Meeting Title: AI Planning & Grooming (Interlude Tickets) Date: 2025-09-19 Meeting participants: Samuel Roberts, Rico Rejoso, Mustafa Raja, Uttam Kumaran
WEBVTT
1 00:00:21.520 ⇒ 00:00:22.610 Rico Rejoso: Hey, Sam.
2 00:00:26.130 ⇒ 00:00:27.880 Samuel Roberts: I’m still muted. Okay, there we go. Hey!
3 00:00:29.580 ⇒ 00:00:31.790 Rico Rejoso: Hey, how are you?
4 00:00:32.350 ⇒ 00:00:33.220 Samuel Roberts: Alright.
5 00:00:33.420 ⇒ 00:00:35.400 Samuel Roberts: It’s been pretty good.
6 00:00:36.730 ⇒ 00:00:41.270 Samuel Roberts: Yeah, I mean, I think the interlude call went well, getting a better sense of what they might…
7 00:00:42.360 ⇒ 00:00:46.789 Samuel Roberts: a lot, explore some other tools.
8 00:00:48.100 ⇒ 00:00:50.279 Samuel Roberts: Or maybe generating some visuals, but…
9 00:00:51.010 ⇒ 00:00:59.160 Rico Rejoso: Yeah, oh, great. Actually, I used the platform to create the tickets based on the meeting. I’m trying to utilize it as much as I can.
10 00:00:59.160 ⇒ 00:01:06.920 Samuel Roberts: Oh, great, yeah, let me… any feedback is good feedback. So if there’s something weird, if it’s a layout thing, if it’s a…
11 00:01:07.170 ⇒ 00:01:14.229 Samuel Roberts: The tickets aren’t… because I’m actually curious now, because this is all, you know, all from the old platform, but if we have a…
12 00:01:14.440 ⇒ 00:01:19.310 Samuel Roberts: new, like, linear… ticket,
13 00:01:19.950 ⇒ 00:01:25.850 Samuel Roberts: what did he call it? What did Justin call it on the call? Like, a guide, or whatever?
14 00:01:26.960 ⇒ 00:01:31.649 Samuel Roberts: We definitely want to add that. The SOP, but he had, like, a linear… I forget what it was.
15 00:01:31.650 ⇒ 00:01:32.550 Rico Rejoso: template.
16 00:01:32.550 ⇒ 00:01:37.939 Samuel Roberts: Standard template, that’s what it was. So I’m thinking we want to add that to the linear ticket generation step.
17 00:01:38.870 ⇒ 00:01:40.870 Rico Rejoso: Yeah, so you can format it that way, right?
18 00:01:41.260 ⇒ 00:01:48.599 Samuel Roberts: Yeah, that’s what I want to try to figure out, if we can do that. I’m sure we can, I just… I need to see… yeah, this is… this is all just,
19 00:01:50.270 ⇒ 00:01:55.429 Samuel Roberts: what am I trying to say? This is all… Just generated from the…
20 00:01:55.870 ⇒ 00:01:57.380 Rico Rejoso: Meeting…
21 00:01:57.710 ⇒ 00:02:02.509 Samuel Roberts: And then probably some little stuff about how the ticket should look.
22 00:02:03.160 ⇒ 00:02:07.040 Samuel Roberts: But there’s definitely a lot of bugs, like, if you look, like, scroll up again…
23 00:02:08.930 ⇒ 00:02:11.809 Samuel Roberts: So, for some reason, Sam Roberts loads my picture.
24 00:02:13.280 ⇒ 00:02:14.729 Samuel Roberts: But if you scroll down.
25 00:02:15.280 ⇒ 00:02:16.259 Rico Rejoso: The other isn’t.
26 00:02:16.340 ⇒ 00:02:19.150 Samuel Roberts: his doesn’t, but if Samuel Roberts is there, too…
27 00:02:19.900 ⇒ 00:02:20.640 Rico Rejoso: It is.
28 00:02:20.640 ⇒ 00:02:21.939 Samuel Roberts: I’ll go all the way to the bottom.
29 00:02:22.280 ⇒ 00:02:33.909 Samuel Roberts: No, no, not that one. Just scroll down. There’s a Samuel Roberts one. So I think there’s probably a lot of cleanup like that to figure out. Doesn’t really matter either way at this point, but,
30 00:02:34.170 ⇒ 00:02:36.090 Samuel Roberts: Would love to… Yeah.
31 00:02:36.280 ⇒ 00:02:45.040 Samuel Roberts: So, please, any feedback is good there. If it’s finding too many people, the wrong people, if it’s finding,
32 00:02:46.430 ⇒ 00:02:52.429 Samuel Roberts: if it’s not matching the format, obviously, like, at this point, it doesn’t know that yet.
33 00:02:53.070 ⇒ 00:03:02.430 Samuel Roberts: But, yeah, I’m glad… I’m glad you’re trying it out, because I… I think there’s a lot to be gained from really digging in and trying to use the platform a bit and see what we can add, because…
34 00:03:02.970 ⇒ 00:03:04.830 Samuel Roberts: It could be really powerful, also.
35 00:03:04.830 ⇒ 00:03:12.769 Rico Rejoso: keeping the transcript, but I find it ineffective, since why use it… why use ChatGPT when you already have it built in on this, right?
36 00:03:12.770 ⇒ 00:03:18.829 Samuel Roberts: That’s the plan, and this is connected to all the Client Hub stuff, so it knows not just this meeting, but it should know
37 00:03:19.200 ⇒ 00:03:21.640 Samuel Roberts: Like, historical stuff, too, theoretically.
38 00:03:21.640 ⇒ 00:03:22.370 Rico Rejoso: Hmm.
39 00:03:22.780 ⇒ 00:03:24.290 Rico Rejoso: I think one thing would be this.
40 00:03:24.580 ⇒ 00:03:27.750 Rico Rejoso: I think interlude… I mean, I saw the interlude here.
41 00:03:28.260 ⇒ 00:03:32.480 Rico Rejoso: Right? But it’s directing to AI team.
42 00:03:33.000 ⇒ 00:03:37.469 Samuel Roberts: Yeah, that I don’t know. Scroll down again for a second? What’s that first one? Interlude Studio?
43 00:03:37.470 ⇒ 00:03:37.910 Rico Rejoso: one.
44 00:03:37.910 ⇒ 00:03:42.690 Samuel Roberts: You did that one? Okay. Yeah, that’s one thing I’m thinking about a little bit, is how do we know…
45 00:03:42.840 ⇒ 00:03:49.529 Samuel Roberts: what… like, we can probably infer…
46 00:03:49.950 ⇒ 00:03:52.010 Rico Rejoso: From, like, a meeting title.
47 00:03:52.010 ⇒ 00:04:00.650 Samuel Roberts: Yeah, but, like, yeah, so, like, it already knows, like, it’s in… it puts it in the right client hub, so we should be able to pass that in here, too, yeah.
48 00:04:01.520 ⇒ 00:04:07.060 Rico Rejoso: I think… I think it should be the proper title for each meeting, if we can set up something like that.
49 00:04:07.250 ⇒ 00:04:16.220 Samuel Roberts: Yeah, I think as long as it mentions a client, I don’t know exactly how it’s doing it right now, I’d have to dig in, but that’s a whole other big project I want to do, because the client hubs are,
50 00:04:16.860 ⇒ 00:04:24.839 Samuel Roberts: complicated to set up. I think I… we talked about this the other day, so… I wanna… I want to streamline that a little bit, so that’s definitely, something to help.
51 00:04:25.730 ⇒ 00:04:27.370 Rico Rejoso: Improve this.
52 00:04:27.900 ⇒ 00:04:34.290 Rico Rejoso: from, I mean, from the general idea of what must be done, or what we can include to the,
53 00:04:34.500 ⇒ 00:04:47.699 Rico Rejoso: work for the client as well. Maybe from this one, we can branch it out to different action steps that we can take to get to this process. But the idea of having this in linear is a great feat already.
54 00:04:48.070 ⇒ 00:04:54.939 Samuel Roberts: Yeah, yeah, just getting a few things in there, because, you know, sometimes we have these meetings, and then you have to go and parse through it all, and yeah.
55 00:04:55.110 ⇒ 00:05:01.030 Rico Rejoso: Yeah, and it’s hard for me to, you know, note everything. Of course, yeah.
56 00:05:01.650 ⇒ 00:05:05.170 Samuel Roberts: Definitely, definitely, the transcripts are good. They’re good, hopefully.
57 00:05:05.450 ⇒ 00:05:08.200 Rico Rejoso: Not sure if Mustafa would be joining.
58 00:05:08.200 ⇒ 00:05:11.450 Samuel Roberts: Yeah, I don’t know, I was just checking to see…
59 00:05:12.970 ⇒ 00:05:15.999 Samuel Roberts: Mustafa and Utam are on this meeting, right? But I’m like…
60 00:05:16.310 ⇒ 00:05:18.500 Rico Rejoso: Not sure about who Thompson.
61 00:05:18.500 ⇒ 00:05:23.700 Samuel Roberts: Yeah, yeah, I just saw him. He’s the only other one listed there, so… .
62 00:05:23.700 ⇒ 00:05:24.879 Rico Rejoso: At least Mustafa.
63 00:05:25.210 ⇒ 00:05:27.010 Samuel Roberts: Yeah, let me ping him real quick.
64 00:05:31.370 ⇒ 00:05:33.999 Samuel Roberts: He was having internet problems earlier, so I don’t know.
65 00:05:34.660 ⇒ 00:05:36.950 Rico Rejoso: But adventure, I mean, I think it’s good.
66 00:05:37.290 ⇒ 00:05:41.840 Samuel Roberts: Yeah, on the… It was just burning on me.
67 00:05:42.520 ⇒ 00:05:46.289 Rico Rejoso: Maybe I can discuss how we can set up the stand-up, because right now the
68 00:05:46.540 ⇒ 00:05:55.699 Rico Rejoso: process or program that we have is you go through with each, team members and, you know, discuss with your clients
69 00:05:55.880 ⇒ 00:05:58.840 Rico Rejoso: what they’re doing, what, I mean, the blockage.
70 00:06:00.380 ⇒ 00:06:13.730 Rico Rejoso: then afterwards, maybe that’s when we can go through the AI stuff, or the AI team, because for client things, I think I… I’m not involved in that, since I don’t have… I mean, I’m not technical with such things, and I cannot provide.
