Meeting Title: AI-App Standup Date: 2025-12-19 Meeting participants: Samuel Roberts, Mustafa Raja, Casie Aviles, Gabriel Lam, Uttam Kumaran
WEBVTT
1 00:00:41.790 ⇒ 00:00:42.680 Mustafa Raja: Hay.
2 00:00:46.020 ⇒ 00:00:46.800 Samuel Roberts: Ayy.
3 00:00:49.240 ⇒ 00:00:50.340 Samuel Roberts: How’s it going?
4 00:00:51.300 ⇒ 00:00:53.089 Mustafa Raja: Yeah, doing good. How are you?
5 00:00:53.640 ⇒ 00:00:59.399 Samuel Roberts: Good, good, just… Cranking away on this Lulu stuff for a meeting today.
6 00:01:00.340 ⇒ 00:01:02.460 Mustafa Raja: Oh, it’s a, it’s a client meeting.
7 00:01:03.060 ⇒ 00:01:08.399 Mustafa Raja: Yeah, it’s like their weekly check-in, and I’m trying to make some good progress. Yeah, yeah, yeah, makes sense.
8 00:01:08.590 ⇒ 00:01:13.520 Samuel Roberts: And I wanted to, there’s, like, the… they basically have a…
9 00:01:14.310 ⇒ 00:01:27.419 Samuel Roberts: a little chat app, and they want to be able to talk to data from Klaviyo, and using the MCP server and stuff, and they had another app, and we’re kind of migrating it over, and I’m trying to get, like, a few more things done with the Klaviyo stuff, so it looks good.
10 00:01:27.730 ⇒ 00:01:30.390 Samuel Roberts: We’re making good progress, though.
11 00:01:31.160 ⇒ 00:01:32.249 Mustafa Raja: Yeah, that’s nice.
12 00:01:34.590 ⇒ 00:01:35.739 Samuel Roberts: Ugh, yeah.
13 00:01:37.890 ⇒ 00:01:41.229 Samuel Roberts: Let’s give it a few more minutes for other people, I guess, I don’t know…
14 00:01:42.380 ⇒ 00:01:45.290 Samuel Roberts: It was even… I guess I didn’t… Absolutely.
15 00:01:46.370 ⇒ 00:01:49.720 Mustafa Raja: Yeah, so for the f- Okay, let’s feed.
16 00:01:50.210 ⇒ 00:01:51.829 Samuel Roberts: No, I mean, was there something…
17 00:01:52.680 ⇒ 00:01:57.999 Mustafa Raja: Yeah, it’s just an update for the fixes, in ABC.
18 00:01:58.220 ⇒ 00:02:03.350 Mustafa Raja: There would be… Three things.
19 00:02:03.470 ⇒ 00:02:04.590 Mustafa Raja: So…
20 00:02:06.400 ⇒ 00:02:18.020 Mustafa Raja: So, first, linking, linking the, service area names and services names, because these are exactly the same thing, but named differently.
21 00:02:18.470 ⇒ 00:02:35.730 Mustafa Raja: Then, we want to update, NATN forms, because these were developed to, these were developed, so that, APC people would adopt to self-serve updates in database, but, NN hasn’t been…
22 00:02:36.260 ⇒ 00:02:43.069 Mustafa Raja: A good experiment, so we need to migrate those forms, and then me and Casey will be updating the database.
23 00:02:43.330 ⇒ 00:02:43.940 Samuel Roberts: Okay.
24 00:02:44.250 ⇒ 00:02:48.099 Mustafa Raja: With the new data, since they are not doing it, we’ll have to do it.
25 00:02:49.580 ⇒ 00:02:51.689 Samuel Roberts: What are you thinking for the new forms?
26 00:02:52.940 ⇒ 00:02:57.490 Mustafa Raja: I haven’t looked into the new forms, Casey said that he’s going to take that.
27 00:02:57.650 ⇒ 00:02:59.910 Samuel Roberts: Okay, okay, cool. I’m thinking we could…
28 00:02:59.910 ⇒ 00:03:05.970 Mustafa Raja: But of course, we… before moving with anything, we’re going to loop you in and all, you know.
29 00:03:05.970 ⇒ 00:03:09.450 Samuel Roberts: Yeah, yeah. Cause I’m thinking we, like, we might need to do a little more…
30 00:03:09.680 ⇒ 00:03:13.069 Samuel Roberts: Custom, and what we… like, depending on what other forms are out there.
31 00:03:13.760 ⇒ 00:03:20.310 Samuel Roberts: Because if we build, like, a custom form, now that we’re hosting on Heroku anyway,
32 00:03:20.420 ⇒ 00:03:26.300 Samuel Roberts: It could, like, fetch all the service areas, fetch all the people, and make it a little easier to fill in, and…
33 00:03:26.940 ⇒ 00:03:33.009 Samuel Roberts: Even give them a look into where, like, what the changes were, and make sure things are updated right.
34 00:03:33.450 ⇒ 00:03:38.630 Samuel Roberts: There’s potentially a few things we can do there, so keep that in mind as you’re looking, that…
35 00:03:39.110 ⇒ 00:03:45.160 Samuel Roberts: like, we have a little more flexibility, maybe, with Heroku, or, you know, whatever hosting we want.
36 00:03:47.010 ⇒ 00:03:49.480 Samuel Roberts: a GCP eventually, I think, but yeah.
37 00:03:52.210 ⇒ 00:03:52.910 Mustafa Raja: Okay.
38 00:03:59.610 ⇒ 00:04:01.930 Samuel Roberts: Oh, dude.
39 00:04:02.090 ⇒ 00:04:02.790 Samuel Roberts: Boom.
40 00:04:03.380 ⇒ 00:04:04.490 Samuel Roberts: My name is Chateau.
41 00:04:05.360 ⇒ 00:04:10.070 Samuel Roberts: is… Okay.
42 00:04:12.110 ⇒ 00:04:15.989 Mustafa Raja: And then, of course, we’re going to go through the… the way…
43 00:04:16.140 ⇒ 00:04:22.679 Mustafa Raja: Central docs are set up, because, of course, there are some issues that we discussed.
44 00:04:22.830 ⇒ 00:04:23.250 Samuel Roberts: Yeah.
45 00:04:23.250 ⇒ 00:04:23.900 Mustafa Raja: one.
46 00:04:24.050 ⇒ 00:04:26.560 Mustafa Raja: The one they’re not structured?
47 00:04:26.740 ⇒ 00:04:38.050 Mustafa Raja: Good enough for some of these answers, and then, and then we also need to, on our side, make sure that, the…
48 00:04:40.670 ⇒ 00:04:52.290 Mustafa Raja: the heading tree is properly configured, you know, so we… since we are looking at the bottom of the information, we need to be into the top of the tree, right?
49 00:04:52.440 ⇒ 00:04:57.209 Mustafa Raja: Yeah, definitely. How that heading is structured in the document.
50 00:04:57.210 ⇒ 00:05:05.199 Samuel Roberts: Yes, yes. One of these… yeah, yeah, we discussed it yesterday, so… Yeah, we did. Yeah, I think you’re right. I think that’s definitely something important.
51 00:05:06.560 ⇒ 00:05:12.659 Mustafa Raja: Okay, let me… Okay, let me share, share my screen, and…
52 00:05:12.870 ⇒ 00:05:17.930 Mustafa Raja: We could look into the Gantt chart, right? We have time, let’s just use.
53 00:05:17.930 ⇒ 00:05:20.289 Samuel Roberts: Yeah, yeah, my whole… yeah, focus on ABCs and why we’re here.
54 00:05:20.290 ⇒ 00:05:21.160 Mustafa Raja: Oh, okay.
55 00:05:24.720 ⇒ 00:05:34.280 Mustafa Raja: Okay, so a lot of these issues, we do not have any proper assignments or ideas on how we’re going to take this. For this, we need to discuss…
56 00:05:35.190 ⇒ 00:05:38.520 Mustafa Raja: This shouldn’t be a big fuss, to be honest.
57 00:05:40.500 ⇒ 00:05:43.059 Mustafa Raja: eval logging… I think this is done.
58 00:05:43.650 ⇒ 00:05:45.929 Mustafa Raja: No, no, this isn’t done, sorry.
59 00:05:47.790 ⇒ 00:05:48.130 Samuel Roberts: Yeah.
60 00:05:48.150 ⇒ 00:05:51.670 Mustafa Raja: Error and latency monitoring.
61 00:05:51.860 ⇒ 00:05:53.120 Mustafa Raja: Hooks.
62 00:05:53.250 ⇒ 00:06:05.479 Mustafa Raja: Alerting… yeah, yeah, we need to… we need to look into that. Anyways, what I did was… So we now have the instances from ABC,
63 00:06:05.590 ⇒ 00:06:09.579 Mustafa Raja: Tim did configure those. We have BigQuery and all.
64 00:06:09.810 ⇒ 00:06:10.840 Mustafa Raja: So…
65 00:06:11.980 ⇒ 00:06:25.450 Mustafa Raja: I was thinking, I could do the new routing logic for the vector store today, and then that would enable us to implement the entry point on…
66 00:06:27.150 ⇒ 00:06:28.040 Mustafa Raja: On this side.
67 00:06:28.040 ⇒ 00:06:28.580 Samuel Roberts: Okay.
68 00:06:30.900 ⇒ 00:06:35.610 Mustafa Raja: So, this would be some progress on… What’s it going?
69 00:06:36.230 ⇒ 00:06:42.320 Mustafa Raja: Migration. But I, which, which instance are we going to go with?
70 00:06:43.500 ⇒ 00:06:49.620 Mustafa Raja: Are we going with the basic Andy version, or are we going with the new, new architecture?
