Meeting Title: AI Service Standup Date: 2026-01-16 Meeting participants: Samuel Roberts, Uttam Kumaran, Casie Aviles, Mustafa Raja, Amber Lin, Gabriel Lam


WEBVTT

1 00:00:06.820 00:00:08.000 Samuel Roberts: Hello.

2 00:00:10.700 00:00:11.510 Uttam Kumaran: Thanks, Anna.

3 00:00:13.200 00:00:14.090 Samuel Roberts: How’s it going, man?

4 00:00:18.170 00:00:19.370 Uttam Kumaran: Doing better?

5 00:00:20.080 00:00:22.000 Samuel Roberts: Oop, sounded far away.

6 00:00:22.300 00:00:23.890 Uttam Kumaran: Alright, is it? Are you feeling better?

7 00:00:24.510 00:00:28.500 Samuel Roberts: And I don’t… yeah, I am, I’m not hearing you super well. I don’t know if it’s me or you.

8 00:00:28.800 00:00:32.440 Uttam Kumaran: Wait, hold on. I just got up, but I have my AirPods in.

9 00:00:33.880 00:00:35.880 Samuel Roberts: I must be whatever the mic’s been using.

10 00:00:36.050 00:00:36.700 Uttam Kumaran: Oh.

11 00:00:37.120 00:00:38.550 Samuel Roberts: Yeah, that sounds better.

12 00:00:38.720 00:00:39.480 Uttam Kumaran: Better?

13 00:00:39.670 00:00:40.350 Samuel Roberts: Yes.

14 00:00:40.740 00:00:42.360 Uttam Kumaran: Okay, no, I said, how are you feeling?

15 00:00:42.850 00:00:48.359 Samuel Roberts: Oh, yeah, I’m feeling better. I think what I… it must have been something I ate, because it was a pretty quick,

16 00:00:48.880 00:00:50.669 Samuel Roberts: Through me saying, yeah.

17 00:00:52.490 00:00:55.860 Samuel Roberts: I was like, oh no, is this another bug of some sort, but I think it was just a…

18 00:00:56.280 00:01:00.049 Samuel Roberts: something wasn’t, you know, you know what I mean? Forget it.

19 00:01:00.050 00:01:05.230 Uttam Kumaran: We didn’t know that has kids, and Mike’s girlfriend works with me, and she’s, like, preparedly sick.

20 00:01:05.239 00:01:06.029 Samuel Roberts: That’s what I’m…

21 00:01:06.030 00:01:15.169 Uttam Kumaran: I’m trying to come up with, like, a cocktail of vitamins. Well, I’ll… I need it, because I get sick, so I’m always, like, taking as many,

22 00:01:15.590 00:01:18.319 Uttam Kumaran: Like, weird, like, elder flowers.

23 00:01:18.610 00:01:22.039 Uttam Kumaran: I can see, like, shit as I possibly can.

24 00:01:22.340 00:01:38.919 Samuel Roberts: Yeah, I used to get sick all the time in college, and, like, I looked like I was just, like, not a well person, and then I… I stopped, like, being around people that often, and, like, that many people, and this thing’s going around, and I don’t get sick nearly as much, and now as soon as he starts going to daycare, I’m sure it’s gonna happen again.

25 00:01:38.920 00:01:41.270 Uttam Kumaran: What did you, what did you study in school?

26 00:01:41.670 00:01:43.459 Samuel Roberts: Mechanical engineering, physics.

27 00:01:43.460 00:01:44.160 Uttam Kumaran: Mr. Foxy.

28 00:01:44.160 00:01:45.000 Samuel Roberts: science?

29 00:01:45.600 00:01:51.570 Samuel Roberts: So I was in a lot of classes with a lot of people, all of the… you know what I mean? It wasn’t like I was in the same class with everyone, you know, one thing.

30 00:01:52.040 00:01:57.279 Uttam Kumaran: Yeah. How was, how is, Philosophy of Science? How’s the course going?

31 00:01:58.000 00:02:00.669 Samuel Roberts: Oh, that’s over. That was just a few, it was like,

32 00:02:01.850 00:02:06.700 Samuel Roberts: End of November, right before Thanksgiving, I think, was the last one. It was awesome. It was so cool. I…

33 00:02:06.700 00:02:09.370 Uttam Kumaran: Did you… did you ever do a lab chair on it, or no?

34 00:02:09.370 00:02:10.990 Samuel Roberts: No, I could do that, though.

35 00:02:10.990 00:02:14.019 Uttam Kumaran: Yes, dude, please. You know how many people I’ve told me, I was like.

36 00:02:14.250 00:02:19.449 Uttam Kumaran: Well, every time I’m talking about him, like, Sam is, like, the low-key smartest person here, it just doesn’t, like…

37 00:02:19.750 00:02:34.590 Uttam Kumaran: really share it, but he’s, like, in this course called Philosophy of Science, which is, like, like, why would you combine two horror things together like that? But, you know, Greg is a philosophy PhD.

38 00:02:34.930 00:02:38.850 Samuel Roberts: I saw that, actually, yeah, I haven’t chatted with him yet, but I’d be curious to.

39 00:02:38.850 00:02:47.720 Uttam Kumaran: That is also, like, a really smart philosophy, but he was like, dude, they’re not also product analytics. Yeah.

40 00:02:47.860 00:02:50.990 Samuel Roberts: Oh, you’re cutting out a little bit. I don’t know if that’s me, my headphones, or not.

41 00:02:50.990 00:02:55.130 Uttam Kumaran: We were… we were talking about logic proofs… talking about logic proofs yesterday.

42 00:02:55.130 00:02:56.180 Samuel Roberts: Oh, nice.

43 00:02:56.530 00:02:58.249 Uttam Kumaran: Yeah, not all of that.

44 00:02:58.670 00:02:59.720 Samuel Roberts: That’s sick.

45 00:03:00.880 00:03:01.770 Samuel Roberts: Sweet.

46 00:03:01.880 00:03:07.080 Samuel Roberts: Alright, well, welcome everyone. Not getting sick is be a hater and stay home. Yeah, pretty much.

47 00:03:09.410 00:03:10.340 Samuel Roberts: Pretty much.

48 00:03:12.600 00:03:13.800 Samuel Roberts: Cool, cool.

49 00:03:14.530 00:03:16.150 Samuel Roberts: How’s everyone doing this morning?

