 00:14:44.440 ⇒ 00:14:51.930 Abhishek Varma: Trying to forecast the cost, which is a little bit of a challenge, early on in the relationship.
171 00:14:52.330 ⇒ 00:14:57.209 Uttam Kumaran: Yeah, I’m almost trying to see whether there’s something I can share about, like.
172 00:14:57.490 ⇒ 00:14:59.710 Uttam Kumaran: A way we’ve thought about this.
173 00:15:00.300 ⇒ 00:15:05.520 Uttam Kumaran: Maybe this is a good example.
174 00:15:06.280 ⇒ 00:15:07.460 Uttam Kumaran: Just…
175 00:15:10.990 ⇒ 00:15:24.070 Uttam Kumaran: Let’s see… yeah, so, like, okay, here’s a good example. This is, like, a large agency based out of LA called, David and Goliath, and, like, this is a perfect example of some…
176 00:15:24.400 ⇒ 00:15:26.219 Mike Klaczynski: Which models are you? I’m sorry, yep.
177 00:15:26.730 ⇒ 00:15:37.150 Uttam Kumaran: This is a perfect example of, like, a large agency client that can totally afford this, but just, like, 2 days ago said, like.
178 00:15:37.440 ⇒ 00:15:57.250 Uttam Kumaran: They said, like, we’re so slammed, but, like, this is the right directional thing. So basically, like, to give you guys the overview, like, they have all these creative and production systems, they’re trying to, like, they’re copying and pasting things into AI tools, they want… they want one single place where they can interact with AI, which is, like, Teams.
179 00:15:57.880 ⇒ 00:16:05.389 Uttam Kumaran: Or, like, something standalone that, like, can retrieve work and help them execute, modify, or update work, right? So…
180 00:16:05.520 ⇒ 00:16:11.049 Uttam Kumaran: we sort of put together a big scope. I think the biggest thing I want to share is, like, sort of the ROI model. Yeah.
181 00:16:11.610 ⇒ 00:16:19.100 Uttam Kumaran: exactly this, like, at Steady State, with 150 people, at 600 hours per week, like, and so we do some, like.
182 00:16:19.150 ⇒ 00:16:32.489 Uttam Kumaran: we kind of draw the line out, but we… we met with them and tried to, like, understand what are the tasks we’re automating, and really the couple things that we’re attacking are client meeting prep, reef creation, asset retrieval.
183 00:16:32.720 ⇒ 00:16:44.250 Uttam Kumaran: you know, and, like, approval and feedback status across projects. We say, like, here’s the sort of, like, time that those groups are taking to do this thing, and, like, here’s sort of, like.
184 00:16:44.690 ⇒ 00:17:02.399 Uttam Kumaran: the way we think about the ROI story. And so, ultimately, like, for our deals, we’re typically, like, trying to craft it with the person, because for the most part, our deals, like, go one or two ways. Either the person we’re selling to doesn’t get it, and, like, that’s done, or they get it, and they’re like, help me sell it.
185 00:17:02.520 ⇒ 00:17:13.480 Uttam Kumaran: And so we try really hard to give them the ROI story, which is, like, you have 150 people, let’s say they’re just using this, like, well, they’re just saying they’re using it, like, a couple times a week.
186 00:17:13.500 ⇒ 00:17:25.689 Uttam Kumaran: and they save 20 minutes each time, like, bang, like, it’s pretty easy. These are really expensive brand marketing people. You know, and so it’s not only, like, capacity saved, but then
187 00:17:25.740 ⇒ 00:17:29.009 Uttam Kumaran: There’s a potential, like, revenue, they can go get more revenue.
188 00:17:29.270 ⇒ 00:17:43.360 Uttam Kumaran: this is sort of the next thing, which is, like, build capacity recovery, which is, like, hey, if you save them this amount of time, they can go build this much more time. And, like, the numbers are, like… it’s like a complete no-brainer.
189 00:17:43.600 ⇒ 00:17:43.960 Abhishek Varma: Yeah.
190 00:17:43.960 ⇒ 00:17:56.759 Uttam Kumaran: It’s, like, disgusting. Like, they should totally do this. Yeah. And still, like, they said no, but this one, I think, like, continues to prove that we’re getting closer, because we, one, like, I’ll give our team credit, we’re getting better at, like.
191 00:17:56.820 ⇒ 00:18:09.550 Uttam Kumaran: coming not only with, like, all the things we’re gonna build, like APIs, MCP, whatever, but then we’re also, like, here’s also the ROI story for you to go, like, this is the screenshot to be, like, why this thing matters, you know?
192 00:18:09.590 ⇒ 00:18:10.810 Abhishek Varma: Yeah.
193 00:18:10.970 ⇒ 00:18:17.359 Uttam Kumaran: So, I guess to put it one way, it’s, like, less about, for us, we think about tokens, we actually think about the unit of work.
194 00:18:17.460 ⇒ 00:18:18.680 Uttam Kumaran: Getting done.
195 00:18:18.970 ⇒ 00:18:21.330 Uttam Kumaran: And, like, the time savings?
196 00:18:21.670 ⇒ 00:18:29.899 Uttam Kumaran: Which, coming from my world in data, it’s really hard. I always assumed we’ve actually, like, we’re underselling most of our deals, because, like.