71 00:06:50.560 ⇒ 00:06:57.890 Samuel Roberts: Well, this is where I wanted to kind of test it a little bit more and get some good inputs. I don’t know if that’s kind of in here or not.
72 00:06:58.820 ⇒ 00:06:59.400 Mustafa Raja: Yeah.
73 00:06:59.400 ⇒ 00:06:59.939 Samuel Roberts: But I…
74 00:06:59.940 ⇒ 00:07:01.129 Mustafa Raja: Yeah, this is…
75 00:07:02.510 ⇒ 00:07:04.709 Samuel Roberts: I’d like to have it where we have, like, at least
76 00:07:06.090 ⇒ 00:07:09.839 Samuel Roberts: Some good, you know, happy paths that…
77 00:07:10.210 ⇒ 00:07:14.990 Samuel Roberts: Are, like, we can confirm against, and, like, some things we know it can’t do, and confirm against that.
78 00:07:14.990 ⇒ 00:07:19.710 Mustafa Raja: what we… What we could do is, we have this sheet.
79 00:07:19.840 ⇒ 00:07:22.969 Mustafa Raja: Run all of these questions from both of those.
80 00:07:24.230 ⇒ 00:07:25.990 Mustafa Raja: And see the output.
81 00:07:26.760 ⇒ 00:07:31.869 Samuel Roberts: That’s a start, yeah, we definitely want that, and we probably want to compile, maybe go through some of the…
82 00:07:32.250 ⇒ 00:07:34.410 Samuel Roberts: Thumbs-up ones, too.
83 00:07:35.140 ⇒ 00:07:38.119 Mustafa Raja: Yeah, I could do that. I could do that. I could…
84 00:07:38.510 ⇒ 00:07:43.799 Mustafa Raja: Yeah, we have those in… in what’s it called Snowflake. I could… I could put that.
85 00:07:43.800 ⇒ 00:07:52.429 Samuel Roberts: So if we could get a list of, like, ones that it did good, ones that it did bad, and then we can run that against the two different Moscra flows.
86 00:07:55.650 ⇒ 00:07:56.769 Samuel Roberts: You know what I mean?
87 00:07:58.100 ⇒ 00:07:58.940 Mustafa Raja: Okay.
88 00:08:00.190 ⇒ 00:08:03.070 Mustafa Raja: Yeah, yeah, yeah, makes sense, makes sense.
89 00:08:03.070 ⇒ 00:08:04.490 Samuel Roberts: But then I’ll feel more comfortable being.
90 00:08:04.490 ⇒ 00:08:05.489 Mustafa Raja: could do that.
91 00:08:05.910 ⇒ 00:08:07.460 Samuel Roberts: Being like, okay, this one is…
92 00:08:07.940 ⇒ 00:08:14.709 Samuel Roberts: is the kind of on par with standard ANDI, and this one is the improved Andy, and we can see that is improved, and that’ll be really good.
93 00:08:17.270 ⇒ 00:08:19.840 Mustafa Raja: Yeah, yeah, I get, I get this.
94 00:08:20.770 ⇒ 00:08:21.540 Samuel Roberts: Okay.
95 00:08:25.300 ⇒ 00:08:31.800 Mustafa Raja: In the fixes, I haven’t mentioned the embeddings.
96 00:08:33.549 ⇒ 00:08:34.770 Mustafa Raja: Improve…
97 00:08:44.780 ⇒ 00:08:45.639 Mustafa Raja: In that time.
98 00:08:51.090 ⇒ 00:08:54.590 Mustafa Raja: I’m going to take this on Monday.
99 00:08:56.080 ⇒ 00:08:56.780 Mustafa Raja: Most of…
100 00:09:00.490 ⇒ 00:09:07.040 Mustafa Raja: That’s see how much… The central docs we would need to first remove.
101 00:09:10.240 ⇒ 00:09:11.429 Samuel Roberts: We would need to do what?
102 00:09:12.370 ⇒ 00:09:16.009 Mustafa Raja: We first need to review the improvements we need to make, right?
103 00:09:16.450 ⇒ 00:09:17.730 Samuel Roberts: Yes, yes.
104 00:09:18.830 ⇒ 00:09:20.330 Mustafa Raja: This makes more sense.
105 00:09:21.790 ⇒ 00:09:22.350 Samuel Roberts: Yeah.
106 00:09:32.570 ⇒ 00:09:33.230 Casie Aviles: Hey, guys.
107 00:09:33.230 ⇒ 00:09:35.030 Samuel Roberts: Hey, Casey, how’s it going?
108 00:09:35.840 ⇒ 00:09:37.409 Casie Aviles: Yeah, doing good.
109 00:09:39.080 ⇒ 00:09:43.400 Samuel Roberts: We were just going through the Gantt chart a little bit and talking about, like, migrating.
110 00:09:44.070 ⇒ 00:09:44.940 Casie Aviles: Yeah.
111 00:09:48.000 ⇒ 00:09:51.569 Samuel Roberts: But, yeah, any other updates on your end?
112 00:09:53.110 ⇒ 00:09:57.169 Casie Aviles: Yeah, we were also just talking about this earlier. Okay.
113 00:09:57.860 ⇒ 00:10:02.090 Casie Aviles: But yeah, other than that, for yesterday, yeah, I think…
114 00:10:02.640 ⇒ 00:10:09.690 Casie Aviles: what we did was we worked on a couple of the P0 fixes, and it was mostly, you know, just prompt.
115 00:10:10.180 ⇒ 00:10:20.739 Casie Aviles: Prompt fixes that we did, and then updating, like, outdated records in the database, and one of the things, one of the problems we surfaced there was
116 00:10:21.090 ⇒ 00:10:27.640 Casie Aviles: We didn’t have a… We didn’t sync, you know, with the spreadsheet that they have.
117 00:10:28.410 ⇒ 00:10:29.120 Samuel Roberts: Okay.
118 00:10:29.830 ⇒ 00:10:34.520 Casie Aviles: So that’s something we… we… we need to kind of think about.
119 00:10:35.170 ⇒ 00:10:37.320 Casie Aviles: Like, determining the source of truth.
120 00:10:37.750 ⇒ 00:10:38.580 Samuel Roberts: Yeah.
121 00:10:39.250 ⇒ 00:10:45.639 Samuel Roberts: I think I, yeah, I was just talking about that, like, new forms and stuff, I think Mustafa and I were…
122 00:10:45.860 ⇒ 00:10:52.780 Samuel Roberts: We’re just discussing, like… And it ended up being kind of limited, and now that we’re hosting the…
123 00:10:52.890 ⇒ 00:11:01.370 Samuel Roberts: DB query tool on Heroku, We could potentially put together some more custom form.
124 00:11:01.370 ⇒ 00:11:02.150 Casie Aviles: Yeah.
125 00:11:02.550 ⇒ 00:11:09.099 Samuel Roberts: that could actually pull from Superbase right now, at least. And then, like, that would give us…
126 00:11:10.700 ⇒ 00:11:15.690 Samuel Roberts: You know, populating the names, populating the service areas.
127 00:11:16.420 ⇒ 00:11:22.660 Casie Aviles: Yeah. Potentially even displaying, you know, the current changes and what’s happening in a little bit better way, so…
128 00:11:22.660 ⇒ 00:11:27.149 Samuel Roberts: make sure that those are going through and everything. So I think that’s worth investigating, too.
129 00:11:32.470 ⇒ 00:11:38.330 Casie Aviles: Yeah, the current forms don’t allow for that, so I think that’s also why.
130 00:11:39.110 ⇒ 00:11:45.360 Casie Aviles: You know, there’s, like, there’s, like, some friction to being able to use that.
131 00:11:45.790 ⇒ 00:11:46.280 Samuel Roberts: Definitely.
132 00:11:46.280 ⇒ 00:11:46.840 Casie Aviles: Okay.
133 00:11:48.770 ⇒ 00:11:51.200 Mustafa Raja: Now, let me ask Sam if you want to update this.
134 00:11:53.940 ⇒ 00:11:56.659 Mustafa Raja: Migration flow then.
135 00:11:59.450 ⇒ 00:12:01.119 Samuel Roberts: The migration flow?
136 00:12:03.730 ⇒ 00:12:06.040 Mustafa Raja: Yeah, the migration Gantt chart.
137 00:12:07.160 ⇒ 00:12:09.459 Samuel Roberts: Oh, the game, okay, yeah, yeah.
138 00:12:10.760 ⇒ 00:12:17.209 Samuel Roberts: Yeah, I think we need to add the… what we just talked about with the… the thumbs-up tests and the thumbs-down tests.
139 00:12:21.000 ⇒ 00:12:26.160 Samuel Roberts: And then we can, test the different architectures for the agents and the workflow.
140 00:12:30.360 ⇒ 00:12:33.239 Uttam Kumaran: Hey, Tom. Hey, Gabe. I didn’t see you guys there yet. Hey, guys.
141 00:12:33.240 ⇒ 00:12:34.150 Gabriel Lam: Morning.
142 00:12:36.550 ⇒ 00:12:37.610 Samuel Roberts: How’s it going?
143 00:12:37.610 ⇒ 00:12:38.360 Uttam Kumaran: Good.
144 00:12:40.190 ⇒ 00:12:41.310 Samuel Roberts: Okay.
145 00:12:42.110 ⇒ 00:12:47.889 Samuel Roberts: Good. Good, making progress on the MCP stuff, it’s… It’s looking better.
146 00:12:49.410 ⇒ 00:12:55.299 Samuel Roberts: it’s, roadblocks, but I was testing the different models, and they’re wildly different in terms of tool calls, which was interesting.
147 00:12:55.300 ⇒ 00:12:56.460 Uttam Kumaran: Okay, cool.