50 00:03:18.050 00:03:19.260 Gabriel Lam: Good, good.

51 00:03:21.330 00:03:22.160 Samuel Roberts: rent.

52 00:03:22.420 00:03:24.749 Samuel Roberts: I guess let’s just jump right in.

53 00:03:25.770 00:03:30.570 Samuel Roberts: to… ABC, let me get my linear on the right monitor here.

54 00:03:33.230 00:03:40.120 Samuel Roberts: Shame… Alright.

55 00:03:42.970 00:03:46.680 Samuel Roberts: So, how we feeling? What do we got?

56 00:03:46.840 00:03:48.809 Samuel Roberts: How did the meeting go yesterday?

57 00:03:49.020 00:03:51.230 Samuel Roberts: With the, admin.

58 00:03:51.420 00:03:52.320 Samuel Roberts: Do I.

59 00:03:52.320 00:03:55.449 Casie Aviles: I missed it, I got, like, triple buttoned, maybe.

60 00:03:55.450 00:03:56.290 Uttam Kumaran: Go ahead.

61 00:03:57.690 00:04:03.039 Casie Aviles: Yeah, yesterday I just, I did a brief demo for them.

62 00:04:03.170 00:04:04.330 Samuel Roberts: Okay. And…

63 00:04:04.330 00:04:10.510 Casie Aviles: I just walked them through how it works right now, and asked if there’s anything, like, that they wanted.

64 00:04:10.660 00:04:12.830 Casie Aviles: you know, I just ask for feedback, and…

65 00:04:13.120 00:04:20.760 Casie Aviles: If there’s anything that’s not intuitive for them, I think they were generally… Happy with it, but…

66 00:04:21.320 00:04:30.379 Casie Aviles: They haven’t really tested it for themselves, so that’s what we’re pushing it. That’s what we’re pushing for them to do, which is to try it out.

67 00:04:32.460 00:04:36.679 Casie Aviles: Yeah, and they also had, like, a bunch of questions about

68 00:04:36.820 00:04:49.539 Casie Aviles: something that was inefficient before, so I also told them that, you know, I added flows based on the feedback they gave from the first, you know, the forums that we did.

69 00:04:50.300 00:04:50.910 Samuel Roberts: Cool.

70 00:04:50.910 00:04:55.369 Casie Aviles: So… Yeah, I think that was pretty much it.

71 00:04:55.370 00:04:56.459 Samuel Roberts: Okay, great.

72 00:04:56.890 00:05:00.980 Samuel Roberts: Yeah, so I guess we’ll leave this here, then, until they get back to us.

73 00:05:01.170 00:05:04.139 Samuel Roberts: Did we… the standardized service names, this…

74 00:05:06.410 00:05:09.109 Samuel Roberts: What’s the status on this one?

75 00:05:09.650 00:05:16.609 Casie Aviles: Well, for this one, it’s just the database that I made some changes to. It’s not really live yet, because it’s…

76 00:05:16.610 00:05:18.819 Samuel Roberts: Oh, it’s in the dev database, that’s right, okay.

77 00:05:19.100 00:05:23.179 Casie Aviles: Yeah, it’s the same… it’s where the admin UI is pointing to.

78 00:05:23.180 00:05:28.999 Samuel Roberts: Okay, I just… I forgot, yeah, okay, so those are… these are kind of tied together, that’s fine. Okay, and then the,

79 00:05:29.270 00:05:30.329 Samuel Roberts: I’m sorry, go ahead.

80 00:05:31.300 00:05:34.180 Casie Aviles: Yeah, yeah, that’s true, their lives together.

81 00:05:34.360 00:05:40.589 Samuel Roberts: Okay. And then the other ones, adding them to the…

82 00:05:41.490 00:05:44.359 Casie Aviles: I think this one for the mechanical, we’re waiting on…

83 00:05:44.970 00:05:51.510 Casie Aviles: The mechanical department for the commercial, at least for the commercial one. I haven’t done this, so…

84 00:05:51.790 00:05:55.740 Samuel Roberts: Okay, so this one is, blocked for now.

85 00:05:57.880 00:05:58.570 Samuel Roberts: Cool.

86 00:05:58.810 00:06:03.909 Samuel Roberts: Alright, good on that, Mustafa, how are we doing on yours?

87 00:06:05.540 00:06:09.120 Mustafa Raja: Yeah, so for 1569,

88 00:06:09.430 00:06:14.150 Mustafa Raja: It’s done. I’m going to, I’m going to share a room on that, so team can review.

89 00:06:14.540 00:06:18.110 Mustafa Raja: And then I bel… I haven’t shared the loom yet, I’m going to share.

90 00:06:18.110 00:06:20.680 Samuel Roberts: Yep, yeah, yeah, I just don’t see what we had before, yeah. No, you’re fine.

91 00:06:20.680 00:06:28.739 Mustafa Raja: Okay. Okay. So, for 1563, I believe this is… this is good. Okay.

92 00:06:29.470 00:06:42.120 Mustafa Raja: I believe you took a look at that also. And then I believe we would need a new, ticket, to set up this evaluation dataset with a, I believe, testing pipeline or something?

93 00:06:42.120 00:06:42.760 Samuel Roberts: Okay.

94 00:06:44.050 00:06:49.900 Samuel Roberts: Yeah, let’s, let’s… Get together on that today.

95 00:06:49.900 00:06:53.039 Mustafa Raja: Yeah, I don’t know if that is already in the Gantt chart or not.

96 00:06:53.260 00:06:55.249 Samuel Roberts: Yeah, I think, I think there is something.

97 00:06:55.760 00:06:57.830 Samuel Roberts: But yeah, we’ll add a ticket either way.

98 00:06:58.450 00:06:59.080 Mustafa Raja: Hmm.

99 00:06:59.390 00:06:59.980 Samuel Roberts: Okay.

100 00:07:02.110 00:07:02.720 Samuel Roberts: Okay, cool.

101 00:07:02.720 00:07:04.299 Mustafa Raja: Yeah, that’s pretty much it from a cycle.

102 00:07:04.490 00:07:05.060 Samuel Roberts: Great.

103 00:07:05.880 00:07:09.280 Samuel Roberts: And then, Panav is not here right now.

104 00:07:10.600 00:07:13.319 Samuel Roberts: No, okay. We’ll check in with him then.