71 00:06:13.730 ⇒ 00:06:14.330 Samuel Roberts: Sure.
72 00:06:14.330 ⇒ 00:06:22.050 Rico Rejoso: insight on that, but on the AI team, maybe I can help you, you know, manage some of the tickets that we have, since we have, like, overdue tickets here.
73 00:06:22.450 ⇒ 00:06:26.009 Samuel Roberts: Definitely, yeah, and the… all the tickets and AI team, I feel like…
74 00:06:26.110 ⇒ 00:06:36.079 Samuel Roberts: I think we talked about this a little bit. We need to really go through and see, like, what’s still important, what’s not, at least go way back, you know?
75 00:06:36.080 ⇒ 00:06:41.669 Rico Rejoso: Do you have, like, a good sense if we will be needing such tickets in the future or not?
76 00:06:42.150 ⇒ 00:06:49.640 Samuel Roberts: I… I think if I… now that I’ve, you know, worked on the platform a little bit, and have a better sense, I… I can probably go through and do a first pass.
77 00:06:49.800 ⇒ 00:06:53.340 Samuel Roberts: I’ll be pinging Casey and Mustafa as well.
78 00:06:53.470 ⇒ 00:06:58.979 Samuel Roberts: Maybe Utam. But yeah, I can, I can add that to my to-dos.
79 00:06:59.240 ⇒ 00:07:02.990 Samuel Roberts: So just do, like, a quick scan of it, and try to…
80 00:07:03.130 ⇒ 00:07:08.759 Samuel Roberts: Either ask some questions about why it’s there, or what it is, or,
81 00:07:09.560 ⇒ 00:07:11.869 Samuel Roberts: Or just, you know, put them…
82 00:07:12.050 ⇒ 00:07:18.770 Samuel Roberts: mark them as, you know, irrelevant or future or something, I don’t know. Yeah.
83 00:07:18.980 ⇒ 00:07:20.409 Samuel Roberts: I can definitely take a look at that.
84 00:07:20.610 ⇒ 00:07:21.939 Rico Rejoso: While we’re waiting, since.
85 00:07:21.940 ⇒ 00:07:22.310 Samuel Roberts: Sure.
86 00:07:22.310 ⇒ 00:07:29.029 Rico Rejoso: So, the internal stuff. Let’s go through this, let’s like it’s under internal review. Is there anything that we can close out here? Something that, you know.
87 00:07:29.170 ⇒ 00:07:31.460 Rico Rejoso: Does it require any changes or revision?
88 00:07:36.650 ⇒ 00:07:41.020 Samuel Roberts: I would say… I think the client hubs are good.
89 00:07:41.410 ⇒ 00:07:47.910 Samuel Roberts: I think they’re… Certainly… .
90 00:07:51.050 ⇒ 00:07:54.739 Rico Rejoso: Oh, yeah, I think they’re… they’re fully implemented, so I would say they’re good.
91 00:07:55.090 ⇒ 00:07:59.360 Samuel Roberts: platform architecture document is kind of a living document, so I would say that’s…
92 00:08:00.220 ⇒ 00:08:04.790 Samuel Roberts: I don’t really want to call it done, but… I, I wanna…
93 00:08:05.010 ⇒ 00:08:11.470 Samuel Roberts: keep it kind of somewhere, but I wouldn’t… the, 450.
94 00:08:11.920 ⇒ 00:08:13.200 Rico Rejoso: 415, oh yes.
95 00:08:14.450 ⇒ 00:08:20.849 Mustafa Raja: Sorry, I… Earlier, I saw that it was at, 40?
96 00:08:21.300 ⇒ 00:08:23.830 Mustafa Raja: It was moved behind, I believe.
97 00:08:24.560 ⇒ 00:08:25.650 Samuel Roberts: How do we move this one?
98 00:08:26.170 ⇒ 00:08:27.389 Mustafa Raja: Yeah, the meeting.
99 00:08:27.390 ⇒ 00:08:28.610 Samuel Roberts: Okay.
100 00:08:29.170 ⇒ 00:08:32.630 Rico Rejoso: Oh, I’m sorry, yeah Yeah, I moved it earlier.
101 00:08:33.110 ⇒ 00:08:33.919 Samuel Roberts: Oh, okay.
102 00:08:34.210 ⇒ 00:08:38.889 Rico Rejoso: For this one, the architecture documents sent to team on Wednesday and Sam to host meeting next week.
103 00:08:38.890 ⇒ 00:08:43.570 Samuel Roberts: Yeah, I meant to… we just kinda got deep into the platform stuff, so now…
104 00:08:43.700 ⇒ 00:08:49.770 Samuel Roberts: Yeah, I would actually move that. I need to do a little more work on it now that I have a better…
105 00:08:50.350 ⇒ 00:08:51.450 Rico Rejoso: Okay.
106 00:08:51.450 ⇒ 00:08:54.519 Mustafa Raja: Do we want to… do we want to invite Casey also?
107 00:08:55.220 ⇒ 00:08:58.680 Samuel Roberts: We’re just… we were just chatting through, but this is actually about interladen.
108 00:08:59.290 ⇒ 00:09:00.340 Mustafa Raja: Oh, okay, okay.
109 00:09:00.340 ⇒ 00:09:03.040 Samuel Roberts: So yeah, no, we were just chatting while we were waiting.
110 00:09:03.040 ⇒ 00:09:14.789 Rico Rejoso: I mean, yeah, we can invite Casey of stuff, since for Interlude, you weren’t able to meet with Micah, right? Or Mika, and he wasn’t able to, create or identify any tickets that we should add for next week?
111 00:09:16.040 ⇒ 00:09:21.179 Mustafa Raja: I guess for next week, we do want to look at, what’s it called, the platforms.
112 00:09:21.330 ⇒ 00:09:24.230 Mustafa Raja: For, visuals, right?
113 00:09:24.830 ⇒ 00:09:32.899 Samuel Roberts: I think it’s worth maybe taking a little bit of time to explore them. I don’t know what exactly we’ll end up doing, because I don’t know
114 00:09:33.400 ⇒ 00:09:36.049 Samuel Roberts: How much they’re gonna want on that until he kinda…
115 00:09:36.230 ⇒ 00:09:42.410 Mustafa Raja: Yeah. But I think having an idea of something that, like, at least knowing how we might do something is good, so kind of more exploratory.
116 00:09:42.630 ⇒ 00:09:43.280 Samuel Roberts: Yeah.
117 00:09:43.280 ⇒ 00:09:43.830 Rico Rejoso: Right.
118 00:09:44.900 ⇒ 00:09:55.009 Rico Rejoso: Actually, the meeting was great. You were able to, like, present meetings that we can work on aside from that, I mean, the deck, right? And from… I think it’s the full process that we’re looking to build for them right now, right?
119 00:09:57.890 ⇒ 00:10:05.639 Samuel Roberts: Yeah, I mean, I think there were still some questions, but I think the big things were we’re gonna be getting the,
120 00:10:06.430 ⇒ 00:10:08.610 Samuel Roberts: those final versions.
121 00:10:08.860 ⇒ 00:10:10.970 Samuel Roberts: That will be,
122 00:10:12.110 ⇒ 00:10:17.399 Samuel Roberts: that’ll… is that in here? I don’t know if it’s in this linear tickets already, but adding some.
123 00:10:17.820 ⇒ 00:10:18.719 Rico Rejoso: Here you come.
124 00:10:19.220 ⇒ 00:10:20.000 Uttam Kumaran: Oh, nice.
125 00:10:21.420 ⇒ 00:10:25.130 Uttam Kumaran: I’ve been looking forward to this meeting. I’m glad to be here.
126 00:10:25.450 ⇒ 00:10:26.319 Samuel Roberts: Good, good.
127 00:10:28.040 ⇒ 00:10:31.369 Uttam Kumaran: I was just Just demoing the platform for some people.
128 00:10:31.610 ⇒ 00:10:32.580 Samuel Roberts: Oh, nice.
129 00:10:34.360 ⇒ 00:10:36.669 Samuel Roberts: I lost my train of thought now.
130 00:10:37.160 ⇒ 00:10:41.250 Samuel Roberts: The final decks… yeah, we need to, like,
131 00:10:41.780 ⇒ 00:10:44.879 Samuel Roberts: Probably figure out a good way to ingest those, and…
132 00:10:45.370 ⇒ 00:10:47.370 Samuel Roberts: Make use of them as, like, a…
133 00:10:48.940 ⇒ 00:10:57.499 Uttam Kumaran: Yeah, I guess my point is that, like, what parts of that are you helpful? Because there’s a couple ways I was thinking when I… I assume you’re talking about Interlude.
134 00:10:57.880 ⇒ 00:10:58.339 Samuel Roberts: Yeah, yeah, yeah.
135 00:10:58.670 ⇒ 00:11:04.059 Uttam Kumaran: the Dropbox, so… One is, like, I think you should…
136 00:11:05.170 ⇒ 00:11:12.570 Uttam Kumaran: I guess it sort of depends, I don’t know what the system is currently, on, like, how much judgment you guys are doing with the agents. For example.
137 00:11:12.920 ⇒ 00:11:21.969 Uttam Kumaran: In what parts does the knowledge of what the final outputs help, right? So, one is you should totally at least extract the text and use it in our evals.
138 00:11:21.970 ⇒ 00:11:23.290 Samuel Roberts: That’s exactly, yeah, that’s one…
139 00:11:23.290 ⇒ 00:11:24.970 Uttam Kumaran: So that’s, like, I think a minimum.
140 00:11:25.280 ⇒ 00:11:29.790 Uttam Kumaran: I also think there’s some… there’s something, like…
141 00:11:30.150 ⇒ 00:11:32.739 Uttam Kumaran: About the structure of the deck?
142 00:11:33.170 ⇒ 00:11:46.509 Uttam Kumaran: that Matthew keeps hitting on, which is, like, the narrative, right? And so, how do you judge the narrative agent, I think is also something unique here, which is, like, it’s not just the text, like, if you…
143 00:11:46.700 ⇒ 00:11:52.490 Uttam Kumaran: put all the text in. I think you lose a lot of fidelity of the ordering of slides.
144 00:11:52.700 ⇒ 00:11:56.630 Uttam Kumaran: Yeah. And there’s probably also stuff about the design color that…
145 00:11:56.630 ⇒ 00:11:57.649 Samuel Roberts: That’s what I was thinking.