148 00:12:57.480 ⇒ 00:13:00.169 Samuel Roberts: And, yeah, speed. But yeah, we’ll talk about that in a moment.
149 00:13:05.820 ⇒ 00:13:10.809 Samuel Roberts: Alright, any other, ABC updates from yesterday, or does that cover everything?
150 00:13:11.040 ⇒ 00:13:21.990 Uttam Kumaran: Yeah, I guess I just wanted to make sure you guys feel good about this, and then, really, the feedback I gave is, like, if somehow we can map the tasks
151 00:13:22.220 ⇒ 00:13:25.819 Uttam Kumaran: Are the sections to, like, what this unlocks for…
152 00:13:25.890 ⇒ 00:13:28.919 Samuel Roberts: The end user, basically, is sort of like…
153 00:13:29.100 ⇒ 00:13:32.480 Uttam Kumaran: what I mentioned to Mustafa this morning is, like.
154 00:13:32.970 ⇒ 00:13:38.129 Uttam Kumaran: Okay, if we move… if we move out of… if we move something from NAN forms to…
155 00:13:38.300 ⇒ 00:13:40.140 Uttam Kumaran: Something else, then…
156 00:13:40.140 ⇒ 00:13:40.770 Samuel Roberts: Yeah.
157 00:13:40.770 ⇒ 00:13:45.579 Uttam Kumaran: it unlocks the ability for ABC to directly update, like, services, right? So I guess, like.
158 00:13:46.080 ⇒ 00:13:52.950 Uttam Kumaran: that’s sort of, like, how it would just help me be able to articulate, like, when things are gonna happen. But yeah.
159 00:13:53.900 ⇒ 00:13:54.570 Samuel Roberts: Okay.
160 00:13:56.530 ⇒ 00:13:57.320 Samuel Roberts: Okay.
161 00:14:00.730 ⇒ 00:14:06.229 Samuel Roberts: Yeah, I mean, Casey Mustafa, well, Mustafa was just talking about looking at other options for that, that…
162 00:14:06.630 ⇒ 00:14:09.759 Samuel Roberts: For the forum specifically, I don’t know if you’ve done any of that yet, but…
163 00:14:14.180 ⇒ 00:14:16.320 Samuel Roberts: Keeping that in mind with the,
164 00:14:18.300 ⇒ 00:14:21.570 Samuel Roberts: The current hosting is… is good, okay.
165 00:14:22.950 ⇒ 00:14:24.720 Casie Aviles: He blocked off some.
166 00:14:28.500 ⇒ 00:14:29.100 Samuel Roberts: Okay.
167 00:14:31.170 ⇒ 00:14:31.800 Casie Aviles: Yeah, double.
168 00:14:31.800 ⇒ 00:14:32.410 Samuel Roberts: Excuse me.
169 00:14:34.150 ⇒ 00:14:41.459 Samuel Roberts: Hang. And so then, what’s the… are there other bonkers or plans for today? Like, what is,
170 00:14:41.780 ⇒ 00:14:44.069 Samuel Roberts: The plan today, then if you blocked off some time for that.
171 00:14:49.530 ⇒ 00:14:50.110 Mustafa Raja: You ain’t going.
172 00:14:50.110 ⇒ 00:14:51.019 Casie Aviles: Go ahead, go ahead.
173 00:14:52.550 ⇒ 00:14:55.699 Casie Aviles: No, no, I was just saying, like, we were going to add, like.
174 00:14:56.170 ⇒ 00:15:07.180 Casie Aviles: some of the missing technicians from other departments, because they just sent us over those spreadsheets, so that’s our next step, which is to add those.
175 00:15:08.120 ⇒ 00:15:08.720 Samuel Roberts: Great.
176 00:15:09.210 ⇒ 00:15:09.900 Mustafa Raja: Hmm.
177 00:15:10.130 ⇒ 00:15:15.429 Mustafa Raja: And for me, adding those, and then, I’ll also be…
178 00:15:15.550 ⇒ 00:15:32.080 Mustafa Raja: pulling the data from Snowflake, with thumbs up and thumbs down, maybe 50 records each, and then I’ll run… I’ll run what we have in N10 and what we have in Mastra, the two architectures that we have in Mastra, and then what we have in N10.
179 00:15:33.350 ⇒ 00:15:45.250 Samuel Roberts: I mean, what we have in N8… yeah, I would not worry about rerunning the N8N, I would just focus on the mastering, make sure that the outputs are still good, thumbs up-wise, and then if it’s improved, then the thumbs down ones. Obviously, use the ones we already did.
180 00:15:45.250 ⇒ 00:15:45.640 Mustafa Raja: Okay.
181 00:15:45.640 ⇒ 00:15:46.990 Samuel Roberts: Documented and everything.
182 00:15:47.100 ⇒ 00:15:53.660 Samuel Roberts: But then, once that’s out… once we get that together, we can kind of see which one of the two is good, and where to make changes, and…
183 00:15:53.970 ⇒ 00:15:56.300 Samuel Roberts: That’ll help us with that a lot.
184 00:15:57.980 ⇒ 00:15:59.300 Mustafa Raja: Okay, cool.
185 00:15:59.300 ⇒ 00:16:00.780 Samuel Roberts: Any other ABC?
186 00:16:03.600 ⇒ 00:16:04.800 Mustafa Raja: How do I remove this?
187 00:16:07.690 ⇒ 00:16:09.629 Samuel Roberts: Where are you trying to move it? Up?
188 00:16:09.850 ⇒ 00:16:10.400 Samuel Roberts: Down?
189 00:16:10.400 ⇒ 00:16:11.579 Mustafa Raja: Yeah, this dependency.
190 00:16:12.290 ⇒ 00:16:13.010 Samuel Roberts: Oh.
191 00:16:14.310 ⇒ 00:16:16.880 Samuel Roberts: Been having issues with Instagane as well, so I’m not sure.
192 00:16:17.350 ⇒ 00:16:18.610 Mustafa Raja: Hmm…
193 00:16:18.610 ⇒ 00:16:19.960 Samuel Roberts: And… yeah.
194 00:16:21.170 ⇒ 00:16:22.300 Samuel Roberts: Okay.
195 00:16:22.460 ⇒ 00:16:28.889 Samuel Roberts: If that’s… Good on ABC then for today. I guess I’ll give an update on…
196 00:16:29.450 ⇒ 00:16:34.330 Samuel Roberts: Lilo, so, yeah, yesterday and this morning,
197 00:16:34.520 ⇒ 00:16:46.409 Samuel Roberts: I got everything running that, Pranav had started. I actually had a call with him real quick after I got all the environment variable stuff set up, so it probably would have been better if it had been a little earlier, but he was… he was busy. But,
198 00:16:46.690 ⇒ 00:16:52.199 Samuel Roberts: Got that all set up, running locally, added stuff for the,
199 00:16:52.840 ⇒ 00:16:58.650 Samuel Roberts: MCP, and specifically, like, setting the environment variables and the keys and everything, so,
200 00:16:58.990 ⇒ 00:17:04.349 Samuel Roberts: I think Surface was literally thinking that we were just hard-coding things so that it would be part of the environment variables, but for each
201 00:17:04.980 ⇒ 00:17:08.159 Samuel Roberts: each brand is all different, and so I’ve got a little bit of a…
202 00:17:08.280 ⇒ 00:17:14.199 Samuel Roberts: UI scaffolded up for that, that right now works with Klaviyo, at least, and so that’s how I’m able to actually test it.
203 00:17:14.200 ⇒ 00:17:17.190 Uttam Kumaran: I don’t know, Sam, if you want to share that, because then.
204 00:17:17.190 ⇒ 00:17:17.660 Samuel Roberts: Yeah.
205 00:17:17.660 ⇒ 00:17:20.160 Uttam Kumaran: Just… maybe we can even walk through what
206 00:17:20.510 ⇒ 00:17:23.319 Uttam Kumaran: We can either walk through the slides together.
207 00:17:24.290 ⇒ 00:17:33.980 Uttam Kumaran: Or walk through the demo, I just want to kind of… maybe we do a dry run, and then I can spend… I basically am free the next, like, hour, so I can work or whatever.
208 00:17:34.230 ⇒ 00:17:43.990 Samuel Roberts: Okay, yeah, let me get over to, my other… green here… Excuse me, I want…
209 00:17:44.620 ⇒ 00:17:45.879 Samuel Roberts: I’ll just share on it.
210 00:17:46.300 ⇒ 00:17:52.849 Samuel Roberts: Alright, so we are… Looking at the,
211 00:17:53.030 ⇒ 00:18:05.960 Samuel Roberts: I’m on my laptop, not with my monitor, so everything. Okay. Here we go. Okay, so basically, yeah, here’s the current thing. I have just a few test stores that I was playing around with. Obviously, nothing is tied to anything, but the,
212 00:18:05.990 ⇒ 00:18:18.369 Samuel Roberts: but the Klaviyo and the Opus keys, or the Claude keys that they gave us. So, you know, you can see there’s a ton of testing here where it was not working, and finally it worked and started spitting out good stuff.
213 00:18:18.570 ⇒ 00:18:26.580 Samuel Roberts: I can ask more, specific questions. I think I did that over… Here…
214 00:18:27.090 ⇒ 00:18:30.499 Samuel Roberts: Yeah, so this was actually really interesting for the models, because I asked about
215 00:18:30.610 ⇒ 00:18:43.590 Samuel Roberts: some, like, specific things, and it’s like, great, yeah, I can’t find that. And then I upgraded the model, and it was like, yeah, great, here it is. So that’s a nice little thing to show off. But, the other thing we can do is just ask one of those questions that they had.