197 00:18:30.010 ⇒ 00:18:39.790 Uttam Kumaran: if you didn’t have visibility into, like, connecting your Salesforce with, like, your Stripe data, with, like, your product analytics, what does that work to you, right? Like.
198 00:18:40.050 ⇒ 00:18:49.710 Uttam Kumaran: it’s probably a lot, but can you even, like, get close to measuring, like, what that is? So you… you kind of have to go towards what is the cost to do, plus some margin.
199 00:18:50.540 ⇒ 00:18:59.350 Uttam Kumaran: with the AI stories, we’re really trying to sell this outcome. One, because it allows us to price better, but two, it’s because, like, I think it’s so obvious, like…
200 00:18:59.580 ⇒ 00:19:14.849 Uttam Kumaran: And I really think that this allows us to anchor towards, like, truly, there is, like, a client brief that happens that’s taking half an hour that you should do in 5 minutes. And we’re gonna accomplish that for you, and we’re gonna tick that box. And, like, you know…
201 00:19:14.850 ⇒ 00:19:19.879 Mike Klaczynski: You can send out 50 client briefs a week and build a much bigger book of business, right?
202 00:19:19.880 ⇒ 00:19:25.419 Uttam Kumaran: Exactly, so the first story is actually just, like, let’s say you just didn’t have to hire more people.
203 00:19:25.420 ⇒ 00:19:26.979 Mike Klaczynski: There’s already a saving.
204 00:19:26.980 ⇒ 00:19:36.380 Uttam Kumaran: Yeah. But let’s say you actually, like, those people went and billed another 2 hours a day, or, like, right? And this is, like, one of the premier firms in LA, like.
205 00:19:37.550 ⇒ 00:19:40.710 Uttam Kumaran: So… But, yeah, but…
206 00:19:40.710 ⇒ 00:19:49.859 Mike Klaczynski: So, what about the cost to operate the platform, right? Like, there’s gonna be tokens required, so what’s the ongoing operational cost of this? How do you present that to them?
207 00:19:49.860 ⇒ 00:19:59.749 Uttam Kumaran: So that is… that is all on the client. So, like, basically, we build all the client’s info. So we’re not selling, like, a… we’re not selling a managed service.
208 00:19:59.870 ⇒ 00:20:05.139 Uttam Kumaran: This is just for… for… for our time to build it. But, like, I…
209 00:20:05.610 ⇒ 00:20:11.690 Uttam Kumaran: I don’t think, given, like, the scope of the things that we’re doing for them, the token usage is, like.
210 00:20:12.050 ⇒ 00:20:13.310 Uttam Kumaran: insane.
211 00:20:13.620 ⇒ 00:20:20.149 Uttam Kumaran: But again, like, what I know is, like, for me, I believe the ROI story, I think we’re understating it.
212 00:20:20.370 ⇒ 00:20:26.790 Uttam Kumaran: So the moment we’re in, in a month where they’re like, oh, this is really working, which is what we’ve done for clients.
213 00:20:27.120 ⇒ 00:20:35.759 Uttam Kumaran: like, yeah, it’s… it’s not a… it’s not a problem. I think for us, it’s like, can we scope something that’s deliverable? Because previously, we were like.
214 00:20:35.850 ⇒ 00:20:47.729 Uttam Kumaran: we’ll bring AI, and then… and then, like, some… it was not… we had to actually get to this granularity to be able to get a scope where we’re like, okay, our… our first part of this scope is, like,
215 00:20:47.980 ⇒ 00:20:50.730 Uttam Kumaran: We’re just gonna come and, like, do a bunch of discovery work.
216 00:20:50.840 ⇒ 00:20:52.599 Uttam Kumaran: And make sure that our, like.
217 00:20:52.850 ⇒ 00:21:05.060 Uttam Kumaran: our next three phases are, like, actually accurate, like, oh, actually, client briefs are way more difficult, or, like, there’s 5 other sources with no APIs that, like, we have to figure out how to do. So, we do that, and then…
218 00:21:05.190 ⇒ 00:21:11.660 Uttam Kumaran: we build… You know, and then we build and try to address Things as they come up.
219 00:21:12.120 ⇒ 00:21:26.050 Mike Klaczynski: Yeah, because it’s kind of like with paper, right? Like, when you go and print something out in these agencies, they don’t think of, like, how much does a sheet of paper cost? Similar here, it’s like, if I’m doing a client brief, and it’s going to cost me, whatever, 3 million tokens, and it costs me 75 cents.
220 00:21:26.050 ⇒ 00:21:34.599 Uttam Kumaran: They don’t think about it at all. No, I disagree, Mike, because if you are computing the cost to do something.
221 00:21:34.960 ⇒ 00:21:44.490 Abhishek Varma: Right? We are not sophisticated enough right now, I’m talking about the industry, to really project
222 00:21:45.070 ⇒ 00:21:48.649 Abhishek Varma: the discounting that is happening in AI.
223 00:21:48.870 ⇒ 00:21:58.960 Abhishek Varma: Right? Sure. Like, for example, Claude, right now, released a product that reviews your PRs, okay?
224 00:21:59.280 ⇒ 00:22:12.549 Abhishek Varma: So every… all the engineers are, like, sold on how good Claude is as a coding model, so they purpose-built a tool to help you with your PRs. It costs $25 per code review.