146 00:11:57.650 ⇒ 00:12:09.659 Uttam Kumaran: Right? And I know you… I think that’s what you were thinking about, too, in that meeting. Yeah. I just don’t know how that goes to reinforce. One way is, like, shove that all into our prompt op… like, we have a prompt optimizer prompt.
147 00:12:09.660 ⇒ 00:12:10.190 Samuel Roberts: Right.
148 00:12:10.190 ⇒ 00:12:17.560 Uttam Kumaran: You could take all of those decks, shove it in, shove our existing prompt in, and you could at least probably do one notch of optimization.
149 00:12:17.730 ⇒ 00:12:18.460 Samuel Roberts: Yeah.
150 00:12:18.460 ⇒ 00:12:27.930 Uttam Kumaran: We should totally upgrade all the prompts to a few-shot or multi-shot, right? Like, if we can. And then I do think that evals, improving the evals.
151 00:12:28.590 ⇒ 00:12:34.450 Uttam Kumaran: Is, like, another… is another great way, you know?
152 00:12:34.770 ⇒ 00:12:37.910 Samuel Roberts: Yeah, when I was… thinking…
153 00:12:38.030 ⇒ 00:12:42.280 Samuel Roberts: was a few of those things. Yeah, the text pulling the copy out, and being able to…
154 00:12:42.550 ⇒ 00:12:44.499 Samuel Roberts: Yeah, do some…
155 00:12:44.620 ⇒ 00:12:51.040 Samuel Roberts: well, actually, we could do evals on probably the copy and eventually the narrative and all that stuff. I was thinking…
156 00:12:51.820 ⇒ 00:12:54.290 Samuel Roberts: If we have all these, if we could…
157 00:12:54.630 ⇒ 00:12:57.789 Samuel Roberts: Get the initial inputs as well.
158 00:12:58.700 ⇒ 00:13:00.679 Samuel Roberts: That would give us a lot to, like…
159 00:13:01.840 ⇒ 00:13:06.969 Samuel Roberts: refine the evals in general, because then it would have, like, a… we could actually run
160 00:13:07.230 ⇒ 00:13:09.540 Samuel Roberts: On, like, given these inputs.
161 00:13:10.710 ⇒ 00:13:18.860 Samuel Roberts: this is the copy we wanted, kind of thing. Because right now, we don’t have that, really. We had to just do kind of, like, an LLM as a judge, and not really, like, a…
162 00:13:19.140 ⇒ 00:13:22.900 Samuel Roberts: Well, no, it wasn’t even that, it was just, like, a rubric that we did, right, Mustafa?
163 00:13:23.100 ⇒ 00:13:23.850 Mustafa Raja: Yeah.
164 00:13:24.420 ⇒ 00:13:30.540 Samuel Roberts: Because we didn’t have, like, a good comparison for, like, what the ideal output was at that point.
165 00:13:32.010 ⇒ 00:13:51.060 Mustafa Raja: The thing was that we, the current generation, we cannot really judge that, right? How BrainTrust works is it’s going to run the previous inputs, on our workflow and see, how close the outputs are going to be.
166 00:13:51.660 ⇒ 00:13:53.809 Mustafa Raja: Like, how they were before.
167 00:13:54.500 ⇒ 00:14:01.000 Samuel Roberts: Yeah, so I’m thinking if we could, now that we have, or will have, final… outputs.
168 00:14:01.880 ⇒ 00:14:02.220 Mustafa Raja: Yeah.
169 00:14:02.220 ⇒ 00:14:03.889 Samuel Roberts: Collecting at least a copy from that.
170 00:14:04.110 ⇒ 00:14:13.190 Samuel Roberts: and then hopefully they can give us the things that they would be inputting, or would have input to this data agent. And then that will let us kind of have, like.
171 00:14:13.460 ⇒ 00:14:24.079 Samuel Roberts: the conditions on either end, and we can kind of work within that to, you know, improve the evals, but also, like, just do some general refinement in general, because I think
172 00:14:25.020 ⇒ 00:14:37.399 Samuel Roberts: that’s been a missing piece. We had kind of his examples, but like you said, those were V1, V2s, so all the stuff that we were talking about with, Mika, Micah, I’m actually… I never forget.
173 00:14:37.400 ⇒ 00:14:39.880 Uttam Kumaran: I said Micah, I didn’t know, I’ve just never met…
174 00:14:39.880 ⇒ 00:14:46.299 Samuel Roberts: Yeah, I honestly, I forget now. We met him once last week, and I’ve just been reading his name a bunch, so I’m like, I don’t… I feel bad.
175 00:14:47.140 ⇒ 00:14:52.899 Samuel Roberts: But, like, I think some of that back-and-forth stuff, which is another way that we were gonna be learning how to improve the agent over time.
176 00:14:53.040 ⇒ 00:15:00.720 Samuel Roberts: we might be able to maybe shortcut that a little bit if we say, like, this is the kind of stuff we’re looking for. Because right now, it’s… it’s just based on the…
177 00:15:01.240 ⇒ 00:15:04.159 Samuel Roberts: types of chats that Matthew was having.
178 00:15:05.490 ⇒ 00:15:13.089 Samuel Roberts: So I’m excited to get that data, figure out how to pull that. I think there is some stuff we could do visually, potentially, or, like, structurally for the deck.
179 00:15:13.810 ⇒ 00:15:19.160 Uttam Kumaran: This is also the thing, is, like, one thing that people are talking a lot about is, like, self-healing agents.
180 00:15:19.380 ⇒ 00:15:20.110 Samuel Roberts: Yeah.
181 00:15:20.110 ⇒ 00:15:27.170 Uttam Kumaran: Self-healing indicates that there’s a feedback loop between your evals.
182 00:15:27.440 ⇒ 00:15:33.390 Uttam Kumaran: And you’re, like, a prompt optimizer, and then the prompts, and, like, it kind of loops.
183 00:15:33.390 ⇒ 00:15:33.940 Samuel Roberts: Yeah, yeah, yeah.
184 00:15:33.940 ⇒ 00:15:40.559 Uttam Kumaran: I don’t think, like, that is maybe, like, what the North Star is here, but I just think we need to, like.
185 00:15:40.870 ⇒ 00:15:48.289 Uttam Kumaran: be self-healing in just our engineering efforts here, right? I think evals is the first way for you to isolate
186 00:15:48.800 ⇒ 00:15:53.800 Uttam Kumaran: like… With enough, like, with some level of confidence, the worst
187 00:15:54.120 ⇒ 00:15:56.390 Uttam Kumaran: performing outputs, and then treat it.
188 00:15:56.580 ⇒ 00:15:57.090 Uttam Kumaran: Right.
189 00:15:57.090 ⇒ 00:15:57.740 Samuel Roberts: really.
190 00:15:58.000 ⇒ 00:16:09.200 Uttam Kumaran: So, that’s what I think, like, and I think that’s kind of what we should do, is that, one, I think you should use… I think find a way to extract as much structured information from those finished decks.
191 00:16:09.430 ⇒ 00:16:15.880 Uttam Kumaran: Second is, I think, Like, do one more pass at the core prompts for each agent.
192 00:16:16.960 ⇒ 00:16:24.269 Uttam Kumaran: And then, I think third, we really need to,
193 00:16:27.370 ⇒ 00:16:31.350 Uttam Kumaran: You almost need to kind of, like, backtest, even, maybe.
194 00:16:31.460 ⇒ 00:16:37.599 Uttam Kumaran: Like, if… for those finished… like, I guess my… my point is that…
195 00:16:38.580 ⇒ 00:16:49.519 Uttam Kumaran: for those finished decks, like, we don’t… we don’t… like, of course, you don’t… this is what’s happening in, like, reinforcement learning, is, like, they tend to overfit for, like.
196 00:16:49.840 ⇒ 00:16:59.170 Uttam Kumaran: these, scoring benchmarks, right? But, like, taking that similar mindset, you almost want to say, cool, if we just… if we have this… a lot of these finished decks.
197 00:16:59.320 ⇒ 00:17:02.560 Uttam Kumaran: And let’s say we just had our starting information.
198 00:17:02.720 ⇒ 00:17:05.719 Samuel Roberts: could we have gotten there faster, right? That’s what I’m thinking.
199 00:17:05.720 ⇒ 00:17:06.290 Uttam Kumaran: That’s name.
200 00:17:06.380 ⇒ 00:17:07.220 Samuel Roberts: Exactly, yeah.
201 00:17:07.630 ⇒ 00:17:11.039 Samuel Roberts: That’s… that’s what I’m… I’m wondering, like, if we can run…
202 00:17:11.400 ⇒ 00:17:13.429 Samuel Roberts: You know, our current evals on that.
203 00:17:13.599 ⇒ 00:17:15.239 Samuel Roberts: See if we’re, you know, what…
204 00:17:16.750 ⇒ 00:17:20.639 Samuel Roberts: what the output is, how does it match what is there, but also…
205 00:17:20.839 ⇒ 00:17:22.859 Samuel Roberts: Do a little bit of refinement.
206 00:17:23.250 ⇒ 00:17:28.140 Samuel Roberts: to… Shoot for that as the, like, output, that kind of…
207 00:17:29.450 ⇒ 00:17:38.399 Samuel Roberts: Yeah, I mean, it’s not self-healing, we’re healing it, but it’s that kind of idea where we’re doing… Yeah, I totally see what you’re saying, though. That’s kind of what I was hoping we can extract from these.
208 00:17:38.400 ⇒ 00:17:48.120 Uttam Kumaran: But see, also, like, you got to find the ceiling on that. Like, if it’s better for us to insert a human in the loop step.
209 00:17:48.380 ⇒ 00:17:52.090 Uttam Kumaran: Versus, like, continuing to iterate on evals.
210 00:17:52.220 ⇒ 00:17:55.849 Uttam Kumaran: That’s what I think you guys need to have an opinion about, which is…
211 00:17:55.850 ⇒ 00:17:56.400 Samuel Roberts: Yeah.
212 00:17:57.490 ⇒ 00:18:00.119 Uttam Kumaran: Like, for, like, the very simple way is, like.
213 00:18:00.270 ⇒ 00:18:07.199 Uttam Kumaran: There is a… there is, like, a trade-off between more human in the loop and quality of the output.
214 00:18:07.400 ⇒ 00:18:08.320 Uttam Kumaran: Right?