105 00:07:13.480 00:07:20.519 Gabriel Lam: I will say, on Zoom, this meeting, it says it’s on 11.15, but on calendar, it’s at 10.30, so…

106 00:07:20.520 00:07:23.870 Samuel Roberts: Oh, it doesn’t mean… It doesn’t move good.

107 00:07:23.870 00:07:24.720 Gabriel Lam: fusion.

108 00:07:25.810 00:07:31.600 Samuel Roberts: That’s so… I never use Zoom, I just use the calendar. Let me…

109 00:07:32.790 00:07:40.640 Samuel Roberts: So, that might be a source of confusion. That’s so annoying. Okay, I assumed Zoom just stayed in sync with Google Calendar, but…

110 00:07:41.700 00:07:43.229 Samuel Roberts: Where is that even?

111 00:07:45.870 00:07:48.230 Samuel Roberts: I don’t know, I’ll try to figure that out later then. Okay. Yeah.

112 00:07:49.050 00:07:56.530 Samuel Roberts: Alright, that’s… odd. Okay, where were we? I don’t remember where mine was. Okay.

113 00:07:56.690 00:08:00.459 Samuel Roberts: Alright, cool, so is that good on ABC, then? Any other thoughts ABC-wise? Amber.

114 00:08:04.160 00:08:10.030 Uttam Kumaran: Not really, I would like to talk about evals next week, too, with somebody who’s working on that.

115 00:08:10.320 00:08:11.580 Samuel Roberts: Yeah, I think…

116 00:08:11.580 00:08:13.010 Uttam Kumaran: Stuff and I got…

117 00:08:13.010 00:08:20.850 Samuel Roberts: Mustaf and I are gonna chat today, because we were trying to figure out, what Monster can do, kind of, on its own, versus, like, setting up blank views around it right now.

118 00:08:22.120 00:08:27.690 Samuel Roberts: Because it has some capabilities to do that, but I don’t know if it’s robust enough for on its own.

119 00:08:27.850 00:08:30.429 Samuel Roberts: But we’ll… we’ll sneak on that today and come up with a plan.

120 00:08:30.820 00:08:32.360 Samuel Roberts: And then…

121 00:08:32.360 00:08:34.320 Mustafa Raja: Yeah, I’d love to chat on that, actually.

122 00:08:34.630 00:08:35.150 Samuel Roberts: Yeah.

123 00:08:35.159 00:08:40.329 Uttam Kumaran: Yeah, we should just include all the AI crew, and include Awash and Gabe.

124 00:08:40.770 00:08:41.409 Samuel Roberts: Okay.

125 00:08:42.880 00:08:45.870 Uttam Kumaran: Just, like, we could just do, like, an hour call about evals, because…

126 00:08:46.070 00:08:50.789 Uttam Kumaran: We’ve gone through this, like, process a couple times, and I want to use this for…

127 00:08:51.140 00:08:52.490 Samuel Roberts: internal…

128 00:08:52.790 00:08:55.150 Uttam Kumaran: ABC and… and Lilo.

129 00:08:55.240 00:08:56.640 Samuel Roberts: Yep.

130 00:08:56.980 00:09:01.199 Uttam Kumaran: So… Okay. Yeah, we’ll set something up for next week, then, that’s a good idea.

131 00:09:02.850 00:09:04.080 Samuel Roberts: Cool.

132 00:09:05.510 00:09:12.020 Samuel Roberts: Alright, I think that’s good, ABCwise, then Lilo, any…

133 00:09:12.980 00:09:19.220 Samuel Roberts: Any updates here? I got some stuff sorted out, there were a couple little quick fixes that, I was…

134 00:09:19.580 00:09:28.219 Samuel Roberts: I have a PR in there for… there’s some weird stuff with the PR environments that I was trying to sort out. I tried a few different things yesterday, and nothing really…

135 00:09:28.780 00:09:35.950 Samuel Roberts: was good, but I think we have a good, or at least bad option, probably, for the PR environments. I just gotta get it working right.

136 00:09:37.490 00:09:49.419 Samuel Roberts: But besides that, staging’s pretty good. I think, when I was working on the Slack reports, there were some… it was pulling, like, the wrong data from certain MCPs or something, but… Utam, I saw your question about the…

137 00:09:50.050 00:09:51.840 Samuel Roberts: Calculations in real time.

138 00:09:53.200 00:09:54.310 Samuel Roberts: Was that what you were…

139 00:09:54.970 00:09:58.740 Uttam Kumaran: Yeah, I just was interested in, like,

140 00:09:59.510 00:10:05.350 Uttam Kumaran: Are all of those pre… are we just pulling them straight? Like, it’s… is the app… is the vendor calculating?

141 00:10:05.840 00:10:24.160 Samuel Roberts: Yeah, so I think what… for some of them, it’s… it’s pulling, like, I can… I can show you the, like, daily cost per click, all that sort of stuff. The, like, the things that were combining, like, ads and sales data was being calculated on the fly. So, like, the return on ad spend and…

142 00:10:24.640 00:10:34.049 Samuel Roberts: I forget the other metric offhand, but there were a couple that were Shopify and Medic together calculations, so those are being calculated on the fly. The rest was just data straight from the MCT.

143 00:10:36.320 00:10:37.100 Uttam Kumaran: Okay, okay.

144 00:10:37.840 00:10:44.169 Samuel Roberts: So that could change in the future once we have more of a warehouse, but for now, it’s just pulling the fresh data.

145 00:10:44.230 00:11:01.810 Samuel Roberts: doing the calculations, which is why I think there were some issues with, like, Shopify was still pulling zero data, or zero numbers, and I gotta… I gotta give up front off and figure out where he’s at with that. But I think we’re pretty close on there. The MCP seemed to be working. The, the skills that they asked about…

146 00:11:01.990 00:11:03.789 Samuel Roberts: Super easy to drop right in.

147 00:11:04.130 00:11:09.249 Samuel Roberts: That was really nice, so that they can… they can set up their own…

148 00:11:09.450 00:11:13.050 Samuel Roberts: Claude’s skills in the Anthropic UI, I guess.

149 00:11:13.220 00:11:14.680 Samuel Roberts: And then we can use them.

150 00:11:14.840 00:11:15.900 Samuel Roberts: in Stitch.

151 00:11:16.130 00:11:17.180 Samuel Roberts: Which is nice.

152 00:11:17.470 00:11:22.259 Samuel Roberts: Casey, how are we doing with the, Nano Banana app?