225 00:22:14.460 ⇒ 00:22:15.140 Abhishek Varma: Okay?
226 00:22:15.140 ⇒ 00:22:15.720 Uttam Kumaran: Yeah.
227 00:22:16.160 ⇒ 00:22:17.010 Abhishek Varma: So…
228 00:22:17.200 ⇒ 00:22:29.689 Abhishek Varma: at a certain scale, if you devoted a senior software engineer, right, at Google scale or whatever, right, and that guy is getting paid $300-400K a year.
229 00:22:30.550 ⇒ 00:22:34.309 Abhishek Varma: You still come out ahead by paying a human being, and so then what’s the point?
230 00:22:36.090 ⇒ 00:22:36.420 Uttam Kumaran: Yeah.
231 00:22:36.420 ⇒ 00:22:39.360 Mike Klaczynski: But I think in this case, we’re not look… we’re not looking at a…
232 00:22:39.640 ⇒ 00:22:42.490 Mike Klaczynski: Price per outcome model, it’s literally tokens, right?
233 00:22:43.030 ⇒ 00:22:43.650 Uttam Kumaran: Yeah.
234 00:22:43.650 ⇒ 00:22:51.539 Mike Klaczynski: if you’ve got your $200 a month Claude license, right, your OAuth that you’re using, then that’s a sunk cost, and if you’re paying the API stuff.
235 00:22:51.790 ⇒ 00:22:56.539 Mike Klaczynski: Most of the models aren’t that bad. I mean, if you start to do vision and audio, then it gets more expensive.
236 00:22:56.540 ⇒ 00:23:03.179 Abhishek Varma: You know, our reasoning, like, all these multi-turn things that we’re working on, it’s very cool.
237 00:23:03.550 ⇒ 00:23:14.770 Abhishek Varma: But we need to, and I’m being open with Uttam here, like, we need to, like, understand, you know, exactly like what you wrote in your brief. I love that you were like, how many concurrent users you have.
238 00:23:15.250 ⇒ 00:23:15.840 Uttam Kumaran: Yeah, yeah.
239 00:23:15.840 ⇒ 00:23:18.099 Abhishek Varma: the throughput is the choke, right?
240 00:23:18.100 ⇒ 00:23:18.530 Uttam Kumaran: Yes.
241 00:23:18.530 ⇒ 00:23:23.209 Abhishek Varma: users run, like, a big job, then from our pricing perspective and our.
242 00:23:23.210 ⇒ 00:23:23.780 Uttam Kumaran: Yes.
243 00:23:23.780 ⇒ 00:23:25.739 Abhishek Varma: perspective, we run into a joke.
244 00:23:28.040 ⇒ 00:23:38.639 Abhishek Varma: I will say one thing, about this whole, thing. From your customers, or your prospective customers, do they have, existing, like.
245 00:23:38.870 ⇒ 00:23:42.629 Abhishek Varma: AWS commit, or something that they’re trying to burn.
246 00:23:42.630 ⇒ 00:23:48.809 Uttam Kumaran: Yeah, so, like, at the top level, yes. So, I’ve talked to Mike before about, like, if we can procure through AWS,
247 00:23:48.920 ⇒ 00:23:56.490 Uttam Kumaran: and, like, have that as an availability, that’s another easy thing. It’s not as, like, the less technical the firm.
248 00:23:56.850 ⇒ 00:24:13.409 Uttam Kumaran: And the smaller, the less they care about it. The more senior, and if they have a robust IT, then yeah, they’re trying to consolidate to one cloud, because they’re getting discounts, or they’re just trying to keep it all in one place. So I think, yeah, it has been something that, like, if available, would be great.
249 00:24:14.020 ⇒ 00:24:20.289 Uttam Kumaran: But it’s not always the case for all of our clients. Like, sometimes they don’t have a strategy.
250 00:24:20.650 ⇒ 00:24:22.689 Mike Klaczynski: We’re on the marketplace, ready to go.
251 00:24:22.900 ⇒ 00:24:24.120 Uttam Kumaran: Okay, dope.
252 00:24:24.980 ⇒ 00:24:25.700 Uttam Kumaran: Nice.
253 00:24:25.700 ⇒ 00:24:29.000 Abhishek Varma: There you go Easy win.
254 00:24:29.380 ⇒ 00:24:33.949 Abhishek Varma: Yeah. I unlocked $10 billion worth of damn.
255 00:24:35.580 ⇒ 00:24:36.110 Mike Klaczynski: And the prices.
256 00:24:36.110 ⇒ 00:24:36.610 Uttam Kumaran: Yeah, I mean…
257 00:24:36.610 ⇒ 00:24:37.410 Mike Klaczynski: There is…
258 00:24:37.410 ⇒ 00:24:37.780 Uttam Kumaran: Yeah.
259 00:24:37.780 ⇒ 00:24:42.429 Mike Klaczynski: We used to scare them off, so they contact us, because we want to do private offers.
260 00:24:42.690 ⇒ 00:24:43.260 Uttam Kumaran: Cool.