215 00:18:08.820 ⇒ 00:18:11.389 Uttam Kumaran: There is also a trade-off between us
216 00:18:11.610 ⇒ 00:18:18.030 Uttam Kumaran: There’s also a limitation of what we can do with just the prompts, with prompt optimization, with agent stitching.
217 00:18:18.500 ⇒ 00:18:23.810 Uttam Kumaran: Well, of course, at the plateau, you… with human in the loop, it’s just no… there’s just human.
218 00:18:24.000 ⇒ 00:18:29.220 Uttam Kumaran: Right? So, I’m not saying we’re close to that, but I do think that you can…
219 00:18:29.490 ⇒ 00:18:33.040 Uttam Kumaran: I think you guys should start to build a mental model for…
220 00:18:33.530 ⇒ 00:18:53.249 Uttam Kumaran: what is the order of operations in optimizing this sort of, like, a closed-loop system like this? It’s similar… it’s similar, not in scale, but it’s a similar problem to what we’re seeing in ABC, and we’re gonna see this on every sort of, like, agentic automation we build, is the need to understand the trade-offs of what
221 00:18:53.880 ⇒ 00:19:04.419 Uttam Kumaran: like, what moves the needle, and then what is this, like, natural self-healing that we also implement over time. We’re not gonna get it right, but we need to prove how far we are.
222 00:19:04.550 ⇒ 00:19:15.710 Uttam Kumaran: And then we augment, I think, with Human in the Loop in every step, you know, as much… Many times, again, what is Human in the Loop in our existing process? It’s Matthew giving feedback to the designer.
223 00:19:15.710 ⇒ 00:19:16.550 Samuel Roberts: Right, yeah.
224 00:19:16.550 ⇒ 00:19:24.810 Uttam Kumaran: So, already, it’s not actually that… it’s not that much… it’s not bad to have those, because those already exist, right?
225 00:19:25.450 ⇒ 00:19:29.849 Samuel Roberts: Yeah, no, I think that’s… I mean, really, that’s their value add to their client.
226 00:19:29.850 ⇒ 00:19:30.290 Uttam Kumaran: Yes.
227 00:19:30.290 ⇒ 00:19:43.139 Samuel Roberts: like, it’s… if he’s just putting stuff in Claude and getting outputs and sending that to a designer, like, that… like, what’s the point then? His intuition, his knowledge, his… his craft is that, and I don’t want to remove that at all.
228 00:19:43.140 ⇒ 00:19:43.960 Uttam Kumaran: For sure.
229 00:19:43.960 ⇒ 00:20:03.519 Samuel Roberts: So I think, I think, yeah, thinking of it as more as, like, augmentation, speeding up, and again, like, like you were saying, and like I highlighted at the end, I think the… figuring out, like, what is the most high-leveraged tasks or feedback, you know what I mean? Like, he still should be providing that, but ideally it becomes less, like, rote.
230 00:20:03.650 ⇒ 00:20:04.750 Samuel Roberts: consistent.
231 00:20:04.860 ⇒ 00:20:14.579 Samuel Roberts: oh, no, this is too long, oh, this is too short, or like, it becomes a more, like, okay, this can be crafted in a slightly different way, rather than… like you were saying, that kind of first pass.
232 00:20:14.730 ⇒ 00:20:22.469 Samuel Roberts: that is always the same kind of feedback is probably a good spot to start. But keeping that, like, next step where he’s still
233 00:20:22.980 ⇒ 00:20:28.099 Samuel Roberts: Or someone is still… Like, gut checking it, or just…
234 00:20:28.100 ⇒ 00:20:34.050 Uttam Kumaran: And I think this is where, like, for Matthew, I just don’t think that they have a fixed enough process for him to, like.
235 00:20:34.050 ⇒ 00:20:37.049 Samuel Roberts: For you to propose, hey, if we had the agent.
236 00:20:37.360 ⇒ 00:20:42.580 Uttam Kumaran: ask you for input here, would you be okay with… like, I don’t think he’s gonna be able to answer that. So…
237 00:20:42.580 ⇒ 00:20:42.950 Samuel Roberts: Oh.
238 00:20:42.950 ⇒ 00:20:45.310 Uttam Kumaran: My opinion is to just do.
239 00:20:45.420 ⇒ 00:20:48.889 Uttam Kumaran: Yeah. Like, start to do whatever it, like.
240 00:20:49.020 ⇒ 00:20:54.149 Uttam Kumaran: you can either go one or two ways. You can just start layering on more of that, naturally, or…
241 00:20:54.520 ⇒ 00:20:59.639 Uttam Kumaran: You layer on a lot, and you sort of strip from there, as you sort of, like, figure out
242 00:20:59.900 ⇒ 00:21:06.779 Uttam Kumaran: But easy things to cross off is just maybe now that we have a lot of the finished ones, we could do another pass at prompts.
243 00:21:07.130 ⇒ 00:21:13.350 Uttam Kumaran: And then, I think we sort of, like…
244 00:21:14.140 ⇒ 00:21:20.809 Uttam Kumaran: try to backtest, like, basically some combination of the existing system plus human loop that should get us to our next
245 00:21:20.930 ⇒ 00:21:22.660 Uttam Kumaran: The next mile, you know.
246 00:21:22.910 ⇒ 00:21:25.620 Samuel Roberts: Yeah, no, I think that’s… I think that’s good.
247 00:21:27.780 ⇒ 00:21:37.759 Uttam Kumaran: Because it’s just such a… this is not a… I just think they’re not… they don’t… they don’t… they’re not thinking much… they’re not thinking very systematically. There is processes for them, but…
248 00:21:38.590 ⇒ 00:21:41.670 Uttam Kumaran: I think this is the first time he’s basically broken it down, and so.
249 00:21:41.670 ⇒ 00:21:42.390 Samuel Roberts: Yeah.
250 00:21:42.600 ⇒ 00:21:47.239 Uttam Kumaran: We may know as much as he knows about, like, how this works in their company.
251 00:21:47.240 ⇒ 00:21:47.650 Samuel Roberts: That’s true.
252 00:21:47.650 ⇒ 00:21:52.030 Uttam Kumaran: So here, from here, it’s net new processes we’re proposing, you know?
253 00:21:52.030 ⇒ 00:21:59.060 Samuel Roberts: Yeah, that’s why I was thinking a little bit with that, like, generating some basic layouts or, like, low-fidelity stuff, that might be…
254 00:21:59.280 ⇒ 00:22:00.240 Uttam Kumaran: Oh, yeah.
255 00:22:00.240 ⇒ 00:22:06.629 Samuel Roberts: let’s try it out and see, like, whether or not that’s gamma or that other thing that I saw that maybe is just marked down from…
256 00:22:06.930 ⇒ 00:22:07.390 Uttam Kumaran: Yes.
257 00:22:07.390 ⇒ 00:22:18.780 Samuel Roberts: you know, an LLM, just something that’s like, this is the narrative, this is the structure, now what could that look like in a however-many-slide-deck kind of thing? Might just be something that at least.
258 00:22:19.580 ⇒ 00:22:19.900 Uttam Kumaran: Yeah.
259 00:22:19.900 ⇒ 00:22:23.900 Samuel Roberts: oh, no, no, the designer needs to be the one to decide that. Or, oh, that’s great.
260 00:22:23.900 ⇒ 00:22:26.829 Uttam Kumaran: And I can run with that.
261 00:22:26.830 ⇒ 00:22:27.619 Samuel Roberts: Throwing it out is good.
262 00:22:27.620 ⇒ 00:22:37.270 Uttam Kumaran: Yeah, we should maybe call that person, maybe Sylvia or whatever, we should maybe call her, and then… maybe we should take a first pass, because also, I think you guys should also maybe…
263 00:22:37.400 ⇒ 00:22:39.500 Uttam Kumaran: show this demo to Hannah.
264 00:22:39.830 ⇒ 00:22:42.399 Uttam Kumaran: Because Hannah does a lot of deck building.
265 00:22:42.520 ⇒ 00:22:55.470 Uttam Kumaran: And it almost may be helpful for her to, like, give you her feedback on, like, what she may expect. We have the same problem, where I give her very limited input and transcripts, and she turns it into decks. Of course, we don’t only build decks.
266 00:22:55.480 ⇒ 00:22:59.849 Samuel Roberts: Right. So, but I think she actually has a really good understanding of, like.
267 00:22:59.850 ⇒ 00:23:04.120 Uttam Kumaran: what information she needs as a designer. I think you should…
268 00:23:04.450 ⇒ 00:23:07.909 Uttam Kumaran: Interview her for a sec about this, and then…
269 00:23:08.130 ⇒ 00:23:11.880 Uttam Kumaran: try something and then ship it to Sylvia and see what she thinks at that point.
270 00:23:12.310 ⇒ 00:23:13.389 Samuel Roberts: That’s a great idea.
271 00:23:13.590 ⇒ 00:23:14.250 Uttam Kumaran: Yeah.
272 00:23:15.140 ⇒ 00:23:19.430 Samuel Roberts: Yeah, I feel a little lost with some of that, so having someone that has a little bit of,
273 00:23:19.660 ⇒ 00:23:21.889 Samuel Roberts: Knowledge and experience in that area would be a good.
274 00:23:21.890 ⇒ 00:23:22.610 Uttam Kumaran: Yeah.
275 00:23:22.920 ⇒ 00:23:25.800 Samuel Roberts: Good idea. That’s great. Yeah, we should set that up.
276 00:23:25.910 ⇒ 00:23:29.310 Samuel Roberts: Okay.
277 00:23:29.710 ⇒ 00:23:30.280 Uttam Kumaran: Cool.
278 00:23:30.960 ⇒ 00:23:35.669 Uttam Kumaran: So, I don’t know how… I think you’re gonna probably have to use this transcript to create…
279 00:23:35.670 ⇒ 00:23:39.149 Samuel Roberts: Yeah, I was already like, oh my gosh, there’s so much stuff, but it’s there.
280 00:23:39.150 ⇒ 00:23:48.240 Uttam Kumaran: the tickets, Rico, so maybe, like, use this to create those tickets, maybe propose it in Slack, and then me and Sam could give a little feedback, but yeah.
281 00:23:48.240 ⇒ 00:23:48.820 Samuel Roberts: Yeah.
282 00:23:50.410 ⇒ 00:23:54.510 Uttam Kumaran: Not this transcript, meaning, like, the transcript of this meeting we’re in right now.
283 00:23:56.630 ⇒ 00:23:57.409 Rico Rejoso: Got it.