216 00:18:44.120 ⇒ 00:18:52.549 Samuel Roberts: here… I haven’t tested all these yet, so this is definitely a,
217 00:18:58.070 ⇒ 00:19:01.310 Samuel Roberts: And then right now, this is what I was… I thought was interesting. I… it just…
218 00:19:01.430 ⇒ 00:19:03.180 Samuel Roberts: takes a while, so I thought…
219 00:19:03.580 ⇒ 00:19:05.819 Samuel Roberts: That’s why I was asking about the different models, but…
220 00:19:05.960 ⇒ 00:19:10.140 Samuel Roberts: I want to also have it, you know, if you come over to…
221 00:19:11.020 ⇒ 00:19:20.499 Samuel Roberts: cursor logs here, you can see what’s going on. So, like, it’s making the tool call, it’s doing its thinking. Right now, it’s not logging much of that, or not displaying much of that.
222 00:19:20.990 ⇒ 00:19:23.879 Samuel Roberts: I want to kind of get their feel for how much they liked
223 00:19:25.040 ⇒ 00:19:29.080 Samuel Roberts: like, they said they didn’t like the formatting, but I didn’t know if they appreciated seeing the…
224 00:19:30.180 ⇒ 00:19:34.209 Samuel Roberts: The tool calls, and what’s happening, because it… sometimes you just sit there waiting, and…
225 00:19:34.320 ⇒ 00:19:36.480 Samuel Roberts: There’s stuff happening in the background, but…
226 00:19:37.170 ⇒ 00:19:45.680 Uttam Kumaran: Okay, so let me just take… let me just take a couple notes, because, as you know, the risk with demos is that it just gets derailed, especially with these guys, they just are…
227 00:19:45.940 ⇒ 00:19:46.549 Uttam Kumaran: We’ll be…
228 00:19:46.550 ⇒ 00:19:50.760 Samuel Roberts: True, true. Let’s probably a loom of this was better or not, for them.
229 00:19:50.760 ⇒ 00:20:02.550 Uttam Kumaran: Well, like, I don’t want to do a loom, because I think we’re… I mean, we’re… yeah, I mean, they’re gonna… they care a lot more about output, but I do want a demo where we’re asking… we’re gonna demo
230 00:20:02.720 ⇒ 00:20:07.569 Uttam Kumaran: things in, like, a predictive order, like, I don’t want to do just, like.
231 00:20:07.680 ⇒ 00:20:15.669 Uttam Kumaran: Click around and just, like, do whatever we feel like, so… let me just, like, put together… not that you’re doing that, but I want to give you some…
232 00:20:15.670 ⇒ 00:20:19.269 Samuel Roberts: Oh, no, it’s good. I have a few… I have a few ideas here from what I’ve been testing.
233 00:20:19.270 ⇒ 00:20:25.160 Uttam Kumaran: I want to give you a little bit of a safety net, so we can go, like, this thing, this thing, this thing, this thing, and just get that structured feedback.
234 00:20:25.580 ⇒ 00:20:28.060 Samuel Roberts: Yeah, what we could even do is walk through the, like.
235 00:20:28.320 ⇒ 00:20:30.960 Samuel Roberts: the brand creation and stuff, like… Okay.
236 00:20:31.560 ⇒ 00:20:33.790 Samuel Roberts: You know, if you come here…
237 00:20:34.400 ⇒ 00:20:37.359 Samuel Roberts: ad brand, test 4, I can show…
238 00:20:39.110 ⇒ 00:20:43.069 Samuel Roberts: right now, there’s only one chat, so I’m hoping they’re not too worried about that, but…
239 00:20:43.070 ⇒ 00:20:45.579 Uttam Kumaran: Wait, wait, so can we start from the beginning?
240 00:20:45.740 ⇒ 00:20:48.289 Samuel Roberts: Yeah, so there’s the main dashboard right now.
241 00:20:48.290 ⇒ 00:20:48.900 Uttam Kumaran: Okay.
242 00:20:49.430 ⇒ 00:20:51.329 Samuel Roberts: I can just add a brand.
243 00:20:51.710 ⇒ 00:20:52.340 Uttam Kumaran: bed.
244 00:20:52.690 ⇒ 00:20:58.909 Samuel Roberts: Just calling test. So then when you come in here, this is just set up to talk to…
245 00:20:59.330 ⇒ 00:21:05.249 Samuel Roberts: Claude Opus right now, and obviously, these are all not really functional except for Klaviyo.
246 00:21:05.420 ⇒ 00:21:09.250 Samuel Roberts: what I would do is come in here, Oops.
247 00:21:10.970 ⇒ 00:21:17.140 Samuel Roberts: Come here… set up the API key, Which is…
248 00:21:17.340 ⇒ 00:21:22.840 Samuel Roberts: I’ll probably have this in my… Clipboard, anyway.
249 00:21:23.410 ⇒ 00:21:25.989 Samuel Roberts: Paste that there. Now Klaviyo should be set up.
250 00:21:26.340 ⇒ 00:21:27.000 Uttam Kumaran: Okay.
251 00:21:27.380 ⇒ 00:21:32.029 Samuel Roberts: And then if we want to ask, what were those questions? That’s why.
252 00:21:34.090 ⇒ 00:21:37.619 Samuel Roberts: This is the one I had tested, so I think this one should be good.
253 00:21:43.710 ⇒ 00:21:51.040 Samuel Roberts: And then hopefully this actually returns well. If not, then that changes what we actually want to test, but this is what I was thinking we could show, that it’s… it’s actually…
254 00:21:54.190 ⇒ 00:21:58.760 Uttam Kumaran: So, I guess, like, what I want to do is just… just actually just save…
255 00:21:59.040 ⇒ 00:22:02.150 Uttam Kumaran: The questions, so you just copy-paste them.
256 00:22:02.360 ⇒ 00:22:05.179 Samuel Roberts: That’s what I was saying, yeah, they’ll be ready to go.
257 00:22:05.640 ⇒ 00:22:06.220 Uttam Kumaran: Cool.
258 00:22:06.570 ⇒ 00:22:09.099 Uttam Kumaran: So we’re gonna do the first chat, which is, like.
259 00:22:09.520 ⇒ 00:22:17.619 Uttam Kumaran: Tell me about recipients. And then in this thing, we want to display, like, a basically, like, rich response.
260 00:22:17.850 ⇒ 00:22:21.849 Uttam Kumaran: responses, right? So, we want to show that it does tables.
261 00:22:22.580 ⇒ 00:22:25.989 Uttam Kumaran: The bold, like, we wanted to show them as headers.
262 00:22:26.170 ⇒ 00:22:33.020 Samuel Roberts: Yeah, it’s all… it’s all just the markdown coming out, but it’s formatted, at least. Okay. Initially, it was… if we jump… oh, no, I can’t show it anymore.
263 00:22:33.020 ⇒ 00:22:39.839 Uttam Kumaran: The thing is, like, they’re not gonna… they’re… they’re not gonna know… I wanna almost be like, it’s doing headers, it’s doing tables, they’re.
264 00:22:39.840 ⇒ 00:22:42.210 Samuel Roberts: Oh, I see, yeah, you want to get… yeah, okay.
265 00:22:42.400 ⇒ 00:22:43.869 Uttam Kumaran: For us, yeah, I know it’s just a.
266 00:22:43.870 ⇒ 00:22:48.619 Samuel Roberts: No, I get what you’re saying, yeah, that’s not… yeah, but yeah, it’s formatting, it’s tabling, it’s…
267 00:22:49.020 ⇒ 00:22:50.369 Samuel Roberts: Yeah, I see what you’re saying.
268 00:22:52.690 ⇒ 00:22:53.460 Uttam Kumaran: Okay.
269 00:22:57.670 ⇒ 00:22:59.600 Samuel Roberts: But, yeah, I mean, I’m not sure what else to…
270 00:23:00.840 ⇒ 00:23:08.970 Uttam Kumaran: Can we show, like, out of all the questions that they sent, like, what are all, possible, or what are all showing.
271 00:23:09.270 ⇒ 00:23:17.050 Samuel Roberts: Yeah, I hadn’t gotten through all of them yet, because I had some… I had some other roadblocks when I was testing them, because of the models and stuff, but… Let’s see…
272 00:23:17.050 ⇒ 00:23:21.900 Uttam Kumaran: We’ll see how many of these work, and then let’s just, like, we’ll…
273 00:23:22.290 ⇒ 00:23:27.730 Uttam Kumaran: I basically kind of want to go through their list and be like, okay, we went through, this is your first question, this is your second question.
274 00:23:36.850 ⇒ 00:23:43.189 Samuel Roberts: Yeah, I just reworked the way this is streaming, because it wasn’t streaming before, because of the way the tool calls were.
275 00:23:43.820 ⇒ 00:23:51.860 Samuel Roberts: But it’s now… so, like, it’s… this is actually better than it was an hour ago, where it was just saying a system is typing, and…
276 00:23:53.150 ⇒ 00:23:57.130 Samuel Roberts: And just… dumped it.
277 00:23:57.660 ⇒ 00:23:58.170 Uttam Kumaran: Okay.
278 00:23:58.170 ⇒ 00:24:00.880 Samuel Roberts: I think this… Probably some separate, like…
279 00:24:01.660 ⇒ 00:24:03.890 Samuel Roberts: But that’s, you know, I wanted to…
280 00:24:04.500 ⇒ 00:24:06.599 Samuel Roberts: It’s… you know… oh, hold on.
281 00:24:06.740 ⇒ 00:24:10.880 Samuel Roberts: In the back here, it’s making these tool calls, it’s getting the…
282 00:24:12.080 ⇒ 00:24:19.140 Samuel Roberts: flow reports and in Slack, that I… helpful, as long as we do it in a way that’s not gonna…
283 00:24:19.440 ⇒ 00:24:23.639 Samuel Roberts: Look, crazy. Did it finish? It finished, so an error prompted you to answer something.