153 00:11:23.480 00:11:28.040 Casie Aviles: So, for Nana Banana, yeah, I just have, like, the…

154 00:11:28.640 00:11:33.250 Casie Aviles: the… I just replicated, you know, the functionalities that they mentioned,

155 00:11:33.740 00:11:43.979 Casie Aviles: I think I just need to, you know, figure out, like, the best way I could tie it together with the existing changes right now. I haven’t…

156 00:11:44.190 00:11:47.710 Casie Aviles: I wasn’t able to work too much on this yesterday, but…

157 00:11:47.710 00:11:48.260 Samuel Roberts: Okay.

158 00:11:48.700 00:11:53.899 Casie Aviles: Yeah, this is what I’m… going to work on today, and…

159 00:11:54.490 00:11:57.369 Casie Aviles: Yeah, I didn’t get a chance to sync with you as well, but…

160 00:11:57.370 00:12:01.920 Samuel Roberts: Yeah, definitely feel free to reach out today, I can… I can…

161 00:12:02.080 00:12:05.969 Samuel Roberts: compare a little bit, or I can… I can try to help get some stuff set up in…

162 00:12:06.350 00:12:10.259 Samuel Roberts: Railway, because we’re going to need a bucket and everything to get all that tied together.

163 00:12:11.490 00:12:15.569 Samuel Roberts: At least for staging and… Production. Okay.

164 00:12:15.860 00:12:21.439 Samuel Roberts: Yeah, feel free to ping me at some point. My day is not too crazy, so I should have some time.

165 00:12:22.760 00:12:26.580 Casie Aviles: Okay. And then, Utam, we have a meeting…

166 00:12:26.580 00:12:33.040 Samuel Roberts: with them at 2.30 Eastern, but we also, you and I have time then, too? Can we reschedule ours?

167 00:12:33.470 00:12:36.679 Samuel Roberts: Yeah, I’ll reschedule ours. Okay.

168 00:12:38.690 00:12:39.970 Uttam Kumaran: to sometime.

169 00:12:40.150 00:12:42.970 Uttam Kumaran: And then I may not make the Lilo meeting.

170 00:12:43.210 00:12:43.830 Samuel Roberts: Okay.

171 00:12:45.310 00:12:56.130 Samuel Roberts: I’m not too… I mean, we’re just to kind of show off what’s currently working and make sure that they’re, you know… they’ve been testing things, I’m less worried about, like, demoing as much, but just kind of a sync with them and…

172 00:12:56.340 00:12:59.850 Samuel Roberts: let them… Like, throw out some ideas, and…

173 00:12:59.850 00:13:06.039 Uttam Kumaran: Yeah, and I’m gonna send the update… I’m gonna send the update right now. Thanks, guys, for sending that yesterday, I just,

174 00:13:06.450 00:13:09.600 Uttam Kumaran: Basically, I have, like, 3 hours of calls after 5, so I just.

175 00:13:09.600 00:13:10.620 Samuel Roberts: Oh, jeez.

176 00:13:10.620 00:13:11.860 Uttam Kumaran: Blocked out after that, so…

177 00:13:11.860 00:13:13.320 Samuel Roberts: Yeah, no, no worries.

178 00:13:14.180 00:13:18.189 Samuel Roberts: Yeah, I think we were pretty on top of it with them yesterday in Slack, so I think they.

179 00:13:18.190 00:13:19.580 Uttam Kumaran: Okay, yeah, yeah, I agree.

180 00:13:19.920 00:13:26.250 Samuel Roberts: So, yeah, I think they… and they said they were just gonna show off the, the…

181 00:13:27.000 00:13:31.529 Samuel Roberts: Not… they weren’t gonna use staging for their meeting yesterday, so they were… we just kept pushing stuff, so…

182 00:13:32.300 00:13:42.800 Samuel Roberts: Should be… should be pretty good. I mean, we might… nanobanana notwithstanding, we could probably push a lot of that stuff into production, but, I was kind of hoping to get everything into one big push, rather than…

183 00:13:42.940 00:13:49.010 Samuel Roberts: wait for another one, but we’ll see how we get today, because I’d love to do that. But I also don’t want to push to production on Friday, but it’s not that big a deal right now, so…

184 00:13:49.010 00:13:49.970 Uttam Kumaran: I forget, yeah.

185 00:13:50.360 00:13:52.350 Samuel Roberts: That’s just my internal, like…

186 00:13:52.350 00:13:54.830 Uttam Kumaran: I don’t know how their call, their all-hands went, you know?

187 00:13:54.830 00:13:57.870 Samuel Roberts: Yeah, yeah, we can find out, today.

188 00:13:59.300 00:14:00.110 Samuel Roberts: Cool.

189 00:14:00.290 00:14:05.219 Samuel Roberts: Any other thoughts there? I think it’s pronounced on here, but,

190 00:14:07.060 00:14:09.060 Samuel Roberts: Can I even push this to done.

191 00:14:09.830 00:14:11.910 Samuel Roberts: I think we’re pretty stable there.

192 00:14:13.890 00:14:15.310 Samuel Roberts: Any other thoughts?

193 00:14:15.610 00:14:18.980 Samuel Roberts: Okay, cool. Let’s jump internal.

194 00:14:19.730 00:14:20.220 Gabriel Lam: Yeah.

195 00:14:20.220 00:14:21.939 Samuel Roberts: Yeah, that’s awesome, you’re stuck, yeah.

196 00:14:21.940 00:14:28.510 Gabriel Lam: Yeah, yeah, yeah. So, first of all, thank you for giving me the Slack notification. I just sent that over to Hannah, and they’re like…

197 00:14:28.510 00:14:34.039 Samuel Roberts: Yeah, sorry about that. I was, I was just, like, thinking about Slack, and I was like, there was something else, and I was like, oh, right, I can.

198 00:14:35.100 00:14:36.280 Uttam Kumaran: What is that for? What is that?

199 00:14:36.280 00:14:38.339 Samuel Roberts: It was the, the case studies.

200 00:14:38.730 00:14:39.650 Uttam Kumaran: Oh, okay.

201 00:14:39.650 00:14:51.490 Gabriel Lam: Yeah, so people would get a notification when they had an interview request, but then, like, a sort of quality of life feature was, like, they had no idea when the interviews were done, and so they would have to go in or refresh or whatever.