284 00:23:57.410 ⇒ 00:24:09.999 Mustafa Raja: For the… what’s it called? Back, backtesting the evals, are they only going to send us the decks, the finalized decks, or the inputs for those decks, too? Because we would need those inputs, right?
285 00:24:10.000 ⇒ 00:24:16.759 Samuel Roberts: Yeah, I think the… I think he was only sending us the final decks, but we can probably, at least for some of them, get their initial…
286 00:24:17.340 ⇒ 00:24:23.559 Samuel Roberts: Like, the transcript and the client description. Like, at least something that’s approximate to what they would give for that. Obviously there’s.
287 00:24:23.560 ⇒ 00:24:26.730 Mustafa Raja: The transcripts and the questionnaires.
288 00:24:26.980 ⇒ 00:24:27.410 Samuel Roberts: Yeah, so…
289 00:24:27.410 ⇒ 00:24:30.030 Mustafa Raja: have those, we would be good to go.
290 00:24:30.030 ⇒ 00:24:33.630 Samuel Roberts: I’m willing to bet they still have that for that stuff somewhere, you know, I don’t know how…
291 00:24:35.050 ⇒ 00:24:37.930 Samuel Roberts: how they keep all that, but I doubt they get rid of it.
292 00:24:38.380 ⇒ 00:24:38.710 Mustafa Raja: Again.
293 00:24:38.710 ⇒ 00:24:51.000 Samuel Roberts: We might have to ask them to put that together, and if they can do that over time, at least then we can start adding more, because that did look like a lot of final versions. I think we can ask for that once they send it.
294 00:24:51.310 ⇒ 00:24:56.069 Uttam Kumaran: They’re gonna have it… they’re gonna have all of it in their… whatever that recording platform is.
295 00:24:56.070 ⇒ 00:24:56.740 Mustafa Raja: the circuit.
296 00:24:56.740 ⇒ 00:24:58.620 Samuel Roberts: Oh, circle back, yeah, okay.
297 00:24:58.620 ⇒ 00:25:01.820 Uttam Kumaran: I just… I can ask and say, can you give us circle back access?
298 00:25:02.110 ⇒ 00:25:03.940 Mustafa Raja: I have that. I have…
299 00:25:03.940 ⇒ 00:25:04.350 Uttam Kumaran: Oh.
300 00:25:05.190 ⇒ 00:25:06.350 Uttam Kumaran: It’s gotta be in there, right?
301 00:25:06.680 ⇒ 00:25:08.150 Mustafa Raja: Rafi’s account.
302 00:25:08.500 ⇒ 00:25:08.880 Samuel Roberts: Oh, okay.
303 00:25:08.880 ⇒ 00:25:13.820 Mustafa Raja: I logged in a few days ago for the transcripts, I don’t know if I still have the access.
304 00:25:14.630 ⇒ 00:25:19.179 Samuel Roberts: Okay, yeah, we’ll take a look at that, and then maybe once they send the Dropbox.
305 00:25:19.520 ⇒ 00:25:23.040 Samuel Roberts: Go in and pull… the transcripts…
306 00:25:23.510 ⇒ 00:25:31.670 Samuel Roberts: for as many of those clients as you can find. Yeah. And then we can… figure out…
307 00:25:32.080 ⇒ 00:25:44.590 Samuel Roberts: the best way to pull stuff from the final versions, and whether or not that’s just text to start, and then passing in other things, and having it described, I don’t know. There’s a fun ingestion problem there to figure out, which will be cool.
308 00:25:44.590 ⇒ 00:25:51.180 Mustafa Raja: Yeah. And in terms of prompts, system prompts, I believe that there’s a lot that can be improved.
309 00:25:51.390 ⇒ 00:25:56.320 Mustafa Raja: If you take a look, the system prompts aren’t really…
310 00:25:56.460 ⇒ 00:26:03.249 Mustafa Raja: In a too good of a shape, so we can definitely, make good progress in there.
311 00:26:03.750 ⇒ 00:26:04.320 Samuel Roberts: Yeah.
312 00:26:04.590 ⇒ 00:26:06.410 Samuel Roberts: Yeah, this would be a good way to do that, too.
313 00:26:08.410 ⇒ 00:26:10.420 Samuel Roberts: Oh, great. Okay.
314 00:26:13.470 ⇒ 00:26:20.870 Samuel Roberts: Yeah, I definitely see what you’re saying about their processes, now that we’ve had this… when we had the call earlier, it definitely made…
315 00:26:21.050 ⇒ 00:26:24.740 Samuel Roberts: just, like, creative kind of endeavors, I feel like.
316 00:26:25.810 ⇒ 00:26:33.469 Samuel Roberts: They’re making lots of decisions all the time and doing lots of stuff, but it’s not as structured, and we can help add a little bit of structure by pulling
317 00:26:33.880 ⇒ 00:26:37.739 Samuel Roberts: Them out of the loop and putting them back in in certain spots, and see how it works for them.
318 00:26:39.690 ⇒ 00:26:40.530 Uttam Kumaran: Yeah.
319 00:26:41.880 ⇒ 00:26:43.170 Samuel Roberts: It’s a good way to think about it.
320 00:26:44.810 ⇒ 00:26:45.570 Samuel Roberts: Cool.
321 00:26:46.400 ⇒ 00:26:50.399 Samuel Roberts: What else are we… To talk about this.
322 00:26:50.820 ⇒ 00:26:51.990 Rico Rejoso: Yep,
323 00:26:52.690 ⇒ 00:26:58.389 Rico Rejoso: That’s for interlude. I’ll get the transcript for this meeting later on and present the tickets by…
324 00:26:58.390 ⇒ 00:26:58.720 Samuel Roberts: Cool.
325 00:26:59.040 ⇒ 00:27:00.109 Rico Rejoso: I get that.
326 00:27:00.220 ⇒ 00:27:06.029 Rico Rejoso: Another thing on our agenda would be, look at the… AI team tickets.
327 00:27:06.230 ⇒ 00:27:07.420 Samuel Roberts: Okay, yes.
328 00:27:07.590 ⇒ 00:27:15.509 Samuel Roberts: That’s right, yeah, we were talking about that, yeah, the other day. There’s a lot in the backlog for, like, AI internal.
329 00:27:16.610 ⇒ 00:27:20.540 Samuel Roberts: the question is sifting through that a little bit, because I don’t know…
330 00:27:20.940 ⇒ 00:27:25.669 Samuel Roberts: I don’t… a lot of it predates me. I don’t necessarily know what has been still important.
331 00:27:25.670 ⇒ 00:27:28.120 Uttam Kumaran: Yeah, I think almost, like, where…
332 00:27:29.550 ⇒ 00:27:36.819 Uttam Kumaran: I honestly will do what, like, I’m supposed to do at this moment, which is, say, follow the OKRs.
333 00:27:37.260 ⇒ 00:27:38.400 Rico Rejoso: Perfect.
334 00:27:38.400 ⇒ 00:27:48.339 Uttam Kumaran: Which is, like, if we were to talk about, like, what those are, the things that really need to happen and will actually solve for this
335 00:27:48.440 ⇒ 00:27:56.560 Uttam Kumaran: is… making sure that all of the PM sales marketing workflows are listed in that thing, right?
336 00:27:56.870 ⇒ 00:28:02.979 Uttam Kumaran: The second thing is… Creating some sort of understanding of usage.
337 00:28:03.100 ⇒ 00:28:08.189 Uttam Kumaran: Yeah. So, I would… and then I think a bonus is if we can start to get tickets
338 00:28:08.780 ⇒ 00:28:12.019 Uttam Kumaran: Like, the ticket creation processing is a huge bottleneck right now.
339 00:28:12.540 ⇒ 00:28:13.539 Uttam Kumaran: But again, like.
340 00:28:13.710 ⇒ 00:28:26.930 Uttam Kumaran: I would say that’s a bonus, like, out of… out of those two. So, I think, Rico, I don’t know whether also, like, and this is maybe a question for Sam, you and Mustafa, like, maybe… is it… is it helpful for us to, like.
341 00:28:27.630 ⇒ 00:28:35.299 Uttam Kumaran: map tickets to, like, the OKR, basically. Like, we should just have the OKR… we should have the KRs in here as projects.
342 00:28:35.830 ⇒ 00:28:36.859 Uttam Kumaran: And then…
343 00:28:37.120 ⇒ 00:28:44.690 Uttam Kumaran: Because, like, ideally, we shouldn’t be working on things, for the most part, that aren’t tied to a key result.
344 00:28:44.690 ⇒ 00:28:45.879 Samuel Roberts: For example… That’s a good point.
345 00:28:45.900 ⇒ 00:28:56.589 Uttam Kumaran: If, like, let’s say getting… one ticket is, like, getting departments, that goes to the fact that there’s probably some people that aren’t using the platform because departments isn’t there, right?
346 00:28:56.810 ⇒ 00:28:57.300 Samuel Roberts: Definitely.
347 00:28:57.300 ⇒ 00:28:57.830 Uttam Kumaran: Google.
348 00:28:58.680 ⇒ 00:29:02.209 Uttam Kumaran: I don’t know, I kind of feel like for internal teams.
349 00:29:02.390 ⇒ 00:29:07.349 Uttam Kumaran: And now you’re talking to Rico, who’s running PM for all internal teams.
350 00:29:07.560 ⇒ 00:29:18.799 Uttam Kumaran: I think maybe it’s best we do a project per KR. Maybe, Rico, you can talk to the whole delivery team about this. Yeah. Maybe we do a project per KR, that way you can ladder tasks
351 00:29:19.230 ⇒ 00:29:23.399 Uttam Kumaran: To those, and then it makes easy for us to be like, oh, this doesn’t fit in one of these.
352 00:29:24.170 ⇒ 00:29:29.629 Uttam Kumaran: You know, and… or also, again, like, I can tell you which KR is more important than other.
353 00:29:30.070 ⇒ 00:29:38.559 Uttam Kumaran: So, yeah, and I think… I think we don’t have to delete all these tickets, we just put them in the backlog, but I do think that’s… that’s probably the best way to go.
354 00:29:38.850 ⇒ 00:29:39.450 Samuel Roberts: Yeah.
355 00:29:39.600 ⇒ 00:29:40.230 Uttam Kumaran: I think.
356 00:29:40.230 ⇒ 00:29:40.760 Samuel Roberts: Yeah.
357 00:29:40.760 ⇒ 00:29:43.069 Uttam Kumaran: Documenting is gonna help.