284 00:24:25.000 ⇒ 00:24:30.549 Uttam Kumaran: Okay, okay. I’m gonna ask them what they think about wanting to see the tools or things on the front end, so…
285 00:24:30.550 ⇒ 00:24:36.530 Samuel Roberts: Yeah, I figured we could talk through that, but… Okay, so this is the second time we hit the prompt too long, and this is not a crazy one.
286 00:24:40.190 ⇒ 00:24:44.290 Uttam Kumaran: I mean, she’s probably going through, like, the events, or it’s just a lot of events, right?
287 00:24:44.290 ⇒ 00:24:54.620 Samuel Roberts: Yeah, so I’m thinking maybe we gotta start… this is the first time I’m hitting this in the testing, so what I probably need to do is, you know, catch this maximum and make sure the prompt is, like, truncated from the beginning or something.
288 00:24:54.850 ⇒ 00:24:56.660 Samuel Roberts: For now.
289 00:24:58.200 ⇒ 00:25:02.060 Samuel Roberts: This is the issue, this is the issue with some of the MCP stuff, where you don’t know what’s coming back, but…
290 00:25:02.550 ⇒ 00:25:03.070 Uttam Kumaran: Okay.
291 00:25:03.610 ⇒ 00:25:06.359 Samuel Roberts: But yeah, so like…
292 00:25:06.820 ⇒ 00:25:11.529 Samuel Roberts: I wonder if I just make a new… I can’t make a new chat right now, but I can go somewhere else, but…
293 00:25:11.870 ⇒ 00:25:19.129 Samuel Roberts: Yeah, I don’t know what else to keep demoing. Like, it’s still in, you know, work-in-progress mode. I just wanted to show them that it’s at least connected, but…
294 00:25:22.090 ⇒ 00:25:23.140 Samuel Roberts: Excuse me.
295 00:25:23.700 ⇒ 00:25:28.580 Samuel Roberts: Yeah. I mean, do you want me to keep going through? I don’t know what’s gonna happen.
296 00:25:28.580 ⇒ 00:25:35.059 Uttam Kumaran: Yeah, let’s just go through those questions. I want to see how much we can display today, and yeah, that… Yeah.
297 00:25:35.350 ⇒ 00:25:40.460 Samuel Roberts: The issue is gonna be, I think we’re sending those previous messages still.
298 00:25:40.780 ⇒ 00:25:45.920 Samuel Roberts: So, like, right now we’re gonna get more, more prompt…
299 00:25:47.330 ⇒ 00:25:51.770 Samuel Roberts: Well, let’s try it, let’s just see what happens. I don’t think it’s gonna work…
300 00:25:56.970 ⇒ 00:25:57.960 Samuel Roberts: Oops.
301 00:25:59.960 ⇒ 00:26:01.659 Samuel Roberts: So that’s working, that’s good.
302 00:26:24.410 ⇒ 00:26:32.060 Samuel Roberts: Yeah, this is… We get an error somewhere? No, that’s good. MCP server is running.
303 00:26:33.130 ⇒ 00:26:36.850 Samuel Roberts: It’s not… doing anything. Yeah, I don’t know.
304 00:26:37.700 ⇒ 00:26:40.280 Samuel Roberts: This is the kind of stuff I hadn’t really gotten to yet, so I don’t know.
305 00:26:40.560 ⇒ 00:26:44.800 Uttam Kumaran: No, all good. No, dude, no pressure. I just literally want to see what’s working.
306 00:26:44.800 ⇒ 00:26:46.160 Samuel Roberts: Yeah, well, no, this is good.
307 00:26:46.160 ⇒ 00:26:51.780 Uttam Kumaran: Because in the event they ask, hey, what about the other questions? We need to have already done them, you know?
308 00:26:51.780 ⇒ 00:26:52.619 Samuel Roberts: Oh, you’re right, you’re right.
309 00:26:52.620 ⇒ 00:26:56.880 Uttam Kumaran: So, I don’t want to get… demos are not for surprises, so…
310 00:26:56.880 ⇒ 00:27:02.499 Samuel Roberts: No, absolutely not. There we go, yeah. The other thing is the, you know, the model, we only have so much control over, though.
311 00:27:02.770 ⇒ 00:27:08.289 Samuel Roberts: But testing the faster, cheaper models aren’t good at tool calls, it seems, so…
312 00:27:11.770 ⇒ 00:27:12.490 Uttam Kumaran: Okay.
313 00:27:12.490 ⇒ 00:27:16.440 Samuel Roberts: We’ll see, we’re probably gonna hit the same… Issue, because it’s gonna get…
314 00:27:18.160 ⇒ 00:27:21.099 Samuel Roberts: Get a bunch of stuff in the prompts, but yeah.
315 00:27:23.090 ⇒ 00:27:24.240 Samuel Roberts: Excuse me.
316 00:27:32.490 ⇒ 00:27:34.609 Samuel Roberts: Yeah, these have been so slow, which is…
317 00:27:36.750 ⇒ 00:27:40.639 Uttam Kumaran: No, I mean, dude, the Klaviyo thing is, like, a lot.
318 00:27:40.640 ⇒ 00:27:50.729 Samuel Roberts: Yeah, I was trying to get it to, like, power, like, how, like, the times I was taking for the tool calls versus the model calls, and it got a little… I just kind of nixed it, and was just like, let’s just…
319 00:27:51.090 ⇒ 00:27:53.879 Samuel Roberts: Pump up the model and say… there we go, okay.
320 00:27:58.100 ⇒ 00:28:02.710 Samuel Roberts: But I can see why, like, MCP might not be the right, or the Clamp MCP specific, like, yeah, it…
321 00:28:03.070 ⇒ 00:28:08.279 Samuel Roberts: You know, it would take a lot more to, you know, warehouse it all and do that, but once that’s done, like, it could be…
322 00:28:08.710 ⇒ 00:28:09.989 Samuel Roberts: Much faster.
323 00:28:11.310 ⇒ 00:28:13.580 Samuel Roberts: But yeah, I mean, that one… okay, so that one seemed to work.
324 00:28:14.140 ⇒ 00:28:16.269 Samuel Roberts: That was worth sending time.
325 00:28:17.990 ⇒ 00:28:23.329 Samuel Roberts: Worst sending time and average based on Klaviyo data. Send time? Yeah. Okay.
326 00:28:29.580 ⇒ 00:28:30.570 Samuel Roberts: Alright.
327 00:28:31.320 ⇒ 00:28:32.869 Samuel Roberts: Add some more then, I guess?
328 00:28:36.160 ⇒ 00:28:38.070 Samuel Roberts: Let’s see the best time, yeah, alright, cool.
329 00:28:49.620 ⇒ 00:28:50.810 Samuel Roberts: 24…
330 00:28:56.260 ⇒ 00:28:58.620 Samuel Roberts: Yeah Yeah, I also hit some…
331 00:29:02.400 ⇒ 00:29:05.320 Samuel Roberts: Oh, nice, that one must have been cash, that’s sweet. Okay, cool.
332 00:29:07.530 ⇒ 00:29:11.310 Samuel Roberts: Yeah, based on the data I already retrieved, so that’s good. Alright.
333 00:29:12.170 ⇒ 00:29:16.530 Samuel Roberts: Yeah, alright, so some of these seem to be good. I mean, do you want me to keep going through all of them now, or do you want to wait and do that after?
334 00:29:16.890 ⇒ 00:29:17.660 Samuel Roberts: We can still do some.
335 00:29:17.660 ⇒ 00:29:26.350 Uttam Kumaran: Oh, no, we can just go… we can just go through them, and then maybe… I just want to make sure that we’re good on the demo side, and so… and then…
336 00:29:27.180 ⇒ 00:29:34.329 Uttam Kumaran: Like, to sort of tick the box there, and then we can now switch, basically, to, like, slides or anything else in the platform we want to share.
337 00:29:34.880 ⇒ 00:29:35.710 Samuel Roberts: Sure.
338 00:29:43.020 ⇒ 00:29:44.460 Samuel Roberts: Alright, so that one’s going.
339 00:29:44.460 ⇒ 00:29:46.609 Uttam Kumaran: And can we rip lane fuse for these guys?
340 00:29:48.010 ⇒ 00:29:48.810 Samuel Roberts: Can we what?
341 00:29:48.990 ⇒ 00:29:50.930 Uttam Kumaran: Can we do length views for these guys?
342 00:29:52.110 ⇒ 00:29:57.239 Samuel Roberts: Yeah, I’m a little nervous about landing keys, because we saw that one error the other day, where it…
343 00:29:57.350 ⇒ 00:29:58.320 Samuel Roberts: timed out?
344 00:29:59.310 ⇒ 00:30:04.760 Uttam Kumaran: But I guess, like, what… my question is, like, what is our default for prompt management?
345 00:30:06.210 ⇒ 00:30:10.920 Samuel Roberts: Well, for this, they were… they were gonna have a prompt, like, a prompt library here,
346 00:30:11.240 ⇒ 00:30:15.850 Samuel Roberts: So I didn’t know if we were just gonna have that built into this platform.
347 00:30:16.010 ⇒ 00:30:22.070 Uttam Kumaran: But you’re gonna… but dude, you’re not gonna do, like… I want… if I want to be able to compare, like, 4 different models, like…
348 00:30:22.770 ⇒ 00:30:26.719 Uttam Kumaran: And look at the response times, and try out prompts, like…
349 00:30:27.260 ⇒ 00:30:28.919 Uttam Kumaran: We need a tool for that.
350 00:30:30.780 ⇒ 00:30:34.880 Samuel Roberts: For, like, optimizing the prompts, or, like, for their prompt library that they’re gonna have for different prompts here?