261 00:24:46.050 ⇒ 00:24:48.040 Uttam Kumaran: Yeah, I mean, this is where, like, I think…
262 00:24:48.210 ⇒ 00:24:59.620 Uttam Kumaran: I think because we were able to now isolate, like, the unit of work, I agree with you, Abhishek, like, I think there’s a lot of subsidies happening, but some of these folks are so… these are folks who are two years behind.
263 00:24:59.620 ⇒ 00:25:00.379 Abhishek Varma: Yeah, yeah, yeah.
264 00:25:00.380 ⇒ 00:25:02.920 Uttam Kumaran: They’re not even using basic stuff.
265 00:25:03.180 ⇒ 00:25:06.520 Uttam Kumaran: So, it’s so bad. It’s like… and I’m…
266 00:25:06.730 ⇒ 00:25:14.699 Uttam Kumaran: I can only imagine, like, how insulated you guys are. I’m also very insulated, and even talking to somebody, I’m like, oh my god, they’re, like, 2 years behind.
267 00:25:14.920 ⇒ 00:25:23.700 Uttam Kumaran: So it’s, like, it’s so… it’s, like, even worse, meaning, like, they just… they want to pick someone to help them do anything in this world.
268 00:25:23.700 ⇒ 00:25:24.030 Abhishek Varma: Yeah.
269 00:25:24.030 ⇒ 00:25:31.260 Uttam Kumaran: Some of them are still not even there, and so we’re just trying to be that… the support for them to engage in any AI solution.
270 00:25:31.950 ⇒ 00:25:43.139 Uttam Kumaran: But ultimately, like, the reason for some of our, like, for the data set, we don’t show ROI, like, this is, like, a really robust SOW, because, like, some people just don’t get…
271 00:25:43.350 ⇒ 00:25:53.869 Uttam Kumaran: truly what the AI thing can do, so we have to really boil it down. On the data side, people are familiar with, like, oh, cool, I’m gonna get these Tableau dashboards, I’m gonna get this, like, we don’t have to tell them
272 00:25:54.000 ⇒ 00:25:58.600 Uttam Kumaran: what the ROI is. So there’s some, like, timing mismatch, but…
273 00:25:58.740 ⇒ 00:26:03.000 Uttam Kumaran: I think it’s helpful for you guys to kind of see how we’re trying to do these.
274 00:26:03.360 ⇒ 00:26:07.490 Uttam Kumaran: And yeah, I mean, I think we’re…
275 00:26:07.760 ⇒ 00:26:26.659 Uttam Kumaran: we’re closer and closer, and some of these deals, again, I think we’re… our pricing totally fits, like, where you guys are at, too. So, like, some of these clients, like, we’re not… we’re no longer going after anybody for… for AI stuff that’s in, sort of, like, oh, we want to do something for, like, 10K and move. Like, these are all, like, pretty robust contracts where we’re gonna staff them up and, like, deliver something big, and then
276 00:26:26.990 ⇒ 00:26:33.140 Uttam Kumaran: like, again, I think for the most part, a lot of these, the bigger players, they just want to pick something that works, finally, beyond…
277 00:26:33.300 ⇒ 00:26:37.869 Uttam Kumaran: Copilot, or, like, or ChatGPT, you know?
278 00:26:37.870 ⇒ 00:26:38.420 Mike Klaczynski: Yeah.
279 00:26:38.760 ⇒ 00:26:43.800 Mike Klaczynski: Yeah, me and Abhishek had a call with EY earlier today for their internal use cases.
280 00:26:43.810 ⇒ 00:27:01.689 Mike Klaczynski: Same thing, they’re like, we’re an Azure shop using Copilot, and, like, the business person’s kind of trying to, you know, buffer it, and the technical folks are like, can we get access to this today? Like, can we start working on this today? Because they, like, they understood the power of how quickly they’re able to build the agents, and the level of detail of being able to connect all this, so…
281 00:27:01.690 ⇒ 00:27:02.170 Uttam Kumaran: Yes.
282 00:27:02.170 ⇒ 00:27:07.800 Mike Klaczynski: EY, they’d have access to the biggest and brightest minds in the world. And…
283 00:27:07.800 ⇒ 00:27:08.980 Abhishek Varma: You’d be surprised.
284 00:27:10.820 ⇒ 00:27:13.239 Mike Klaczynski: Well, no offense, but you used to work there, right?
285 00:27:13.240 ⇒ 00:27:13.650 Abhishek Varma: Yes!
286 00:27:13.650 ⇒ 00:27:14.440 Uttam Kumaran: Exactly.
287 00:27:14.440 ⇒ 00:27:15.829 Mike Klaczynski: That’s why you left, but that’s why.
288 00:27:15.830 ⇒ 00:27:31.710 Uttam Kumaran: No, I have some… I have some friends and ex-EY friends, and they… they tend to say the same thing, but that’s… I also… that’s so crazy to me to hear, but I also understood, like, look, a firm like that, it’s so big that rock-the-boat type people, I don’t think they, like, really want
289 00:27:32.000 ⇒ 00:27:35.109 Uttam Kumaran: Internally, they kind of want to just keep things going.
290 00:27:35.430 ⇒ 00:27:37.130 Uttam Kumaran: Yeah, and…
291 00:27:37.130 ⇒ 00:27:47.319 Abhishek Varma: You know, it’s also a question for the… like, just to give a little bit of grace to the business person, right? Their directive is time to value.