358 00:29:43.960 ⇒ 00:29:46.930 Uttam Kumaran: It’s just gonna solve all of those.
359 00:29:47.130 ⇒ 00:29:50.000 Uttam Kumaran: It’s gonna be able to solve a lot of problems.
360 00:29:50.180 ⇒ 00:29:52.880 Uttam Kumaran: By systematically going team by team.
361 00:29:52.880 ⇒ 00:29:54.849 Samuel Roberts: Learning what everybody’s doing.
362 00:29:54.850 ⇒ 00:29:59.519 Uttam Kumaran: Because right now, If we’re to keep going the way we’re going.
363 00:29:59.680 ⇒ 00:30:04.440 Uttam Kumaran: I’m the bottleneck for a lot of the ideation, which is, like, always something that I wanted to sort of
364 00:30:04.810 ⇒ 00:30:06.480 Uttam Kumaran: collaborate with.
365 00:30:06.480 ⇒ 00:30:07.010 Samuel Roberts: Yeah.
366 00:30:07.010 ⇒ 00:30:12.640 Uttam Kumaran: Second is… Because of that, we’re stuck in, like, my…
367 00:30:13.690 ⇒ 00:30:16.269 Uttam Kumaran: Ability to creatively think of the solution.
368 00:30:16.440 ⇒ 00:30:24.209 Uttam Kumaran: Which is the bottleneck. Like, so I want to democratize the fact that we all have a clear understanding of what the problems in the
369 00:30:24.430 ⇒ 00:30:28.279 Uttam Kumaran: The problems the stakeholders’ processes in the business are, you know?
370 00:30:28.280 ⇒ 00:30:28.870 Samuel Roberts: Yeah.
371 00:30:28.870 ⇒ 00:30:32.230 Uttam Kumaran: Like, this team is the only team that has to…
372 00:30:33.330 ⇒ 00:30:36.190 Uttam Kumaran: Pretty much know how the whole company works.
373 00:30:36.190 ⇒ 00:30:37.090 Samuel Roberts: Yeah.
374 00:30:37.090 ⇒ 00:30:39.350 Uttam Kumaran: Like, apart from our internal data team.
375 00:30:40.070 ⇒ 00:30:40.760 Samuel Roberts: Right.
376 00:30:40.760 ⇒ 00:30:42.550 Uttam Kumaran: It’s… I would say they… they’re…
377 00:30:42.820 ⇒ 00:30:45.800 Uttam Kumaran: Daria not even touching on some of the stuff, so…
378 00:30:46.190 ⇒ 00:30:48.910 Uttam Kumaran: that’s, like, I think we have to start with the fact that, like.
379 00:30:49.970 ⇒ 00:30:57.930 Uttam Kumaran: everybody on AI… on internal AI has to know almost how, like, or at least have a way to know, like, it’s documented somewhere.
380 00:30:57.930 ⇒ 00:30:58.710 Samuel Roberts: Yeah.
381 00:30:59.670 ⇒ 00:31:03.530 Samuel Roberts: Yeah, the documentation is definitely gonna have to be more of a priority.
382 00:31:03.530 ⇒ 00:31:06.059 Uttam Kumaran: And we’re adding processes every week.
383 00:31:06.060 ⇒ 00:31:07.750 Samuel Roberts: Yeah, I was gonna say, like, it’s only gonna get.
384 00:31:07.750 ⇒ 00:31:25.349 Uttam Kumaran: Ideally, you guys have to think about, like, what is the SOP creation process, and, like, how do we get more people to create SOPs, you know? And so, I don’t know, I’m more of a fan of that. I still think there are some small bugs and things to fix.
385 00:31:25.490 ⇒ 00:31:26.520 Uttam Kumaran: But, like.
386 00:31:26.990 ⇒ 00:31:30.999 Uttam Kumaran: I think we should do it that way, like, I think that really the focus should be on, like.
387 00:31:31.460 ⇒ 00:31:36.879 Uttam Kumaran: crushing through… Some of these, like, the workflow documentation pieces.
388 00:31:37.290 ⇒ 00:31:43.349 Samuel Roberts: That’s smart. Yeah, we were talking about that the other day, making, like, a plan of attack on that, Rico and Amber and I were.
389 00:31:43.770 ⇒ 00:31:46.860 Samuel Roberts: I would say…
390 00:31:47.550 ⇒ 00:31:56.349 Samuel Roberts: Yeah, I mean, we have, like, kind of the rough priority, because we split, like, engineering and NPM marketing sales, right, on the OKRs.
391 00:31:56.470 ⇒ 00:31:57.510 Uttam Kumaran: Yeah.
392 00:31:57.510 ⇒ 00:32:00.160 Samuel Roberts: So, we definitely have our, like, priority there to go…
393 00:32:00.860 ⇒ 00:32:05.370 Samuel Roberts: Is there, is there, like, a particular priority from, like, PM Marketing Sales.
394 00:32:06.540 ⇒ 00:32:09.790 Uttam Kumaran: Pm sales.
395 00:32:10.360 ⇒ 00:32:10.970 Samuel Roberts: Okay.
396 00:32:11.740 ⇒ 00:32:17.389 Uttam Kumaran: The only reason I say PM is because you have Rico, Amber, and Justin, who are all very, very…
397 00:32:17.640 ⇒ 00:32:20.960 Uttam Kumaran: like, they’ll… they’re all using AI stuff.
398 00:32:20.960 ⇒ 00:32:21.670 Samuel Roberts: Yeah, that’s true.
399 00:32:21.670 ⇒ 00:32:22.280 Uttam Kumaran: boat.
400 00:32:22.630 ⇒ 00:32:26.719 Uttam Kumaran: And then sale, so it… That directly affects
401 00:32:26.940 ⇒ 00:32:33.469 Uttam Kumaran: Our ability to get client success at a high margin, and then sales is our ability to make more money.
402 00:32:33.470 ⇒ 00:32:34.990 Samuel Roberts: Of course, that’s… okay, that’s…
403 00:32:34.990 ⇒ 00:32:36.990 Uttam Kumaran: Those are the only two… those are the two that… that…
404 00:32:36.990 ⇒ 00:32:39.140 Samuel Roberts: Okay. Yeah, no, that’s good.
405 00:32:39.140 ⇒ 00:32:45.789 Uttam Kumaran: But with sales, again, the other limiting factor on sales is, like, Justina is really the only 100% dedicated salesperson.
406 00:32:45.900 ⇒ 00:32:49.889 Samuel Roberts: Okay. Ryan is between marketing and sales. Hannah is also…
407 00:32:50.010 ⇒ 00:32:53.199 Uttam Kumaran: Robert and I, you know, are the other. So…
408 00:32:53.440 ⇒ 00:32:56.929 Uttam Kumaran: Justina has to become that person there, you know?
409 00:32:57.190 ⇒ 00:32:57.890 Samuel Roberts: Yeah.
410 00:32:58.150 ⇒ 00:32:59.050 Samuel Roberts: Okay.
411 00:32:59.050 ⇒ 00:32:59.700 Uttam Kumaran: Yeah.
412 00:33:02.880 ⇒ 00:33:10.240 Samuel Roberts: Yeah, I think… okay, so… the thought was there’s just so many tickets, like, I don’t want to just dump everything into the backlog, my thought was to go through it and.
413 00:33:11.290 ⇒ 00:33:12.240 Uttam Kumaran: Why don’t you go through.
414 00:33:12.240 ⇒ 00:33:13.120 Samuel Roberts: Pull out.
415 00:33:13.130 ⇒ 00:33:14.829 Uttam Kumaran: map it to the KRs.
416 00:33:14.830 ⇒ 00:33:21.479 Samuel Roberts: That’s what I was thinking, yeah, that’s a good… but I was already thinking about, like, triaging all of those, and this is a good framework for that now.
417 00:33:23.180 ⇒ 00:33:23.800 Uttam Kumaran: Yeah.
418 00:33:24.060 ⇒ 00:33:26.920 Samuel Roberts: Okay, cool. That sounds like a good plan.
419 00:33:27.320 ⇒ 00:33:28.080 Samuel Roberts: Yeah, I’m gonna…
420 00:33:28.080 ⇒ 00:33:33.900 Rico Rejoso: create the project for each OPR that we have listed, and afterwards we can go through all the tickets.
421 00:33:34.400 ⇒ 00:33:34.740 Samuel Roberts: Yeah.
422 00:33:34.740 ⇒ 00:33:43.059 Rico Rejoso: tickets as well, and try to categorize them. Then, for some that are not, I’m gonna keep them on backlogs, and for checking with income, if we…
423 00:33:43.330 ⇒ 00:33:45.000 Rico Rejoso: Should have those tickets still.
424 00:33:45.180 ⇒ 00:33:45.870 Rico Rejoso: Underlinear.
425 00:33:45.870 ⇒ 00:33:56.510 Samuel Roberts: Yeah. Yeah, I would say once we organize stuff, if things don’t fit into those OKRs, they can… you know, if they’re bug fixes and stuff, I think it’s different than if it’s, like, whole features that aren’t,
426 00:33:57.460 ⇒ 00:34:06.629 Samuel Roberts: driving anything, necessarily. You know, if they’re just ideas, is one thing, but if they’re ideas that tie to an OKR or something, then it’s… then it makes sense. Yeah.
427 00:34:07.840 ⇒ 00:34:08.719 Samuel Roberts: Okay.
428 00:34:09.750 ⇒ 00:34:17.160 Samuel Roberts: And then, I guess the other… so then I need to start setting up times to chat with people, see what, like… I… is Justina probably someone to…
429 00:34:17.659 ⇒ 00:34:18.849 Samuel Roberts: Talk too early?
430 00:34:19.110 ⇒ 00:34:28.650 Uttam Kumaran: Yeah, Justina, I would… on sales, I would talk to Justina and Hannah, and then, of course, like, if you get… if you… if you have more… just ping me with more questions.
431 00:34:28.659 ⇒ 00:34:29.659 Samuel Roberts: Of course, yeah.
432 00:34:29.659 ⇒ 00:34:32.219 Uttam Kumaran: I’m just thinking, like, process-wise, especially if she’s…
433 00:34:32.689 ⇒ 00:34:34.249 Samuel Roberts: New coming in.
434 00:34:34.419 ⇒ 00:34:37.839 Samuel Roberts: Getting up to speed on current processes, building new processes, he’s probably something.