351 00:30:35.230 ⇒ 00:30:38.139 Uttam Kumaran: Yeah, but the prompt library is different than, like.
352 00:30:38.490 ⇒ 00:30:39.170 Samuel Roberts: Yeah, I know.
353 00:30:39.170 ⇒ 00:30:51.129 Uttam Kumaran: It uses, like, anything, like… I mean, I get that it has that as an option, but, like, I, like, personally, as, like, a developer, I need to be able to be able to try prompts, see potential outputs.
354 00:30:51.290 ⇒ 00:30:54.170 Uttam Kumaran: Look at response times, like, things like that.
355 00:30:54.420 ⇒ 00:30:55.120 Uttam Kumaran: like…
356 00:30:55.590 ⇒ 00:30:56.230 Samuel Roberts: Yeah.
357 00:30:56.410 ⇒ 00:30:57.020 Uttam Kumaran: Yeah.
358 00:30:58.390 ⇒ 00:31:13.289 Uttam Kumaran: I just… again, I just want to be, like, when we go to a new client and develop an app like this, we always just go with this… we just think about, like, what our typical stack is, but for example, if, like, I need to go… we need to try, like, 5 different prompts and see changes, like, what are we going to use for that, you know?
359 00:31:13.630 ⇒ 00:31:17.089 Samuel Roberts: Right, yeah, no, I’d like to use a definite use for that. I thought you meant the, like.
360 00:31:17.280 ⇒ 00:31:20.169 Samuel Roberts: Because they had talked about, like.
361 00:31:20.170 ⇒ 00:31:25.720 Uttam Kumaran: I think we’ll give them some way of doing that, but honestly, they’re not gonna care. Like, if that’s in laying fuse, they’re not gonna care.
362 00:31:27.110 ⇒ 00:31:28.260 Samuel Roberts: Oh, I suppose that’s true.
363 00:31:28.590 ⇒ 00:31:29.420 Uttam Kumaran: Yeah, like.
364 00:31:29.420 ⇒ 00:31:34.620 Samuel Roberts: Yeah, but they’re still gonna want UI here, though, I think, for the different brands, but yeah, I hear what you’re saying.
365 00:31:34.620 ⇒ 00:31:36.060 Uttam Kumaran: Yeah, yeah, yeah.
366 00:31:36.340 ⇒ 00:31:39.520 Samuel Roberts: Okay, so this one definitely… this one hit the prompt limit, too.
367 00:31:39.520 ⇒ 00:31:40.820 Uttam Kumaran: Okay. Look like this.
368 00:31:40.830 ⇒ 00:31:44.610 Samuel Roberts: Few, yeah, big… get metrics, get campaign reports.
369 00:31:45.420 ⇒ 00:32:02.270 Uttam Kumaran: Okay, so one thing I think we should just share is that, like, look, there is… we’re just hitting context limits, so we have, like, one or two directions. We should either get it in, like, chunk, or we basically have to… we have to do aggregations in a warehouse, which is probably the direction we’ll go here.
370 00:32:02.270 ⇒ 00:32:03.540 Samuel Roberts: Yeah, I think so.
371 00:32:03.540 ⇒ 00:32:08.839 Uttam Kumaran: Yep, okay, cool. That’s all I want to share. So I don’t mind, us explaining that.
372 00:32:08.940 ⇒ 00:32:09.810 Samuel Roberts: Okay.
373 00:32:09.810 ⇒ 00:32:14.769 Uttam Kumaran: So let’s just focus on… so, in terms of my list, I have,
374 00:32:17.500 ⇒ 00:32:27.520 Uttam Kumaran: Open the main dashboard, add a brand, open chat, we’re using Cloud Opus, open integrations, add Klaviyo key, send a chat, select… in that thing, you select the…
375 00:32:27.660 ⇒ 00:32:42.830 Uttam Kumaran: you select the integration, you run your tell me about recipients question, we share the responses, we also talk about, like, what happens when you’re waiting for a response, and then my question to them is, like, do we want to see tool calls?
376 00:32:43.430 ⇒ 00:32:46.490 Uttam Kumaran: Open to Data Warehouse…
377 00:32:46.790 ⇒ 00:32:47.130 Samuel Roberts: Yeah.
378 00:32:47.130 ⇒ 00:32:51.039 Uttam Kumaran: Yeah, I mean, that’s… those are the…
379 00:32:51.390 ⇒ 00:32:52.100 Samuel Roberts: Yeah.
380 00:32:52.100 ⇒ 00:32:53.290 Uttam Kumaran: Probably the biggest star.
381 00:32:58.300 ⇒ 00:32:59.400 Uttam Kumaran: Okay, cool.
382 00:33:01.970 ⇒ 00:33:02.600 Samuel Roberts: Alright.
383 00:33:02.740 ⇒ 00:33:05.449 Samuel Roberts: So good on that, then.
384 00:33:05.580 ⇒ 00:33:10.380 Samuel Roberts: Anything else, then, to discuss now before we dive in after this?
385 00:33:11.040 ⇒ 00:33:12.400 Uttam Kumaran: No, I think that’s it.
386 00:33:12.700 ⇒ 00:33:13.520 Samuel Roberts: Okay, cool.
387 00:33:14.600 ⇒ 00:33:17.200 Samuel Roberts: I guess then, yeah, Gabe?
388 00:33:17.400 ⇒ 00:33:18.490 Samuel Roberts: All yours.
389 00:33:18.490 ⇒ 00:33:25.759 Gabriel Lam: Yeah, so, from yesterday, Utam, I don’t want to ask if there’s been any updates with Ian.
390 00:33:25.920 ⇒ 00:33:28.630 Gabriel Lam: And getting our insurance.
391 00:33:29.040 ⇒ 00:33:30.520 Gabriel Lam: forms, data.
392 00:33:30.720 ⇒ 00:33:32.650 Uttam Kumaran: Oh, no. They’ll block on that.
393 00:33:32.840 ⇒ 00:33:35.900 Uttam Kumaran: Yeah, I don’t have… I mean, yeah, I just… maybe just Slack.
394 00:33:36.330 ⇒ 00:33:37.180 Gabriel Lam: Rico.
395 00:33:37.470 ⇒ 00:33:42.110 Uttam Kumaran: Yeah, just Slack me and Rico, some channel, yeah, and then… Yeah.
396 00:33:42.590 ⇒ 00:33:46.630 Uttam Kumaran: I mean, if you get… you can also Slack Rico, and for, like, a few…
397 00:33:46.850 ⇒ 00:33:49.200 Uttam Kumaran: basically, I would email Ian, be like.
398 00:33:49.410 ⇒ 00:33:58.590 Uttam Kumaran: Can we send you an NDA so we can maybe use some of your live data? Like, what would you need? Yeah. And Ricoh can help you with any documents you need. I would just drive it with Ian.
399 00:33:58.590 ⇒ 00:33:59.350 Gabriel Lam: Okay.
400 00:33:59.350 ⇒ 00:34:03.229 Uttam Kumaran: So where I think Ian, like, I don’t mind him using this for his company.
401 00:34:03.360 ⇒ 00:34:07.279 Uttam Kumaran: As long as, like, we’re… he’s, like, more of a design partner with us, you know?
402 00:34:07.280 ⇒ 00:34:07.820 Gabriel Lam: Yep.
403 00:34:07.820 ⇒ 00:34:08.380 Uttam Kumaran: Yeah.
404 00:34:08.790 ⇒ 00:34:19.719 Gabriel Lam: And then the other thing I was working on yesterday was kind of doing an audit on our whole GitHub stack and checking to see which repos we need and don’t. For example, Brainforge files.
405 00:34:20.050 ⇒ 00:34:22.710 Gabriel Lam: Contains, like, our case studies and marketing.
406 00:34:22.710 ⇒ 00:34:30.410 Uttam Kumaran: Yeah, you’re gonna find it funny, like, what I’m doing right now. I’m, like, writing a sales email. It turned out great, by the way, but I have…
407 00:34:30.620 ⇒ 00:34:33.519 Uttam Kumaran: I have it, so let me show you, like, kind of, like, what I did.
408 00:34:33.790 ⇒ 00:34:45.099 Uttam Kumaran: I sent it, like, a referral that we got, and I said, help me respond, consider the files and Brainforge files. It went and looked through it all, built this plan for me.
409 00:34:45.320 ⇒ 00:34:46.840 Uttam Kumaran: And then…
410 00:34:47.239 ⇒ 00:34:59.359 Uttam Kumaran: this was actually so killer. The email was good, and then it said, from what I gathered, the university’s been leading into these things. And then I looked at the email that I got, and none of those things were mentioned, and I was like.
411 00:34:59.630 ⇒ 00:35:06.789 Uttam Kumaran: I was like, did you find anything on the internet about BayPath? And he was like, yeah, this was, like, I found this about… on the internet.
412 00:35:07.020 ⇒ 00:35:09.299 Uttam Kumaran: And I guess it may have, like, looked…
413 00:35:10.730 ⇒ 00:35:13.820 Uttam Kumaran: Yeah, like, it searched the web. Oh, yeah. Oh, wow.
414 00:35:14.010 ⇒ 00:35:19.960 Uttam Kumaran: BayPath AI University initiatives, and then I was like, it’s like, yeah, I found these, and I’m like.
415 00:35:20.300 ⇒ 00:35:25.980 Uttam Kumaran: I was like, Dude, this, like, really helps our credibility, like, let’s link these in, and so…
416 00:35:26.150 ⇒ 00:35:36.049 Uttam Kumaran: it linked, and these are real… these are real things from BayPath’s website or association that they’re doing related to AI, and I wrote this email, and
417 00:35:36.240 ⇒ 00:35:41.010 Uttam Kumaran: to give you guys, like, what I would have done before,
418 00:35:41.370 ⇒ 00:35:47.650 Uttam Kumaran: I would have tried to use my sales prompt that’s in ChatGPT, but it doesn’t have access to, like.