292 00:27:47.860 ⇒ 00:27:53.060 Abhishek Varma: And if they have to go through a big procurement process, then
293 00:27:53.300 ⇒ 00:27:55.839 Abhishek Varma: That’s just more work for them, and a lot of.
294 00:27:55.840 ⇒ 00:27:56.270 Uttam Kumaran: That’d be.
295 00:27:56.270 ⇒ 00:28:02.489 Abhishek Varma: who are working at EY, like, even the partner, like, especially the partners, they’re, like.
296 00:28:02.630 ⇒ 00:28:06.909 Abhishek Varma: they just want to make the money and retire, right? They’re not, like, investing.
297 00:28:06.910 ⇒ 00:28:09.650 Uttam Kumaran: So the government’s government, almost pseudo-government.
298 00:28:10.050 ⇒ 00:28:10.700 Abhishek Varma: Yeah.
299 00:28:11.220 ⇒ 00:28:21.480 Abhishek Varma: I want to get my money and my equity in EY, and then retire in the next 2-3 years. I’m not here to, like, establish some omega tech stack agent.
300 00:28:21.480 ⇒ 00:28:23.549 Uttam Kumaran: So, like…
301 00:28:23.550 ⇒ 00:28:26.889 Abhishek Varma: Microsoft already has a deal with us, let’s just…
302 00:28:27.250 ⇒ 00:28:30.080 Uttam Kumaran: Turn on whatever stage.
303 00:28:30.530 ⇒ 00:28:34.870 Abhishek Varma: Yeah. And you’ll… I mean, this is the case with a lot of big enterprises.
304 00:28:35.110 ⇒ 00:28:38.810 Abhishek Varma: The decision maker is, like, trying to, like, get value as soon as possible.
305 00:28:38.810 ⇒ 00:28:43.309 Uttam Kumaran: But the noise is increasing for them, because I think their business model’s getting disrupted.
306 00:28:43.310 ⇒ 00:28:43.890 Abhishek Varma: Yeah.
307 00:28:43.890 ⇒ 00:28:49.469 Uttam Kumaran: You know, and their clients are asking for stuff, and they’re not able… they’re like.
308 00:28:49.670 ⇒ 00:28:52.419 Uttam Kumaran: Stuck still doing, like, co-pilot deployments, so…
309 00:28:52.730 ⇒ 00:28:53.790 Abhishek Varma: Yeah, I’m…
310 00:28:53.790 ⇒ 00:28:55.130 Uttam Kumaran: It’s getting louder.
311 00:28:55.130 ⇒ 00:29:02.330 Abhishek Varma: I know, like, YC is, like, backing a lot of startups that are, like, AI first.
312 00:29:02.330 ⇒ 00:29:19.089 Uttam Kumaran: That’s basically what our company is. So, like, this whole SOW we wrote mostly with AI, but, like, this is why, like, a company like ours, because we’re a services-first company, but then, like, we sort of are eating the inside with AI,
313 00:29:19.100 ⇒ 00:29:25.400 Uttam Kumaran: our product is still a service, which is still very hard. Like, service ultimately is, like, the meaning, and the trust and the relationship.
314 00:29:25.860 ⇒ 00:29:42.389 Uttam Kumaran: But it… but it’s a hard business model to automate, to get a foothold in, but for them, they’re the bill… it’s just the billable hour is starting to get degraded, which is why, and I think, like, you just hit on that, why do I have an outcome story? Because I want to price for the outcome. I don’t want to price the hour, because
315 00:29:42.390 ⇒ 00:29:46.360 Uttam Kumaran: Who knows how long it’s gonna take me to do this today, or in 6 months?
316 00:29:46.750 ⇒ 00:29:47.230 Abhishek Varma: Correct.
317 00:29:47.230 ⇒ 00:30:01.250 Uttam Kumaran: I know it’s gonna be faster, I bet you it’s gonna be faster. I’m like, why am I gonna lose because of that, right? And so, that’s also why we try to show an ROI outcome story, selfishly for us, but actually the AI
318 00:30:01.550 ⇒ 00:30:07.999 Uttam Kumaran: the AI services, the way we sell it, actually necessitates that, too. Otherwise, they’re not getting it at all, like…
319 00:30:08.190 ⇒ 00:30:15.770 Uttam Kumaran: Because I couldn’t… it’s not a… it’s not about how many hours… some of this stuff is also a frontier. You don’t know how long things are gonna take to get perfect.
320 00:30:15.910 ⇒ 00:30:19.340 Uttam Kumaran: You know, in, like, 6 months, we may have to change things.
321 00:30:19.970 ⇒ 00:30:22.149 Uttam Kumaran: So you can’t do that, yeah.
322 00:30:22.440 ⇒ 00:30:33.759 Abhishek Varma: And I’ll add that, you know, there’s a world where you… your firm becomes, like, the consultants are AI agents.
323 00:30:34.500 ⇒ 00:30:35.090 Uttam Kumaran: Yeah.
324 00:30:35.490 ⇒ 00:30:45.249 Abhishek Varma: You know, that’s, like, you’re a services business, right? And the service is being done, you’re, like, quarterbacking the deal-making, but a lot of the work is being done by agents.