435 00:34:37.840 ⇒ 00:34:50.789 Uttam Kumaran: She’s the one. She’s gonna own… she’s gonna own a lot of the coordination on sales, and it’s gonna be a lot of… something into HubSpot, something out of HubSpot, something into… from the handoff also from sales to ops, sales to finance.
436 00:34:50.790 ⇒ 00:34:51.420 Samuel Roberts: Yep.
437 00:34:51.429 ⇒ 00:34:53.869 Uttam Kumaran: Fails to delivery is gonna be her.
438 00:34:54.440 ⇒ 00:34:59.689 Samuel Roberts: Okay, great. Yeah, we just… we just caught something interesting with the HubSpot integration.
439 00:35:00.280 ⇒ 00:35:09.020 Samuel Roberts: when she… so she and Mustafa chatted about adding a new one, I don’t… I think it was the enrichment, right? Yeah.
440 00:35:09.780 ⇒ 00:35:14.840 Samuel Roberts: Mustafa discovered that you only get one Trigger.
441 00:35:15.320 ⇒ 00:35:16.260 Uttam Kumaran: Oh, really?
442 00:35:16.260 ⇒ 00:35:30.260 Samuel Roberts: into N8N. So apparently, like, what you… Tom, when you and I were talking about, like, the data getting into Supabase, and that not being current, and maybe connecting right to the HubSpot API, I’m not sure that Supabase stuff has been updated at all once we added this new trigger.
443 00:35:30.260 ⇒ 00:35:31.910 Uttam Kumaran: Oh, we’re…
444 00:35:31.910 ⇒ 00:35:46.290 Samuel Roberts: So, the plan right now, is to build a little routing workflow so that we can still automate things from HubSpot and not just have one new thing at a time. And then, Mustaf also looked at
445 00:35:46.740 ⇒ 00:35:56.569 Samuel Roberts: the HubSpot integration into the platform a little bit, just to map it out, because I think… unless there’s a reason to have things in Supabase for, like.
446 00:35:56.980 ⇒ 00:35:59.599 Samuel Roberts: Tracking over time, or…
447 00:35:59.730 ⇒ 00:36:06.989 Samuel Roberts: additional metadata we want that HubSpot doesn’t have. My gut is to just go with HubSpot as the source of truth for a lot of things, and…
448 00:36:07.240 ⇒ 00:36:10.280 Samuel Roberts: Not duplicate just a display kind of thing.
449 00:36:10.620 ⇒ 00:36:16.499 Uttam Kumaran: Yeah, I mean, I also suggest… yeah, I’d rather not duplicate just the display, I think that was just a gap night out.
450 00:36:16.500 ⇒ 00:36:18.349 Samuel Roberts: Yeah, exactly, exactly, but now it won’t…
451 00:36:18.350 ⇒ 00:36:21.200 Uttam Kumaran: Pull it, then totally pull it. The only thing is, like.
452 00:36:22.160 ⇒ 00:36:27.859 Uttam Kumaran: I mean, this is where, like, I don’t want to introduce more outside of CRM fields.
453 00:36:27.970 ⇒ 00:36:36.540 Uttam Kumaran: So, like, every… and I know Flexibility HubSpot, we can add more properties there, and I want the… I want the source of… I don’t want there to be…
454 00:36:36.680 ⇒ 00:36:40.749 Uttam Kumaran: Other places for data on customers or contacts.
455 00:36:40.750 ⇒ 00:36:42.890 Samuel Roberts: Totally, yeah, no, I think that’s… if you wanted…
456 00:36:42.890 ⇒ 00:36:46.069 Uttam Kumaran: We need to be able to write back, probably, at some point.
457 00:36:46.070 ⇒ 00:36:46.850 Samuel Roberts: Yeah, definitely.
458 00:36:46.850 ⇒ 00:36:48.429 Uttam Kumaran: be able to pull reliably.
459 00:36:48.730 ⇒ 00:36:49.390 Samuel Roberts: Okay.
460 00:36:49.630 ⇒ 00:37:04.789 Samuel Roberts: Oh no, my headphones are gonna die. Just heard the little blither. Yeah, so we are… yeah, the HubSpot things will be good, that’s all. Maybe, I’ll get with Justina to learn a little bit more about what she’s thinking and how things are going that way, map some stuff out.
461 00:37:06.000 ⇒ 00:37:07.219 Samuel Roberts: Alright, cool.
462 00:37:07.500 ⇒ 00:37:12.500 Samuel Roberts: Go through the backlog, yeah, add the project.
463 00:37:12.500 ⇒ 00:37:27.940 Mustafa Raja: While we’re at it, Justina, for Justina’s enrichment, I used the Apollos instance that we have, in our clay, for the, contact enrichment. Let me know if we want to keep doing that.
464 00:37:29.030 ⇒ 00:37:33.249 Uttam Kumaran: That’s fine, I don’t have another… I don’t have another,
465 00:37:34.190 ⇒ 00:37:43.960 Uttam Kumaran: sort of enrichment solution for you. The only other thing that we’ve tried is previously we used EXA,
466 00:37:44.280 ⇒ 00:37:46.419 Uttam Kumaran: And then you can also use perplexity.
467 00:37:48.790 ⇒ 00:37:51.699 Uttam Kumaran: Probably what I would suggest for now. I can… I mean, like.
468 00:37:52.260 ⇒ 00:38:00.250 Uttam Kumaran: we’re not at the scale of, like, default, where we need, like, a bunch, so, like… and we’re not enriching, like, anything much crazy, so I just need to know their, like…
469 00:38:00.420 ⇒ 00:38:04.990 Uttam Kumaran: the company… Ideally, their name, where they’re located.
470 00:38:05.130 ⇒ 00:38:09.299 Uttam Kumaran: And then the company enrichment, we can pull from a bunch of places, but Apollo’s fine.
471 00:38:09.300 ⇒ 00:38:09.860 Samuel Roberts: room.
472 00:38:11.880 ⇒ 00:38:19.969 Mustafa Raja: She has about 2,500 more accounts that she wants to be enriched, so I’ll run that then.
473 00:38:20.350 ⇒ 00:38:23.249 Uttam Kumaran: Can you filter to just, can you.
474 00:38:24.630 ⇒ 00:38:31.250 Mustafa Raja: Yeah, if you have the company and name and all, I’ll filter that. I wouldn’t enrich that contact, right?
475 00:38:33.690 ⇒ 00:38:39.939 Uttam Kumaran: Yeah, I’m just worried some of those contacts are, like, just random people, but I guess it’s fine.
476 00:38:40.620 ⇒ 00:38:45.490 Samuel Roberts: What, I mean, so right now, it’s enriching new… Entries on the phone.
477 00:38:45.490 ⇒ 00:38:45.960 Mustafa Raja: So the…
478 00:38:45.960 ⇒ 00:38:46.959 Samuel Roberts: Is that what it was?
479 00:38:46.960 ⇒ 00:38:51.159 Mustafa Raja: Yeah, so the workflow, what it does is it listens
480 00:38:51.160 ⇒ 00:39:09.939 Mustafa Raja: To any new contacts, that have been added in HubSpot via, any form, right? But they are already, according to Justina, there are, 2,500 about, accounts sitting in HubSpot that she needs to get enriched.
481 00:39:09.990 ⇒ 00:39:13.090 Mustafa Raja: So, so this is actually…
482 00:39:13.150 ⇒ 00:39:16.089 Mustafa Raja: Some backlog that we need to enrich.
483 00:39:16.610 ⇒ 00:39:17.800 Samuel Roberts: Yeah, okay.
484 00:39:18.370 ⇒ 00:39:25.699 Samuel Roberts: Yeah, I don’t know if there’s a good… I mean, I don’t know the details of where they all came from and not, but if there’s a good way to, like, prioritize better ones, or…
485 00:39:25.860 ⇒ 00:39:34.250 Samuel Roberts: I don’t know, I don’t know how… I don’t know where they’re all coming from, but… And I also don’t know how much that would be on Clay, like, I don’t remember the credit system they had.
486 00:39:36.210 ⇒ 00:39:42.959 Mustafa Raja: It wouldn’t cost anything because, the cost, the cost is being veered by our Apollo key.
487 00:39:43.470 ⇒ 00:39:44.840 Samuel Roberts: Right, okay, but, yeah.
488 00:39:44.840 ⇒ 00:39:49.620 Mustafa Raja: The rest of the workflow doesn’t really use any credits at all.
489 00:39:49.920 ⇒ 00:39:53.219 Samuel Roberts: Okay, but Apollo then, is there anything… okay.
490 00:39:53.810 ⇒ 00:39:55.680 Mustafa Raja: It’s going to be Apollo Coast.
491 00:39:56.260 ⇒ 00:39:58.780 Samuel Roberts: Okay, do we have a sense of… How they work?
492 00:39:58.780 ⇒ 00:39:59.300 Mustafa Raja: Aww.
493 00:39:59.300 ⇒ 00:40:01.450 Samuel Roberts: I don’t know them in detail. Okay.
494 00:40:01.450 ⇒ 00:40:08.730 Mustafa Raja: don’t, have, the access to, upload also.
495 00:40:08.860 ⇒ 00:40:10.899 Mustafa Raja: the key was there, and I just used it.
496 00:40:12.120 ⇒ 00:40:14.899 Samuel Roberts: Just wondering what their, like, pricing plan… are we gonna hit anything?
497 00:40:14.900 ⇒ 00:40:20.449 Uttam Kumaran: Oh, yeah, we… well, we were… we were on their startup thing, and then I think I paid for a year.
498 00:40:20.660 ⇒ 00:40:21.309 Uttam Kumaran: I’ll bet.
499 00:40:21.310 ⇒ 00:40:21.640 Mustafa Raja: Totally.
500 00:40:21.640 ⇒ 00:40:22.230 Samuel Roberts: Okay.
501 00:40:22.230 ⇒ 00:40:24.709 Uttam Kumaran: basket of out. You have a basket of credits.
502 00:40:25.390 ⇒ 00:40:27.360 Samuel Roberts: Right, okay, I’m seeing that here in the pressing now.
503 00:40:27.360 ⇒ 00:40:27.960 Uttam Kumaran: Yeah.
504 00:40:28.640 ⇒ 00:40:31.789 Samuel Roberts: Don’t know what they’re… The criticisms are always so, like…
505 00:40:31.790 ⇒ 00:40:37.619 Uttam Kumaran: I don’t know, Apollo is the worst, they just change it again. So, there is just a vast amount of, like.