202 00:14:51.590 00:14:59.920 Gabriel Lam: And they were just like, is this just a small addition that we could have? And they’ve been asking for that for a couple weeks. So, glad we got around to doing that.

203 00:15:00.090 00:15:14.040 Gabriel Lam: Moving on to cursor integrations, you… I’m sure you saw the Loom basically ran into an issue with Railway having this, like, loop, this authentication loop, where

204 00:15:16.150 00:15:31.619 Gabriel Lam: Yeah, I just asked her to write a bug report, but it was just stuck on validation indefinitely, which I think seemed to be a sort of web-specific issue, so when I just tried to replicate what you did on a desktop client, that worked.

205 00:15:31.760 00:15:39.500 Gabriel Lam: And I also, committed a document to the playbook, so I have multiple computers, and then

206 00:15:39.680 00:15:55.759 Gabriel Lam: I tried it on another computer this morning, and it works well. So that’s sort of the temporary fix that I’ve made so far. So if, like, people wanted to just get it done and set it up on the locally… on their local computer, like, it’s doable.

207 00:15:56.110 00:16:04.280 Gabriel Lam: So that’s sort of the… the, the… the plus, and then, for…

208 00:16:05.380 00:16:23.310 Gabriel Lam: for Utam, I’m just finishing off the demos for later today for Ian. We’re gonna call in about 45 minutes, so… I’ll show him a bunch of good stuff there. I would like to still get some documents from him, mainly to do with, like, templates and, like, the types of formatting he wants there.

209 00:16:24.190 00:16:32.030 Gabriel Lam: Like, basically, what’s the artifact that he wants to see exactly, and like, what’s it… what is the shape of it that it would be good for him to have?

210 00:16:32.220 00:16:34.560 Gabriel Lam: So that, that, I’m gonna need him to…

211 00:16:35.430 00:16:39.969 Gabriel Lam: send us, really. Aside from that, I think we…

212 00:16:40.090 00:16:43.809 Gabriel Lam: are… I think that’s really good for me. I looked into the…

213 00:16:44.040 00:16:46.579 Gabriel Lam: the video you sent yesterday, Utom, so…

214 00:16:47.390 00:16:49.699 Gabriel Lam: That seems really cool for Omni. I don’t know if.

215 00:16:49.700 00:16:52.080 Uttam Kumaran: What was the… what was the video? Sorry, what did I say?

216 00:16:52.080 00:16:58.389 Gabriel Lam: It was… it was the Omni, client, the Omni agent, oh, the Google.

217 00:16:58.390 00:16:58.840 Samuel Roberts: the Google…

218 00:16:58.840 00:16:59.740 Gabriel Lam: Yeah.

219 00:16:59.740 00:17:07.109 Uttam Kumaran: Oh, yeah, yeah, yeah. Well, the thing there is, like, we use Omni internally, so…

220 00:17:08.650 00:17:14.409 Uttam Kumaran: Yeah, I don’t know, I was just like, maybe you could play around there. There’s also, like, a couple of other op… like…

221 00:17:15.220 00:17:25.140 Uttam Kumaran: I feel like we should prob… we’ll probably end up just building our own, but if you like the Omni one, I can… I can ask them to connect us with whoever built it on their site to tell us how they did it.

222 00:17:27.409 00:17:35.779 Samuel Roberts: Yes, but it’s not too similar from what you’ve been trying to run anyway. They might have just written their own MCP that does specifically what they need, rather than, like, a…

223 00:17:36.160 00:17:38.760 Uttam Kumaran: Well, they’re doing it for, like, they have to scale it to, like, a bunch.

224 00:17:38.760 00:17:41.209 Samuel Roberts: Exactly, yeah, so they probably did it a little differently.

225 00:17:41.210 00:17:45.030 Uttam Kumaran: For us, we’ll, like, hardcode it once, and it’s, like, as long as it works.

226 00:17:45.030 00:17:53.029 Samuel Roberts: Is it worth taking a look at the railway? Like, do you want me to take a look at the… the loop that it got stuck in, or do you want to just keep doing it locally?

227 00:17:53.030 00:18:11.119 Gabriel Lam: That’d be great. I think maybe this is more a priority question, that I feel like if it’s so urgent for everyone to not… to try to have a web client set up, then sure. I mean, the current desktop setup is, like, kind of a stopgap solution for… The MCP.

228 00:18:11.120 00:18:11.520 Samuel Roberts: for.

229 00:18:11.520 00:18:13.070 Gabriel Lam: the MCP for Google.

230 00:18:13.330 00:18:15.220 Gabriel Lam: It’s a bit of a stopgap solution that works.

231 00:18:15.220 00:18:16.329 Samuel Roberts: Yeah, because we have to share.

232 00:18:16.330 00:18:16.780 Gabriel Lam: cold.

233 00:18:16.780 00:18:17.950 Samuel Roberts: And everything, and then…

234 00:18:17.950 00:18:18.860 Gabriel Lam: Yeah.

235 00:18:19.190 00:18:21.940 Samuel Roberts: Okay. I mean, I think either way, we do have to share a key.

236 00:18:22.100 00:18:31.520 Gabriel Lam: if… so there’s another problem, which is, like, I’m on the free trial version for Google Cloud, the APIs, and so, like, we’re limited.

237 00:18:31.720 00:18:32.690 Gabriel Lam: there.

238 00:18:32.690 00:18:33.929 Uttam Kumaran: Why, dude?

239 00:18:33.930 00:18:34.300 Samuel Roberts: Wait.

240 00:18:34.300 00:18:37.789 Uttam Kumaran: Dude, join our thing, I’ll pay for whatever you need.

241 00:18:37.790 00:18:38.560 Samuel Roberts: Yeah, I mean.

242 00:18:38.560 00:18:41.979 Gabriel Lam: As in, I found this out, like, last night, I was like, why.

243 00:18:41.980 00:18:42.580 Uttam Kumaran: Oh.

244 00:18:42.580 00:18:46.440 Gabriel Lam: I’m on trial. Like, I’ll just, like, message Rico today, and, like.

245 00:18:46.440 00:18:46.860 Uttam Kumaran: Yeah, yeah.

246 00:18:46.860 00:18:48.670 Gabriel Lam: Because, dude, they hate the API?

247 00:18:48.670 00:18:55.839 Uttam Kumaran: It’s, like, literally, like, negative… it’s, like, one, like, trillionth of a penny to call the Google Drive API.