325 00:30:45.250 ⇒ 00:30:47.550 Uttam Kumaran: Well, that’s my bet, is that, like.
326 00:30:48.060 ⇒ 00:30:52.250 Uttam Kumaran: I don’t think… like, I have a lot of some friends or kind of colleagues that are…
327 00:30:52.540 ⇒ 00:30:57.489 Uttam Kumaran: they’re like, oh, we’ve built this AI agent, let’s give it to our clients. I’m like, no, no, no, now you’re a product company.
328 00:30:57.640 ⇒ 00:31:11.660 Uttam Kumaran: very different game. Right. Everything in my business, like, goes through a human filter, and we’re not even, like, that automated. Like, we’re doing really well, but, like, because still, the person’s hiring us because it’s someone on the phone, they can
329 00:31:12.010 ⇒ 00:31:13.970 Uttam Kumaran: Blame, ultimately.
330 00:31:13.970 ⇒ 00:31:14.730 Abhishek Varma: Yeah, yeah, yeah.
331 00:31:14.730 ⇒ 00:31:17.480 Uttam Kumaran: Or, like, someone you can hold accountable, like, you can’t…
332 00:31:17.700 ⇒ 00:31:23.830 Uttam Kumaran: relinquish some of this… I don’t have some timeline, maybe, but, like, I think so for now. But you’re right, like.
333 00:31:24.000 ⇒ 00:31:28.729 Uttam Kumaran: I want to spin up a data model faster than I did last year, and like, that’s…
334 00:31:28.880 ⇒ 00:31:32.500 Uttam Kumaran: That’s the cost of goods in my business, is, like, the time it takes to do that.
335 00:31:32.500 ⇒ 00:31:34.980 Abhishek Varma: You know, so we’re… we try to move towards, like.
336 00:31:34.980 ⇒ 00:31:51.900 Uttam Kumaran: subscription models for a fractional team, or, like, performance-based for an outcome. Like, for example, if we can get, like, if we can get your client briefs down to this, then I want X percent… I want, like, X dollars per client brief generated. Like, I can start to, like.
337 00:31:52.300 ⇒ 00:31:53.949 Uttam Kumaran: Play games like that?
338 00:31:55.330 ⇒ 00:32:01.909 Uttam Kumaran: And it’s like a win-win. It’s like, I’m actually isolating a true cost, and I’m finding a path to reducing it.
339 00:32:02.020 ⇒ 00:32:03.910 Uttam Kumaran: You know,
340 00:32:04.130 ⇒ 00:32:17.170 Uttam Kumaran: it seems more fair. I just think this industry and consulting is really cursed by, like, the billable hour, and it’s sort of just, like, a relic of something everybody just does. Because everybody does it, everybody buys that way.
341 00:32:17.450 ⇒ 00:32:20.710 Uttam Kumaran: And so, it’s sort of like… Tough, yeah.
342 00:32:20.710 ⇒ 00:32:22.979 Abhishek Varma: Are you selling mainly to consulting companies?
343 00:32:23.070 ⇒ 00:32:33.229 Uttam Kumaran: No, we do have… we have sold, we have some people that are agencies, but it’s mostly, like, we have e-com, some, like, traditional businesses.
344 00:32:33.290 ⇒ 00:32:45.540 Uttam Kumaran: sort of a mix. I like selling to professional services because we can just… we know that really well, and, like, all the AI stuff that we did for ourselves, we’re probably, like, a year ahead of, like.
345 00:32:45.800 ⇒ 00:32:50.330 Uttam Kumaran: what they even are asking for, or longer. So it’s like, oh, I’ll…
346 00:32:50.330 ⇒ 00:32:51.050 Abhishek Varma: Yeah.
347 00:32:51.520 ⇒ 00:32:55.550 Abhishek Varma: Yeah, like, super, super, super well. Yeah.
348 00:32:55.550 ⇒ 00:33:02.730 Uttam Kumaran: And so I think we’re gonna try to go further there, for sure. And professional service is so broad, right? So, like, I think there’s a lot there, but…
349 00:33:03.490 ⇒ 00:33:04.599 Uttam Kumaran: I don’t know, I think so…
350 00:33:04.600 ⇒ 00:33:07.389 Abhishek Varma: Small shops, right? There are so many small shops that…
351 00:33:07.390 ⇒ 00:33:13.330 Uttam Kumaran: But yeah, the problem is it’s just always such a mess, like, consulting companies are a big mess, and
352 00:33:13.770 ⇒ 00:33:14.570 Uttam Kumaran: Like…
353 00:33:16.120 ⇒ 00:33:26.869 Uttam Kumaran: I don’t know, it’s… it’s… we have to go… we… our… the echelon of companies that we have to go sell these to has to be, like, pretty large. The smaller consulting companies, they’re typically just trying to stay afloat, because…
354 00:33:26.870 ⇒ 00:33:30.930 Abhishek Varma: It’s a big, like, cash flow swing cycle. It’s a huge…
355 00:33:30.930 ⇒ 00:33:44.519 Uttam Kumaran: It’s typically not very high margin until you, like, grow, and so oftentimes they don’t invest in a lot of technology, and, like, I think because of my background is in product, and I’m an engineer, we run Brainforge kind of like a product company.