419 00:35:47.770 ⇒ 00:35:49.359 Uttam Kumaran: A lot of what we do.
420 00:35:49.670 ⇒ 00:35:56.559 Uttam Kumaran: I would also have maybe tried to write this in the platform, but our sales agent doesn’t have, like, a lot of the context.
421 00:35:57.030 ⇒ 00:36:03.959 Uttam Kumaran: And then I… or I probably would have just, like, done it myself, like, the… the usual way.
422 00:36:04.200 ⇒ 00:36:11.249 Uttam Kumaran: And so… Yeah, like, this was, like, this crushed, like.
423 00:36:11.890 ⇒ 00:36:15.379 Uttam Kumaran: I’m gonna send this basically word for word, except maybe a couple things.
424 00:36:15.690 ⇒ 00:36:16.440 Samuel Roberts: miss.
425 00:36:16.440 ⇒ 00:36:23.390 Uttam Kumaran: And this is not… it used files. So basically, yes, like, we should probably consolidate a bunch of these.
426 00:36:25.740 ⇒ 00:36:27.039 Gabriel Lam: Yeah, so I’ll probably…
427 00:36:27.040 ⇒ 00:36:28.439 Samuel Roberts: Kyle’s the one that…
428 00:36:29.380 ⇒ 00:36:34.470 Samuel Roberts: Go on. I think Files is the one that is currently the, like, the marketing asset page on the platform, too.
429 00:36:34.470 ⇒ 00:36:34.840 Gabriel Lam: Yup.
430 00:36:34.840 ⇒ 00:36:38.159 Samuel Roberts: So, I don’t… yeah, don’t… don’t…
431 00:36:38.300 ⇒ 00:36:44.600 Samuel Roberts: Don’t touch that one too much, but, like, if we want to move things into the other repo as well, so that it has it all in one place, that’s not a bad idea.
432 00:36:44.960 ⇒ 00:37:00.370 Gabriel Lam: Yeah, I saw it more as, like, I’m gonna write down everything I see, and everything that… I’m just gonna flag a bunch of repos that I’m like, I don’t think this is necessary, and why that’s the case, and then we can maybe look through all of it, and just, like, in one go, just say, like, that’s what we need, that’s what we don’t need, and then…
433 00:37:00.370 ⇒ 00:37:00.790 Samuel Roberts: Yeah, yeah.
434 00:37:00.790 ⇒ 00:37:01.570 Gabriel Lam: go from there.
435 00:37:03.060 ⇒ 00:37:03.790 Samuel Roberts: Okay, cool.
436 00:37:04.240 ⇒ 00:37:05.040 Gabriel Lam: Yeah.
437 00:37:06.200 ⇒ 00:37:12.569 Gabriel Lam: But glad to hear the playbook’s working. I think I’m gonna refine some of the documentation still, because I noticed
438 00:37:13.010 ⇒ 00:37:14.319 Gabriel Lam: Especially if we’re moving the ball.
439 00:37:14.320 ⇒ 00:37:20.730 Uttam Kumaran: Yeah, the long… the longer you wait to refine it, the more crappy stuff I’m gonna ship out, so…
440 00:37:21.050 ⇒ 00:37:23.200 Uttam Kumaran: All on your shoulders. I’m using this…
441 00:37:23.200 ⇒ 00:37:23.760 Gabriel Lam: I’m on it.
442 00:37:23.760 ⇒ 00:37:24.520 Uttam Kumaran: 3 minutes.
443 00:37:24.520 ⇒ 00:37:25.669 Gabriel Lam: I am on it.
444 00:37:25.670 ⇒ 00:37:31.649 Uttam Kumaran: I’m using this every 30 minutes for major, major things. There’s hundreds of thousands of dollars on the line.
445 00:37:32.090 ⇒ 00:37:51.690 Uttam Kumaran: I’m just… I’m joking, just saying. But, like, literally, please don’t, like, wait to batch stuff, just, like, throw shit in there as you fix stuff, like, because I… I am using it, and… and soon I will… as I have, like, sort of arrived at, like, how to use it for a couple things, I’m… in most of my meetings, I’m trying to share this with people, so…
446 00:37:51.690 ⇒ 00:37:52.230 Gabriel Lam: Yep.
447 00:37:52.540 ⇒ 00:37:53.090 Uttam Kumaran: Yeah.
448 00:37:54.380 ⇒ 00:37:54.900 Gabriel Lam: Yeah.
449 00:37:55.790 ⇒ 00:38:02.950 Gabriel Lam: I guess a question that I have is, like, in the future, how do we imagine putting files in from…
450 00:38:03.210 ⇒ 00:38:12.230 Gabriel Lam: From our platform, like, maybe a wider question of… saving things from Superbase, or… What that might look like.
451 00:38:15.240 ⇒ 00:38:22.969 Samuel Roberts: Yeah, I mean, we could put them right in the repo. The other thing is, like, we could make more use… like, Carson’s just, like, new to call…
452 00:38:23.690 ⇒ 00:38:25.529 Samuel Roberts: And do a web search?
453 00:38:26.240 ⇒ 00:38:29.940 Samuel Roberts: We could add stuff that could just hit Supabase, I suppose.
454 00:38:30.050 ⇒ 00:38:32.209 Samuel Roberts: Like, I don’t know…
455 00:38:32.730 ⇒ 00:38:41.030 Samuel Roberts: I don’t know enough about, like, the way Cursor knows to make certain tool calls for, like, searching files versus searching the internet versus something that we want to custom add to it.
456 00:38:43.940 ⇒ 00:38:53.160 Samuel Roberts: But, yeah, I mean, if everything’s already in Superbase, it seems a little, you know, crazy to just dump that all into a repo. Right. Just to have all the text files there, if we could just have Cursor know to go look.
457 00:38:54.890 ⇒ 00:38:58.920 Gabriel Lam: I can do a spike on the SuperBase MCP, maybe.
458 00:38:59.240 ⇒ 00:39:00.589 Samuel Roberts: Yeah, I mean, even if we have to…
459 00:39:00.590 ⇒ 00:39:05.120 Uttam Kumaran: I don’t… I don’t know, like, I just think, I feel like…
460 00:39:05.420 ⇒ 00:39:08.010 Uttam Kumaran: I would need to understand what the,
461 00:39:09.160 ⇒ 00:39:14.719 Uttam Kumaran: Like, having all the files locally seems like the faster, better option.
462 00:39:14.860 ⇒ 00:39:18.729 Uttam Kumaran: I… but of course, maybe we hit a point where it’s like, okay, there’s, like.
463 00:39:18.880 ⇒ 00:39:24.610 Uttam Kumaran: Hundreds of thousands of transcripts, and we need more, like… we need a database, because…
464 00:39:24.910 ⇒ 00:39:30.299 Uttam Kumaran: we need to, like, give AI the ability to, like, search, and then… and then extract.
465 00:39:30.420 ⇒ 00:39:31.719 Samuel Roberts: Yeah.
466 00:39:32.280 ⇒ 00:39:38.839 Uttam Kumaran: you know, for example, like, I would like us to start to tag transcripts with, like, meeting types.
467 00:39:39.040 ⇒ 00:39:47.689 Uttam Kumaran: like, what was in the meeting, like, short, long, summaries, and… but, like, again, like, it’s also…
468 00:39:48.120 ⇒ 00:39:57.989 Uttam Kumaran: if that’s gonna take a month or two months to do, I’m telling you that, like, having everything open in my GitHub folder and just telling AI to look through everything and figure it out is working perfectly.
469 00:39:57.990 ⇒ 00:39:59.800 Samuel Roberts: Yeah, I mean, that’s something, like.
470 00:40:00.250 ⇒ 00:40:10.179 Uttam Kumaran: more of what the shirt… the last thing I’ll say is more of the short-term alpha, Gabe, is gonna come from the fact that I sent that screenshot of who’s using Cursor, and it’s basically everyone here.
471 00:40:10.330 ⇒ 00:40:16.809 Uttam Kumaran: Plus, plus, like, plus you, plus, like, one other person. As you can tell, nobody on data except for Awash.
472 00:40:16.930 ⇒ 00:40:23.800 Uttam Kumaran: probably is using Cursor, and certainly nobody else on the internal team is using it, and…
473 00:40:24.340 ⇒ 00:40:27.549 Uttam Kumaran: that is, like, like, horrendous, you know, that’s, like, more…
474 00:40:28.430 ⇒ 00:40:29.030 Gabriel Lam: Yeah.
475 00:40:29.030 ⇒ 00:40:29.620 Samuel Roberts: Yeah.
476 00:40:29.850 ⇒ 00:40:35.889 Samuel Roberts: I mean, it’s also nice because we, like, we’re using, like, they’re focused on it being able to interact with these files, like, they’re…
477 00:40:36.220 ⇒ 00:40:43.540 Samuel Roberts: you know, whole job is to make this a good experience, and like, if we’re gonna try to replicate things on the platform, like, it does seem a little bit like we’re not gonna have the…
478 00:40:44.520 ⇒ 00:40:59.230 Samuel Roberts: the manpower, the brainpower, the resources to, like, replicate the experience from here, if it really is just able to search these files as well as it can, and that’s their whole thing, is being able to do that, because, like, they want to handle large codebases, they want to handle all this stuff. It just seems like…
479 00:40:59.380 ⇒ 00:41:01.979 Samuel Roberts: But if it’s working well, that makes total sense, you know?