248 00:18:55.840 00:18:56.450 Gabriel Lam: Yeah.

249 00:18:56.480 00:19:01.899 Uttam Kumaran: But he just may not have linked, you may not have linked it to our billing account, so… yeah.

250 00:19:01.900 00:19:08.909 Gabriel Lam: I think that’s really it, so I’ll get it to Rico, and then we’ll get that sorted out. But yeah, good progress, I think.

251 00:19:10.230 00:19:12.189 Gabriel Lam: I think we’ve got it pretty…

252 00:19:12.190 00:19:20.679 Samuel Roberts: Yeah, I think it is… I think you’re right, the local’s a good, like, stopgap, but if people could just, like, point it to that, and then OAuth with a Brainforge email, like, that seems…

253 00:19:21.130 00:19:26.149 Samuel Roberts: Yeah, which is… which is what it’s done. I literally just asked Cursor to, like, run that document.

254 00:19:26.870 00:19:29.880 Gabriel Lam: Yeah. It, like, tells you what you need to do, as well. Okay.

255 00:19:31.140 00:19:33.689 Gabriel Lam: That’s been great, so… Alright, for the…

256 00:19:34.200 00:19:39.610 Samuel Roberts: Actually, I’ll message you later, because if I can try… if I get a chance to get into the railway and the logs and see what’s going on there, maybe I can…

257 00:19:39.610 00:19:40.200 Gabriel Lam: Yep.

258 00:19:40.740 00:19:41.440 Samuel Roberts: Take a look.

259 00:19:42.210 00:19:47.939 Samuel Roberts: Sure. Just to see if it’s something, like, easy, or if it is, like, a weird railway networking thing that’s not worth it.

260 00:19:48.070 00:19:59.820 Samuel Roberts: The other side of it is it doesn’t have to be railway, it could be… if, like, Railway specifically has something weird, we could do a Heroki, because, like, Railway has been really nice for Lilo, because it has, like, the multiple services, but if this is just one MCP server.

261 00:20:00.070 00:20:00.620 Samuel Roberts: Doesn’t matter.

262 00:20:00.620 00:20:01.170 Gabriel Lam: I don’t know.

263 00:20:01.170 00:20:02.979 Samuel Roberts: So… But I’ll take a look.

264 00:20:03.470 00:20:04.070 Gabriel Lam: Okay.

265 00:20:05.330 00:20:12.890 Uttam Kumaran: Yeah, I think a couple things. Also, Gabe, you should connect with Sheshu, who just joined. He’s joining to kind of, like, lead ops.

266 00:20:14.470 00:20:20.829 Uttam Kumaran: like, you guys would have a ball talking about, like, sort of what we’re doing. He’s probably gonna grab time with you if he hasn’t already.

267 00:20:21.200 00:20:33.769 Uttam Kumaran: He’s probably someone that you just need to kind of, like, brain dump everything we’re doing on the AI side. And a couple particular things, that I just, like, want to say out loud. One is, like, I’m still…

268 00:20:34.210 00:20:37.009 Uttam Kumaran: Not happy that everyone is not using Whisper.

269 00:20:37.240 00:20:43.130 Uttam Kumaran: Like, Sam messaged me, it was like, I’m running out of my Super Whisper trial to use VoicesText. I’m like.

270 00:20:43.270 00:20:43.780 Uttam Kumaran: Yes.

271 00:20:43.780 00:20:46.119 Samuel Roberts: I didn’t know what you had. Yeah.

272 00:20:46.120 00:20:48.350 Uttam Kumaran: Oh, but less about you, Sam, more of, like…

273 00:20:48.350 00:20:48.839 Samuel Roberts: Totally right.

274 00:20:48.840 00:20:53.969 Uttam Kumaran: We have not… like, more people in our company need to be using voice-to-text.

275 00:20:53.970 00:20:54.380 Samuel Roberts: Yeah.

276 00:20:54.380 00:21:02.339 Uttam Kumaran: Because it is literally, like, life-changing, and I am, like, not really, like, willing to debate that anymore, so…

277 00:21:02.840 00:21:08.319 Uttam Kumaran: I want to, like… push that, and I’m willing to buy everybody Whisperer licenses.

278 00:21:09.380 00:21:24.139 Uttam Kumaran: Second thing is, a lot of folks on the sales and marketing side are using Granola. Granola released, like, ability to share recipes in an org. Their default recipes kind of suck for, like, generate follow-up email, things like that.

279 00:21:24.140 00:21:26.610 Samuel Roberts: It may take, like, 1 or 2 hours to just…

280 00:21:27.180 00:21:30.479 Uttam Kumaran: do some great Brain Forge recipes.

281 00:21:31.020 00:21:38.830 Uttam Kumaran: So that anyone who’s using Granola can… can do that. In particular, I use Granola for off-platform

282 00:21:39.220 00:21:47.509 Uttam Kumaran: meetings, because I’m in a lot of sales meetings where it’s someone else’s Google Meet or Zoom, or I’m in person, and so I turn it on, like, just, like, in my meeting when I’m sitting with somebody.

283 00:21:48.680 00:21:49.180 Samuel Roberts: Yeah.

284 00:21:49.180 00:21:55.309 Uttam Kumaran: So… those are, like, sort of managed services. In a very similar vein, Cursor.

285 00:21:55.450 00:21:57.720 Uttam Kumaran: I wanna think more about, like.

286 00:21:58.100 00:22:01.750 Uttam Kumaran: let’s say, like, Luke on sales was gonna install Cursor.

287 00:22:02.450 00:22:08.179 Uttam Kumaran: how does he go from, like, a blank VS Code to, like, a configured cursor workspace?

288 00:22:08.670 00:22:12.960 Uttam Kumaran: Like, through, like, our cursor settings file that we can just throw into the playbooks.

289 00:22:13.290 00:22:15.750 Uttam Kumaran: You know, cause…

290 00:22:15.750 00:22:16.410 Samuel Roberts: shoulder.

291 00:22:16.590 00:22:18.220 Samuel Roberts: Yeah, yeah, absolutely.

292 00:22:18.810 00:22:23.540 Uttam Kumaran: I also know they released Claude for work, I’m, like, less, like…

293 00:22:23.860 00:22:37.849 Uttam Kumaran: what we’re gonna see in the market, guys, is you’re gonna see a lot of tools that are, like, babying people around. Our folks are very smart, so I don’t, like, necessarily, like, impress with Cloud for Work, and that, yeah, I think a lot of people are gonna use it.