356 00:33:44.640 ⇒ 00:33:47.459 Uttam Kumaran: Where there’s a lot of interaction across teams.
357 00:33:47.700 ⇒ 00:34:03.130 Uttam Kumaran: even though, like, one client just interacts with, like, two, three people, internally, there’s a… everybody knows everybody. That’s just because, like, that’s the teams I’m used to, like, working with. But that actually worked out really well for this model, where we have a lot of shared contacts, and so ultimately.
358 00:34:03.180 ⇒ 00:34:13.260 Uttam Kumaran: this shared context layer where, for example, someone can say, when’s the last time we, like, did contextual for a client? What was our process? Where did we get tripped up?
359 00:34:13.340 ⇒ 00:34:18.009 Uttam Kumaran: how should I draft this SOW to, like, reflect that? It’s not possible.
360 00:34:18.139 ⇒ 00:34:24.110 Uttam Kumaran: That’s all, like, knowledge in someone’s head, or someone leaves, and it goes, and… You know?
361 00:34:24.650 ⇒ 00:34:31.759 Uttam Kumaran: So, that’s how we’re thinking about it, and then yes, exactly, like, we can go to a client like D&G and say, like, here’s how we’re accelerating
362 00:34:31.920 ⇒ 00:34:37.340 Uttam Kumaran: Like, how we prep for meetings, or how we put together, like, briefs or assets, you know?
363 00:34:37.580 ⇒ 00:34:38.230 Abhishek Varma: Yeah.
364 00:34:40.199 ⇒ 00:34:48.940 Abhishek Varma: Great. I, I totally agree. Let’s, let’s definitely, like, I think what you are doing is really cool, and, you know, would love to share.
365 00:34:48.949 ⇒ 00:34:54.129 Uttam Kumaran: I just wish I could sell more contextual, I, like, we’re so close, it’s, like, it’s so sad that, like.
366 00:34:54.329 ⇒ 00:35:02.399 Uttam Kumaran: some of these companies, I just think as long as we wait it out, like, we’re gonna be there. I don’t usually… I’m very mostly pessimistic when it comes to sales, but, like.
367 00:35:02.929 ⇒ 00:35:16.349 Uttam Kumaran: seeing some of the ways we’re able to price now, and how far these conversations are getting, we’re a lot closer, especially, like, the higher we go. The noise is so big. I still think, though, that, like, fundamentally, they’re not gonna understand the way it works.
368 00:35:16.719 ⇒ 00:35:18.689 Uttam Kumaran: Which is, like, good and bad.
369 00:35:18.889 ⇒ 00:35:24.199 Uttam Kumaran: But, like, we’re… We’re figuring it out, you know?
370 00:35:24.659 ⇒ 00:35:26.549 Uttam Kumaran: Yeah.
371 00:35:26.759 ⇒ 00:35:32.079 Uttam Kumaran: But yeah, I mean, I think I would love to, like, have Luke on my team share, like, how we’re doing this web. We’re actually…
372 00:35:32.219 ⇒ 00:35:41.009 Uttam Kumaran: doing, like, a three-part webinar on professional services and agencies. So, like, we presented, like, we had a couple of people that ran some big agencies on last week.
373 00:35:41.029 ⇒ 00:35:53.079 Uttam Kumaran: it was a good discussion about, like, using AI for various parts in professional services. And so, like, I think we’re gonna highlight Omni is another partner of ours where we do a lot of business intelligence work.
374 00:35:53.119 ⇒ 00:36:01.469 Uttam Kumaran: With their tool, and so we’re gonna highlight a lot of our partners in that, and like, with demos and vibe-coded prototypes that we bring to the table.
375 00:36:01.829 ⇒ 00:36:08.339 Uttam Kumaran: So maybe I can have Luke just send a little bit of, like, a recap of the one we just did, and, like, I mean, we should just totally do one, it’s like…
376 00:36:08.569 ⇒ 00:36:09.539 Uttam Kumaran: So cheap.
377 00:36:10.320 ⇒ 00:36:10.650 Abhishek Varma: Yeah.
378 00:36:10.650 ⇒ 00:36:11.560 Mike Klaczynski: Yeah, so…
379 00:36:15.340 ⇒ 00:36:15.900 Abhishek Varma: Right?
380 00:36:16.020 ⇒ 00:36:17.340 Abhishek Varma: That sounds awesome.
381 00:36:17.640 ⇒ 00:36:19.060 Uttam Kumaran: Yeah, awesome.
382 00:36:19.220 ⇒ 00:36:23.830 Uttam Kumaran: Thank you for the time, guys. Yeah, and let me know if, like, if anything from this conversation, like.
383 00:36:23.960 ⇒ 00:36:30.760 Uttam Kumaran: is interesting, like, I’m happy to, you know, even send you guys a version of this, but, like, this is how we’re, like, our pitch is evolving, too, so…
384 00:36:30.920 ⇒ 00:36:32.309 Abhishek Varma: Yeah, and funeral…
385 00:36:32.310 ⇒ 00:36:34.979 Mike Klaczynski: to share that, that, SOW, that’d be cool.
386 00:36:36.180 ⇒ 00:36:36.860 Uttam Kumaran: Cool.