480 00:41:02.120 ⇒ 00:41:11.230 Uttam Kumaran: I think more, Gay, what… what… I think there’s gonna be what… what is… how does the cursor platform exist in our company? For example, like, what can you govern?
481 00:41:11.400 ⇒ 00:41:16.290 Uttam Kumaran: Like, can you… like, the playbooks thing is one thing that if everybody has playbooks.
482 00:41:16.530 ⇒ 00:41:25.089 Uttam Kumaran: in their folder, then as you add more playbooks, your first thing is to tell people, always tell AI to look at the playbooks.
483 00:41:25.090 ⇒ 00:41:30.779 Samuel Roberts: You know, or like, the second thing is, I don’t know what admin functionality exists in Cursor, but maybe you can…
484 00:41:30.780 ⇒ 00:41:42.729 Uttam Kumaran: modify everybody’s cursor rules, or, like, load people up with prompts, or we can, like, have everybody use the same shortcuts. I think those are what’s gonna get… move people faster.
485 00:41:42.730 ⇒ 00:41:46.610 Gabriel Lam: Yeah, it’s like an adoption blocker more than a technical blocker.
486 00:41:46.770 ⇒ 00:41:47.420 Uttam Kumaran: Yeah.
487 00:41:47.420 ⇒ 00:41:48.140 Gabriel Lam: Yeah.
488 00:41:48.250 ⇒ 00:41:51.390 Uttam Kumaran: Because, dude, I’m telling you, I’m doing some stuff that even…
489 00:41:51.670 ⇒ 00:41:58.279 Uttam Kumaran: a few weeks ago would have taken me so long. And, we just got off of, like.
490 00:41:58.440 ⇒ 00:42:06.829 Uttam Kumaran: to tell you the truth, I just drafted an SOW for CTA, In, like, 90 seconds.
491 00:42:07.080 ⇒ 00:42:12.029 Uttam Kumaran: that I shared, and she was like, this is exactly on point.
492 00:42:12.290 ⇒ 00:42:16.249 Samuel Roberts: And, like, that’s gonna drive probably another $100,000.
493 00:42:16.340 ⇒ 00:42:26.390 Uttam Kumaran: And it’s not the fact that, like, I was lazy, it’s actually more of, like, I did a good job as a cursor user to make sure all the transcripts were there.
494 00:42:26.500 ⇒ 00:42:42.639 Uttam Kumaran: And I told it to look at our last call with her, where she outlined what she wanted to see in that SOW. I then had it do plan mode. I reviewed the plan, I made one more change to the plan, and then I had it use Sonnen to draft it.
495 00:42:42.780 ⇒ 00:42:47.980 Uttam Kumaran: And it basically one-shotted it. I read it, it was perfect. I showed it in the meeting.
496 00:42:50.460 ⇒ 00:43:01.599 Uttam Kumaran: And, like, that’s, like, it’s, like, really complicated to do that right now. And if I did that, and I… and I did that in, like, 90 seconds.
497 00:43:01.940 ⇒ 00:43:08.619 Uttam Kumaran: that just, like, shows there’s so much possible, you know? Whether it is, like, more people being able to draft.
498 00:43:08.640 ⇒ 00:43:25.449 Uttam Kumaran: sales emails from Robert and I’s Gmail, whether it’s people using it to, like, build strategy and SOPs faster, whether it’s, like, you know, engineering teams being able to do things that may be more difficult for them, like decks and project plans and Gantt charts.
499 00:43:25.540 ⇒ 00:43:30.540 Uttam Kumaran: like, I want us to think about that. I feel like I’m really enjoying the…
500 00:43:31.390 ⇒ 00:43:36.829 Uttam Kumaran: the function of using, like, a cursor for that stuff, and then there will be things like the stand-up assistant.
501 00:43:36.960 ⇒ 00:43:39.360 Uttam Kumaran: And things where a UI is necessary.
502 00:43:40.160 ⇒ 00:43:42.300 Uttam Kumaran: And then we moved to the platform, you know?
503 00:43:42.490 ⇒ 00:43:43.100 Gabriel Lam: Alright.
504 00:43:44.250 ⇒ 00:43:44.910 Uttam Kumaran: Yeah.
505 00:43:48.110 ⇒ 00:43:48.670 Samuel Roberts: Coom.
506 00:43:48.960 ⇒ 00:43:49.550 Gabriel Lam: Cool.
507 00:43:50.620 ⇒ 00:43:52.649 Samuel Roberts: No, that’s… yeah, anything else?
508 00:43:52.880 ⇒ 00:43:56.220 Samuel Roberts: Anyone? I guess it’s… we’re at noon now, but we’re at 12 here.
509 00:43:59.260 ⇒ 00:43:59.950 Samuel Roberts: Alright.
510 00:43:59.950 ⇒ 00:44:00.750 Gabriel Lam: Alright.
511 00:44:01.100 ⇒ 00:44:10.419 Uttam Kumaran: Cool. I think, Sam, maybe we slack, and if you want to… do you want to take a stab at the deck first? I’m trying… sorry, I’m not… I’m trying to just get everybody a little bit familiar.
512 00:44:10.420 ⇒ 00:44:14.329 Samuel Roberts: No, I appreciate that. I know, I know, I was, like, a little bit like, oh, I got this other… I’m, like, trying to get this thing…
513 00:44:14.330 ⇒ 00:44:22.300 Uttam Kumaran: But I would like you to maybe… maybe chat with the wage. I want to get more engineers familiar with decoration, because…
514 00:44:22.450 ⇒ 00:44:34.269 Uttam Kumaran: as long as… if everybody here is able to get a deck 60% of the way there, marketing and me and Robert can take it from there. It’s actually the initial creation is like, I want you guys to start to own
515 00:44:34.440 ⇒ 00:44:41.180 Uttam Kumaran: That piece of communication, and we’re not looking for, like, beauty. I’m just looking for… Something, you know?
516 00:44:41.180 ⇒ 00:44:41.700 Samuel Roberts: Yeah.
517 00:44:41.700 ⇒ 00:44:46.360 Uttam Kumaran: And so I do think that I want you to give it a go, if possible. Okay.
518 00:44:46.500 ⇒ 00:44:55.730 Uttam Kumaran: And then ask Awayish and Ashwini. Ashwini just did one this morning that was really good. Awish also has done a couple, so they’re really good resources.
519 00:44:56.000 ⇒ 00:44:59.029 Samuel Roberts: Cool. That way, also, once you guys start building decks.
520 00:44:59.040 ⇒ 00:45:04.099 Uttam Kumaran: the requirements for Gabe on the deck side will be a lot more clear.
521 00:45:04.600 ⇒ 00:45:05.130 Samuel Roberts: Yeah.
522 00:45:05.340 ⇒ 00:45:08.430 Gabriel Lam: It seems like Gamma’s the one that most people are using at the moment, but…
523 00:45:09.010 ⇒ 00:45:11.990 Gabriel Lam: Yeah, I haven’t had the chance to fully…
524 00:45:13.010 ⇒ 00:45:15.230 Gabriel Lam: Try it out in all those different ways.
525 00:45:15.230 ⇒ 00:45:15.940 Uttam Kumaran: Cool.
526 00:45:16.450 ⇒ 00:45:19.379 Mustafa Raja: Yeah, Google Notebook does really good, too.
527 00:45:19.970 ⇒ 00:45:28.360 Uttam Kumaran: Yeah, and I put… I threw in all the transcripts into… into our repo, too, Sam, so you could probably… Oh, if I was… if I was you, I would…
528 00:45:28.690 ⇒ 00:45:35.070 Uttam Kumaran: If… If I was doing this, I would export the deck as a PDF,
529 00:45:35.220 ⇒ 00:45:43.570 Uttam Kumaran: throw it into Cursor, make sure all the transcripts from, Lilo are there, and… and then I would basically…
530 00:45:44.100 ⇒ 00:45:45.640 Uttam Kumaran: Use plan mode.
531 00:45:45.810 ⇒ 00:45:47.489 Samuel Roberts: To have it help you with…
532 00:45:48.080 ⇒ 00:45:50.269 Uttam Kumaran: Creating it and giving you all the copy.
533 00:45:50.440 ⇒ 00:45:51.130 Uttam Kumaran: And…
534 00:45:51.130 ⇒ 00:45:51.929 Samuel Roberts: Okay, yeah.
535 00:45:52.200 ⇒ 00:45:54.080 Uttam Kumaran: Hi, that’s… That’s how I.
536 00:45:54.080 ⇒ 00:45:54.699 Samuel Roberts: Yeah, no, the transfer.
537 00:45:54.700 ⇒ 00:45:55.989 Uttam Kumaran: That’s all I would cheat.
538 00:45:55.990 ⇒ 00:46:13.150 Samuel Roberts: I did a… I did something in that vein, but I realized I was on the branch that I was working on, and not the, probably, main branch where the transcripts are. So it just, like, gave me this master, like, of, like, all the code stuff I did. I’m like, that is not the Jason I was looking for cursor, but if the transcripts are there, I’ll bump… I’ll bump over to branches and do that. That’ll be fine.
539 00:46:13.400 ⇒ 00:46:17.870 Uttam Kumaran: Okay, yeah, and I’ll make sure all my recent stuff, yeah, all my stuff is in there, so…
540 00:46:17.870 ⇒ 00:46:21.239 Samuel Roberts: Okay, cool Yeah, I’ll take a swing.
541 00:46:21.480 ⇒ 00:46:22.470 Uttam Kumaran: Okay, alright.
542 00:46:23.050 ⇒ 00:46:24.000 Samuel Roberts: Alright, thanks, y’all.
543 00:46:24.000 ⇒ 00:46:24.600 Gabriel Lam: Thanks, guys.