294 00:22:37.850 00:22:46.029 Uttam Kumaran: But it’s, like, it’s too basic. Like, our folks, like, should be in cursor, because the workflows are much more advanced.

295 00:22:48.000 00:22:59.669 Uttam Kumaran: So yeah, just, like, a couple things to keep in mind. One last piece is Gmail released some AI features. I feel like we have them turned on already, but again, only some of us are using that.

296 00:22:59.780 00:23:01.409 Samuel Roberts: On occasion.

297 00:23:02.100 00:23:06.820 Uttam Kumaran: And then I did ask Mustafa for a favor on helping me fix codecs.

298 00:23:07.020 00:23:14.179 Uttam Kumaran: I do need to flick that on, Mustafa, the internet thing, so I can run npm install, so I’ll do that right after this.

299 00:23:14.320 00:23:20.760 Uttam Kumaran: Just to see whether that remedies it. But…

300 00:23:20.970 00:23:23.699 Uttam Kumaran: Yeah, I feel like that’s another platform thing.

301 00:23:23.910 00:23:25.590 Uttam Kumaran: Sorry, I’m gonna ramble, maybe this.

302 00:23:25.590 00:23:25.950 Gabriel Lam: Oh, you’re.

303 00:23:25.950 00:23:26.480 Uttam Kumaran: transcript.

304 00:23:27.590 00:23:37.559 Uttam Kumaran: One more thing, I asked ops to help me make a decision if we want to buy Notion AI. It’s expensive, like, and Notion charges per person.

305 00:23:37.680 00:23:43.270 Uttam Kumaran: But if we’re doing a lot of work in Notion, and having AI there.

306 00:23:43.440 00:23:49.119 Uttam Kumaran: is helpful, like, if you could prove the ROI, I’m… I’m open to considering it.

307 00:23:49.510 00:23:53.199 Uttam Kumaran: I didn’t buy the Slack AI, because I thought it’s kind of…

308 00:23:53.640 00:23:56.059 Uttam Kumaran: I feel like it’s pretty mids right now, so…

309 00:23:56.060 00:23:56.480 Samuel Roberts: Yeah.

310 00:23:56.900 00:24:00.589 Uttam Kumaran: Again, I can be convinced, if, like, the ROI is there.

311 00:24:01.550 00:24:11.309 Uttam Kumaran: Two other things that I chatted with a prospective client about yesterday. One is, he started using Perplexity’s Comet browser for some, like, knowledge work.

312 00:24:12.080 00:24:20.389 Uttam Kumaran: I feel like maybe our sales and marketing folks may get a better kick out of that than our engineering folks, because they’re doing a lot of, like, online UI form filling.

313 00:24:20.660 00:24:21.400 Samuel Roberts: Got it.

314 00:24:21.700 00:24:26.129 Uttam Kumaran: And then the second piece is Claude released an extension, like a browser control extension.

315 00:24:27.060 00:24:41.699 Uttam Kumaran: Similarly, some of our folks on sales and marketing may find that very valuable to help speed up their workflows. Like, I’m not doing… like, I actually have… I’m doing some UI, like, clicking into softwares and stuff like that, but our delivery team is mostly not.

316 00:24:42.300 00:24:44.890 Uttam Kumaran: So there’s, like, some other thoughts there.

317 00:24:45.210 00:24:51.660 Uttam Kumaran: Yeah, that’s all, that’s all. For now.

318 00:24:51.750 00:24:56.229 Gabriel Lam: I guess a quick question would just be, like, what is the benefit of Notion AI?

319 00:24:56.710 00:24:58.680 Gabriel Lam: That you’re… that you’re, like…

320 00:24:58.680 00:25:04.710 Uttam Kumaran: Well, you basically can do, like, at Notion AI, write me a doc like this, reformat this, improve the grammar.

321 00:25:04.940 00:25:07.780 Uttam Kumaran: Like…

322 00:25:08.060 00:25:15.000 Uttam Kumaran: create me a database that does all these things. Like, so, for example, if you’re going to create a database with, like, 10 columns, certain types.

323 00:25:15.650 00:25:22.830 Uttam Kumaran: Doing that is really annoying. It takes, like, probably 10-15 minutes. It could just be, like, Notion AI, like… I can basically, like.

324 00:25:23.480 00:25:26.430 Uttam Kumaran: Click Notion AI, click My Whisper.

325 00:25:26.830 00:25:31.599 Uttam Kumaran: word vomit about a database I want, click go, and get it in probably, like.

326 00:25:31.970 00:25:35.250 Uttam Kumaran: 30 seconds versus whatever 10-15 minutes, right?

327 00:25:35.560 00:25:36.839 Uttam Kumaran: But…

328 00:25:37.240 00:25:42.260 Uttam Kumaran: If still people… if I buy and people aren’t gonna use it, then that’s tough, because I have to buy for everybody.

329 00:25:42.510 00:25:42.830 Samuel Roberts: FC.

330 00:25:42.830 00:25:43.759 Uttam Kumaran: Buy one off?

331 00:25:44.750 00:25:50.460 Uttam Kumaran: But if we’re doing a lot of work in Notion, Maybe it’s valuable, like…

332 00:25:50.840 00:25:56.930 Uttam Kumaran: You know? Because I find myself in Notion writing, and I’m like, but again, that’s… Just a thought.

333 00:25:58.350 00:26:00.270 Samuel Roberts: Yeah, interesting.

334 00:26:01.590 00:26:05.270 Samuel Roberts: Especially if we’re gonna be pulling things out of Notion at all into the cursor or anything, like…

335 00:26:06.290 00:26:09.979 Samuel Roberts: a lot of… a lot of, like, AI from different places being…

336 00:26:10.290 00:26:13.319 Samuel Roberts: Plugged into things and paid for, and… Yeah,

337 00:26:13.320 00:26:15.839 Gabriel Lam: I just, like, don’t want to double…

338 00:26:16.300 00:26:16.890 Uttam Kumaran: Yeah. Yeah.

339 00:26:16.890 00:26:20.380 Gabriel Lam: back, where I think Notion

340 00:26:20.650 00:26:38.619 Gabriel Lam: But at least what Cursor has been good at has been, like, reading documentation and not so much pushing documentation, or Notion, the MCP. And so that’s been the primary use case that’s been working really well. It’s just, like, hyper-specific, like, I want to know what’s going on in this document, as opposed to, like, write, like, build me a database, you know?