387 00:36:36.860 ⇒ 00:36:40.270 Abhishek Varma: Is your team in Austin, by the way?
388 00:36:40.270 ⇒ 00:36:43.969 Uttam Kumaran: Kind of… it’s… no, we have, like, 4 or 5 people in LA.
389 00:36:44.160 ⇒ 00:36:45.389 Uttam Kumaran: Sort of scattered.
390 00:36:45.570 ⇒ 00:36:47.509 Uttam Kumaran: Okay. Some people in, like, New York.
391 00:36:47.720 ⇒ 00:36:50.569 Uttam Kumaran: Okay. Like, yeah, just, like, everywhere.
392 00:36:50.870 ⇒ 00:36:59.560 Uttam Kumaran: We have one… we have… I would love to hire more people here in Austin for some reason. We hired a lot of people in LA. I’m a Lakers fan, I don’t really like LA that much.
393 00:36:59.560 ⇒ 00:37:00.580 Abhishek Varma: Yeah, at least so.
394 00:37:00.920 ⇒ 00:37:06.329 Uttam Kumaran: But I, whatever, so, yeah, just, like, wherever I can…
395 00:37:06.330 ⇒ 00:37:08.159 Abhishek Varma: JWT, man, what are you doing?
396 00:37:08.450 ⇒ 00:37:14.319 Uttam Kumaran: Huh? I know, well, dude, you connect me, connect me to… I actually, I judged a hackathon there.
397 00:37:14.430 ⇒ 00:37:17.380 Uttam Kumaran: I judged a… no, not a hackathon, a,
398 00:37:17.590 ⇒ 00:37:27.190 Uttam Kumaran: it was, like, a senior design for, engineering there, but I would love… yeah, connect me if you… if you have any contacts. I would love to… we’re interviewing for a bunch of people.
399 00:37:27.190 ⇒ 00:37:27.770 Abhishek Varma: Several?
400 00:37:28.490 ⇒ 00:37:33.869 Uttam Kumaran: Yeah, I mean, if they’re… especially on the AI side, if they’re, like, if they’re crafty.
401 00:37:33.870 ⇒ 00:37:34.670 Abhishek Varma: Yeah.
402 00:37:35.480 ⇒ 00:37:46.149 Abhishek Varma: I went to… I went to the business school for the MIS, right, which is half tech, half business. I still, I think, have the email address of some of the professors I can put you in touch with.
403 00:37:46.150 ⇒ 00:37:57.179 Uttam Kumaran: Please, yeah, I would love nothing more than to hire locally. It’s just sometimes really tough to take… some people just have so little background, and we’re a client service that I’m like, dude, I’m gonna get…
404 00:37:57.310 ⇒ 00:38:09.429 Uttam Kumaran: smashed if I put this person in front of a client. But some people really, like, they got it, or they have the sauce, and I’m like, I’ll just show you how to do the AI stuff. Especially on the AI side, I’m much more willing to take risks, because it’s super new.
405 00:38:09.430 ⇒ 00:38:09.980 Abhishek Varma: Yeah.
406 00:38:09.980 ⇒ 00:38:17.619 Uttam Kumaran: And a lot of our stuff is building the systems. You don’t need a super deep technical knowledge, and folks will get it.
407 00:38:17.620 ⇒ 00:38:28.219 Abhishek Varma: That is tough. I think for your business, the MIS, like, my program is exactly right, because the technical depth is nowhere near a computer science degree, but…
408 00:38:28.220 ⇒ 00:38:34.120 Uttam Kumaran: a lot of… It’s like, can you speak to the next hardest technical thing, or at least, like, show that your path to, like, if you spent…
409 00:38:34.260 ⇒ 00:38:38.150 Uttam Kumaran: a week learning about LLMs, like, you’d at least, like, Correct.
410 00:38:38.150 ⇒ 00:38:40.780 Abhishek Varma: I’m more looking for aptitude. Yeah, yeah, yeah.
411 00:38:40.780 ⇒ 00:38:42.259 Uttam Kumaran: Yeah, yeah, yeah, yeah.
412 00:38:42.990 ⇒ 00:38:46.400 Abhishek Varma: Yeah, I’ll do that for you. Let me go and find.
413 00:38:47.200 ⇒ 00:38:47.910 Uttam Kumaran: Amazing.
414 00:38:47.910 ⇒ 00:38:48.550 Abhishek Varma: Yeah.
415 00:38:49.140 ⇒ 00:38:52.080 Abhishek Varma: Clint Tuttle, I’ll… I’ll hit him up for you.
416 00:38:52.400 ⇒ 00:38:53.090 Uttam Kumaran: Okay.
417 00:38:53.090 ⇒ 00:38:53.740 Abhishek Varma: Cool.
418 00:38:55.160 ⇒ 00:38:56.399 Abhishek Varma: Awesome, man.
419 00:38:56.950 ⇒ 00:38:58.240 Uttam Kumaran: Thank you both for the time.
420 00:38:58.240 ⇒ 00:38:59.960 Mike Klaczynski: Thanks, Utah, always a pleasure.
421 00:38:59.960 ⇒ 00:39:01.720 Uttam Kumaran: Yeah, of course. Talk to you soon.
422 00:39:02.230 ⇒ 00:39:02.720 Abhishek Varma: Bye.