506 00:40:38.670 ⇒ 00:40:45.239 Uttam Kumaran: credits in there that you guys can use for whatever. Okay. We’re not gonna… I don’t think we’ll hit the limits, we just don’t have, like, that much.
507 00:40:45.240 ⇒ 00:40:53.450 Samuel Roberts: I mean, that’s a good point. We’re not doing stuff at, like you said, the scale to fall. I was even thinking that when I was setting up the post hoc, and I was looking at the limits, and I’m like, this is built for, like.
508 00:40:53.450 ⇒ 00:40:54.419 Uttam Kumaran: Yeah, yeah, yeah.
509 00:40:54.420 ⇒ 00:40:58.280 Samuel Roberts: Consumer-facing, like, they have a pretty generous free tier for especially our use case.
510 00:40:58.280 ⇒ 00:40:59.270 Uttam Kumaran: Yeah.
511 00:40:59.610 ⇒ 00:41:03.279 Samuel Roberts: Which is also a nice thing to keep in mind in general as we’re building, like, we’re not…
512 00:41:03.740 ⇒ 00:41:07.849 Samuel Roberts: The production… our production-ready thing for ourselves, at least.
513 00:41:07.850 ⇒ 00:41:09.340 Uttam Kumaran: No, it’s very different, yeah.
514 00:41:09.340 ⇒ 00:41:11.159 Samuel Roberts: It’s… yeah, I gotta keep that in mind a little bit.
515 00:41:11.160 ⇒ 00:41:14.580 Uttam Kumaran: That’s why we’re running Heroku, I think, basically for free.
516 00:41:14.580 ⇒ 00:41:15.469 Samuel Roberts: Yeah. Okay.
517 00:41:15.850 ⇒ 00:41:19.940 Uttam Kumaran: most of the stuff, I think, will end up running for free, and additionally, like.
518 00:41:20.690 ⇒ 00:41:26.499 Uttam Kumaran: If we’re implementing the tool for customer, let me know, because I could probably get us just free instance of it.
519 00:41:27.450 ⇒ 00:41:28.210 Samuel Roberts: Hmm.
520 00:41:28.440 ⇒ 00:41:35.939 Uttam Kumaran: like, we have Omni for free, we have Real for free, like, we have a bunch of the stuff we do for clients for free, because.
521 00:41:35.940 ⇒ 00:41:39.489 Samuel Roberts: I call them, and we do a partnership. Thank you, that’s smart.
522 00:41:39.490 ⇒ 00:41:44.779 Uttam Kumaran: They gave us a demo instance, so… Very, very…
523 00:41:44.970 ⇒ 00:41:47.459 Uttam Kumaran: Very, very trying to be cheap in that way.
524 00:41:47.460 ⇒ 00:41:49.319 Samuel Roberts: It works, it works, you know?
525 00:41:49.460 ⇒ 00:41:50.500 Samuel Roberts: Shannon.
526 00:41:51.140 ⇒ 00:41:51.960 Samuel Roberts: Cool.
527 00:41:52.830 ⇒ 00:41:53.540 Uttam Kumaran: Okay.
528 00:41:55.090 ⇒ 00:41:56.270 Samuel Roberts: Right?
529 00:41:57.020 ⇒ 00:42:00.959 Uttam Kumaran: I better get some food, but I guess, like.
530 00:42:02.060 ⇒ 00:42:06.339 Uttam Kumaran: I don’t know, I guess, Rico, do you want to take a first pass of stuff, or are you…
531 00:42:06.580 ⇒ 00:42:12.500 Uttam Kumaran: And same, take a first pass, and then, yeah, I think Sam, I mean, we should just spend maybe, like, an hour or two next week.
532 00:42:12.970 ⇒ 00:42:19.869 Uttam Kumaran: kind of thinking through… I can just give you… just be around and sort of work through this, or work on platform stuff, so…
533 00:42:19.870 ⇒ 00:42:20.530 Samuel Roberts: Yeah.
534 00:42:20.820 ⇒ 00:42:21.450 Uttam Kumaran: Yeah.
535 00:42:21.850 ⇒ 00:42:30.910 Samuel Roberts: Yeah, so, yeah, Rico, make those projects, and then either we can go together, or I can just go through them, probably, and just start grouping things in the right manner.
536 00:42:32.570 ⇒ 00:42:34.900 Samuel Roberts: for… for that, and if I have questions.
537 00:42:35.450 ⇒ 00:42:47.120 Samuel Roberts: I’ll ping whoever maybe made the ticket, or… because I don’t know how… you know, I’m sure they’re not super groomed, it’s probably just titles and stuff that I might be like, what was this even referring to back in, you know, April or something, but…
538 00:42:48.790 ⇒ 00:42:49.200 Rico Rejoso: Okay.
539 00:42:49.200 ⇒ 00:42:50.360 Samuel Roberts: That sounds like a good plan.
540 00:42:50.500 ⇒ 00:42:51.690 Samuel Roberts: Yep.
541 00:42:53.020 ⇒ 00:42:55.320 Samuel Roberts: And then… oh, go ahead.
542 00:42:55.790 ⇒ 00:43:08.260 Rico Rejoso: Yeah, I’m gonna create the projects, I’ll finalize them first, also run the transcript for this meeting for Interlude, post it to our channel, internal channel, for your checking as well. And…
543 00:43:09.010 ⇒ 00:43:10.200 Rico Rejoso: Aside from that.
544 00:43:10.500 ⇒ 00:43:15.079 Rico Rejoso: Yeah, maybe we can set up a meeting by Monday or Tuesday to groom through the tickets as well.
545 00:43:17.180 ⇒ 00:43:27.610 Samuel Roberts: Yeah, yeah, we can do that. I’m gonna try to take a pass this afternoon, at least on some of them, so we’ll hopefully make some progress, but there’s a chunk there, but… Okay.
546 00:43:27.610 ⇒ 00:43:28.180 Rico Rejoso: Let’s…
547 00:43:28.540 ⇒ 00:43:29.230 Samuel Roberts: Okay.
548 00:43:30.300 ⇒ 00:43:32.640 Samuel Roberts: I don’t know if I have…
549 00:43:33.300 ⇒ 00:43:39.240 Samuel Roberts: I don’t know if I have anything else. Uten, did you see the email from… Tim at ABC.
550 00:43:40.900 ⇒ 00:43:42.150 Samuel Roberts: I don’t know, I don’t know how…
551 00:43:42.150 ⇒ 00:43:47.880 Uttam Kumaran: Yeah, I did. I guess I forgot… I had a thought, and I forgot what I thought about it.
552 00:43:47.880 ⇒ 00:43:51.239 Samuel Roberts: Yeah, I think I’m… I totally understand them wanting, like.
553 00:43:51.690 ⇒ 00:43:52.210 Uttam Kumaran: Oh!
554 00:43:52.440 ⇒ 00:43:53.040 Samuel Roberts: Restrictions.
555 00:43:53.540 ⇒ 00:44:01.760 Samuel Roberts: Because I was… I had that thought, too, and I was just, like, poking around the database, or the 8x8 endpoints, and I was like, there’s a lot here.
556 00:44:01.760 ⇒ 00:44:09.510 Uttam Kumaran: Yeah, totally, I mean, ask them what they want to do, like, if they… if they want our… they don’t want us to store it on our side, or whatever, it’s pretty reasonable.
557 00:44:09.510 ⇒ 00:44:14.920 Samuel Roberts: I wasn’t too worried about it. He said that he wanted to grab 30 minutes with key stakeholders, he said, so I’m hoping that they’ll set something up.
558 00:44:15.250 ⇒ 00:44:18.969 Samuel Roberts: Yeah, like, and again, him is just head of, like, IT, so he provisioned.
559 00:44:19.000 ⇒ 00:44:21.300 Uttam Kumaran: a lot of our Google stuff, and yeah, I don’t…
560 00:44:21.740 ⇒ 00:44:26.970 Uttam Kumaran: I think it’s just helpful for us, or, I mean, you can limit it in many ways, like, we don’t need…
561 00:44:27.960 ⇒ 00:44:34.729 Uttam Kumaran: I don’t know, we don’t think we need, like, all the historical stuff, and we only need certain scopes, so I think you’ll be pretty…
562 00:44:35.450 ⇒ 00:44:36.249 Uttam Kumaran: with, like, what we’re…
563 00:44:36.250 ⇒ 00:44:51.119 Samuel Roberts: Yeah, I’m not too concerned. I totally understand them wanting to, like, have a few considerations there, and I think the provision you said was, like, nothing hits production until they, like, approve something, which is totally fine, because right now we’re just testing and trying to see what we can combine. And then…
564 00:44:51.670 ⇒ 00:45:04.929 Samuel Roberts: Obviously, that would change if we’re going to build the new UI kind of thing, then that would be, like, actually getting something, but that’s a… that’s a bigger discussion for the future. Just right now, I just need access to build a little matching thing and see who’s using Andy, when, and…
565 00:45:05.270 ⇒ 00:45:07.600 Samuel Roberts: Pull some data from that, I think we’ll be fine.
566 00:45:07.810 ⇒ 00:45:09.640 Samuel Roberts: Yes. But, yeah, cool.
567 00:45:09.930 ⇒ 00:45:13.750 Samuel Roberts: Trying to think if there’s anything else.
568 00:45:13.860 ⇒ 00:45:18.399 Samuel Roberts: That I have in top of mind while I got 3 of you here, but I think I’m pretty good.
569 00:45:20.050 ⇒ 00:45:24.499 Rico Rejoso: Yep, thank you so much. I’m good, I’m gonna update you guys on Slack as well.
570 00:45:25.360 ⇒ 00:45:26.010 Samuel Roberts: Awesome.
571 00:45:26.450 ⇒ 00:45:27.150 Rico Rejoso: Alrighty.
572 00:45:27.600 ⇒ 00:45:28.120 Uttam Kumaran: Okay.
573 00:45:28.560 ⇒ 00:45:31.220 Uttam Kumaran: Thank you, guys. Have a great weekend, but I’ll talk to you.
574 00:45:31.220 ⇒ 00:45:31.980 Samuel Roberts: Yeah, totally, you too.
575 00:45:31.980 ⇒ 00:45:33.010 Rico Rejoso: Have a good one.
576 00:45:33.010 ⇒ 00:45:33.819 Mustafa Raja: Thank you.