341 00:26:39.140 00:26:42.150 Gabriel Lam: Okay.

342 00:26:45.310 00:26:47.369 Samuel Roberts: Yeah, everyone’s trying to wall us into their gardens.

343 00:26:49.010 00:27:07.429 Samuel Roberts: like, Notion, Notion AI, like, Cloud for Work, I’m like, like, this is why I kind of prefer Cursor to Cloud Code right now, it’s just because, like, I can pick a different model if I want, I can try a few different things, I can run multiple models, and, like, I don’t want to get tied in, especially because, like, these are… it’s… it’s gonna become much more commoditized. It’s already becoming commoditized, and, like…

344 00:27:08.140 00:27:13.649 Samuel Roberts: Notion AI seems like, no, I don’t… I don’t necessarily need you to do it, I just need something to be able to do it, you know?

345 00:27:15.350 00:27:17.420 Samuel Roberts: But if that’s the, you know, I don’t know.

346 00:27:20.630 00:27:25.080 Samuel Roberts: Cool. Okay, Bobby just responded, something sweet. Alright,

347 00:27:25.710 00:27:28.120 Samuel Roberts: Any other thoughts to get figured out now?

348 00:27:32.300 00:27:35.149 Samuel Roberts: Who’s the new, ops person? I missed, I missed that.

349 00:27:35.150 00:27:41.060 Uttam Kumaran: Keshu. Keshu, okay. Yeah, I’ll start, he’s… he’ll, he’s in some of these meetings.

350 00:27:41.320 00:27:43.320 Uttam Kumaran: So, yeah.

351 00:27:43.320 00:27:47.269 Gabriel Lam: Yeah. I had a chance to talk with him, I think, Rico sent.

352 00:27:47.270 00:27:48.050 Uttam Kumaran: Oh, yeah, yeah.

353 00:27:48.050 00:27:48.750 Gabriel Lam: That…

354 00:27:48.750 00:27:49.090 Uttam Kumaran: Okay.

355 00:27:49.090 00:27:54.680 Gabriel Lam: link over, so I’ve chatted with him, but I think… great to continue that conversation, so…

356 00:27:54.680 00:27:55.000 Samuel Roberts: Cool.

357 00:27:55.000 00:27:55.540 Uttam Kumaran: Cool.

358 00:27:55.910 00:27:56.570 Gabriel Lam: Go.

359 00:27:58.870 00:27:59.640 Uttam Kumaran: Okay.

360 00:27:59.900 00:28:06.180 Uttam Kumaran: Alright guys, thank you. Yeah, Gabe, let me know how the call with Ian goes. I would say this really buys towards, like.

361 00:28:06.340 00:28:09.129 Uttam Kumaran: What is the thing that, like, is the hook?

362 00:28:09.840 00:28:13.220 Uttam Kumaran: And like, what are the core features?

363 00:28:13.400 00:28:19.009 Uttam Kumaran: Because once you have the core logic, we can build a simple UI around your demo, by the way.

364 00:28:19.260 00:28:21.370 Uttam Kumaran: But we just need, like, kind of, like.

365 00:28:22.120 00:28:25.329 Uttam Kumaran: It should just work, you know, and be, like, powerful, so…

366 00:28:25.800 00:28:29.049 Gabriel Lam: Right. Yeah, I think the main thing is just, like, having it…

367 00:28:29.350 00:28:35.469 Gabriel Lam: produce documentation that, like, Ian can take pretty much instantly, which I think requires, like, his…

368 00:28:35.640 00:28:46.829 Gabriel Lam: some sort of documentation from him, or skeleton from him, to be like, this is how I want the information displayed. Because that’s something that I can generate, but I actually cannot validate at the moment.

369 00:28:47.320 00:28:48.680 Gabriel Lam: Correct.

370 00:28:49.700 00:28:50.890 Samuel Roberts: Which, what is this for?

371 00:28:51.160 00:28:51.920 Samuel Roberts: Context.

372 00:28:53.300 00:28:58.919 Gabriel Lam: Yeah, so basically, we have a partnership with Contextual AI, which is, like, a RAG model, and… Right, right.

373 00:28:59.690 00:29:01.860 Gabriel Lam: And so, one of the things that we’re trying to do is

374 00:29:02.230 00:29:18.389 Gabriel Lam: specifically for, like, highly regulated industries where it’s really important to extract, like, word for word, and, like, understand documentation and not hallucinate in any way, like, no inference, really. Right, right. So insurance is one of the…

375 00:29:18.390 00:29:18.820 Samuel Roberts: Oh, God.

376 00:29:18.820 00:29:21.519 Gabriel Lam: Okay. And, there’s a…

377 00:29:21.910 00:29:30.959 Gabriel Lam: who Tom has a connection in the industry, and so we’re working with him and the contractual team to pump out… create a demo that we can hopefully bring to marketing and sales.

378 00:29:31.320 00:29:31.900 Samuel Roberts: Sweet.

379 00:29:32.350 00:29:33.080 Gabriel Lam: Yeah.

380 00:29:33.460 00:29:34.030 Samuel Roberts: Cool.

381 00:29:34.830 00:29:38.189 Samuel Roberts: Interestingly about that on Slack, I just totally forgot the context there. Cool.

382 00:29:38.190 00:29:38.780 Gabriel Lam: You’re good.

383 00:29:38.960 00:29:46.750 Samuel Roberts: Alrighty, I think that’s… that’s the meeting, then. Thank you all. I’ll be…

384 00:29:46.970 00:29:53.569 Samuel Roberts: I don’t have a ton of, like, deep work, hopefully, so if you need me to help with stuff, like, ping me. I’ll be in and out of stuff, but I’m not gonna be, like.

385 00:29:53.710 00:30:02.189 Samuel Roberts: I’ll be configuring stuff on railway for a bit, so, Casey specifically, and Mustafa, I know we want to talk, evals, so,

386 00:30:02.890 00:30:03.640 Samuel Roberts: Cool.

387 00:30:04.260 00:30:06.490 Uttam Kumaran: Perfect. Okay. Alright, thank you all.

388 00:30:06.670 00:30:07.250 Uttam Kumaran: Thank you.

389 00:30:08.150 00:30:09.310 Samuel Roberts: Alright, bye.