Meeting Title: AI-Data Platform Team Standup Date: 2025-08-07 Meeting participants: Sam Roberts, Casie Aviles, Mustafa Raja, Uttam Kumaran
WEBVTT
1 00:00:14.260 ⇒ 00:00:15.340 Sam Roberts: Hello!
2 00:00:19.070 ⇒ 00:00:19.869 Casie Aviles: Hey, guys.
3 00:00:20.110 ⇒ 00:00:22.120 Mustafa Raja: Hey? How are you guys.
4 00:00:22.850 ⇒ 00:00:23.727 Sam Roberts: Doing all right.
5 00:00:25.540 ⇒ 00:00:26.630 Sam Roberts: But you guys.
6 00:00:27.420 ⇒ 00:00:28.670 Mustafa Raja: Yeah. Good.
7 00:00:29.210 ⇒ 00:00:33.580 Sam Roberts: Good, good, cool, cool.
8 00:00:42.020 ⇒ 00:00:46.580 Sam Roberts: Did either you get a chance to look at the spike I put together co-pilots. It.
9 00:00:47.500 ⇒ 00:00:48.110 Mustafa Raja: Yep.
10 00:00:48.860 ⇒ 00:00:49.800 Sam Roberts: Okay. Cool.
11 00:00:52.660 ⇒ 00:00:53.440 Casie Aviles: Got it.
12 00:00:57.080 ⇒ 00:00:59.429 Sam Roberts: Okay, no problem. I can give a quick update
13 00:01:00.640 ⇒ 00:01:02.610 Sam Roberts: and looks like Utam left some.
14 00:01:02.930 ⇒ 00:01:09.039 Mustafa Raja: Yeah, yeah, I had the same question as utham. Actually, yeah. And we just
15 00:01:09.040 ⇒ 00:01:13.440 Mustafa Raja: directly hook up, hook it up with the output of our Internet and agent.
16 00:01:14.220 ⇒ 00:01:21.210 Sam Roberts: I don’t think so, but the other ones he mentioned look a little so copilot Kit is a little more
17 00:01:24.620 ⇒ 00:01:28.230 Sam Roberts: featured but a little more front and back end tied together.
18 00:01:28.570 ⇒ 00:01:29.400 Mustafa Raja: Yeah.
19 00:01:29.400 ⇒ 00:01:37.660 Sam Roberts: Whereas the other ones he mentions I’m looking at the AI SDK from Vercell, which I haven’t looked at in a while, and Assistant ui look like a little more.
20 00:01:40.561 ⇒ 00:01:47.850 Sam Roberts: take a little more work to get up and running, but we can probably use any. Then, you know, copilot Kit is kind of intended. It seems to like
21 00:01:47.960 ⇒ 00:01:53.400 Sam Roberts: drop into an app that already has a bunch of data. And you can just kind of use those react hooks
22 00:01:53.800 ⇒ 00:01:56.360 Sam Roberts: that just like give it access to.
23 00:01:57.240 ⇒ 00:02:00.310 Sam Roberts: You know any of the context. It needs
24 00:02:01.610 ⇒ 00:02:02.250 Mustafa Raja: Yeah.
25 00:02:03.040 ⇒ 00:02:04.179 Uttam Kumaran: Oh, hi guys.
26 00:02:04.910 ⇒ 00:02:05.640 Sam Roberts: Hey?
27 00:02:11.240 ⇒ 00:02:11.820 Sam Roberts: Going all right.
28 00:02:17.025 ⇒ 00:02:25.710 Uttam Kumaran: Cool. Okay, does someone want to on today
29 00:02:30.620 ⇒ 00:02:34.750 Uttam Kumaran: yesterday with software saying, you guys want to try to run today.
30 00:02:35.960 ⇒ 00:02:38.670 Mustafa Raja: Yeah, I can. I can share my screen.
31 00:02:39.030 ⇒ 00:02:39.730 Uttam Kumaran: Okay.
32 00:02:46.820 ⇒ 00:02:48.249 Mustafa Raja: A minute, please.
33 00:02:59.520 ⇒ 00:03:02.750 Mustafa Raja: Jesse, do you know which which views do we use.
34 00:03:04.440 ⇒ 00:03:05.380 Casie Aviles: My team
35 00:03:07.100 ⇒ 00:03:14.270 Casie Aviles: active tickets. Can you check that? Or it’s either that or a item. All statuses with client and internal.
36 00:03:15.170 ⇒ 00:03:16.490 Mustafa Raja: This one.
37 00:03:16.950 ⇒ 00:03:21.499 Uttam Kumaran: Yeah, that one is gonna be all the active ones and the one below it is gonna be
38 00:03:21.810 ⇒ 00:03:23.260 Uttam Kumaran: no matter the style.
39 00:03:25.850 ⇒ 00:03:26.960 Casie Aviles: Yeah, that’s fine.
40 00:03:27.500 ⇒ 00:03:28.529 Mustafa Raja: This one or that one.
41 00:03:29.990 ⇒ 00:03:31.220 Uttam Kumaran: Either one? Was it.
42 00:03:31.560 ⇒ 00:03:32.160 Mustafa Raja: Okay.
43 00:03:32.550 ⇒ 00:03:33.230 Uttam Kumaran: Good.
44 00:03:36.810 ⇒ 00:03:38.819 Mustafa Raja: Okay, let’s go.
45 00:03:42.010 ⇒ 00:03:44.900 Mustafa Raja: Yeah, we can talk about this spike.
46 00:03:49.704 ⇒ 00:03:50.159 Sam Roberts: So
47 00:03:50.870 ⇒ 00:03:58.693 Sam Roberts: I was. I just mentioned a little bit, but the overall is, it’s a pretty good like drop in kind of thing to add
48 00:03:59.890 ⇒ 00:04:06.419 Sam Roberts: chat to kind of existing apps a little bit where there’s data and react already. And you can just add hooks.
49 00:04:06.990 ⇒ 00:04:18.029 Sam Roberts: for to answer some of the questions like Utam like you mentioned, I think, Mustafa had the same question. I don’t really see a way to just have it hit a certain endpoint.
50 00:04:18.980 ⇒ 00:04:24.940 Sam Roberts: It’s kind of tied to its co-pilot kit back end. Either the cloud or the self hosted one.
51 00:04:24.940 ⇒ 00:04:26.220 Uttam Kumaran: Oh, really. Okay.
52 00:04:26.440 ⇒ 00:04:27.809 Sam Roberts: There may be a.
53 00:04:28.480 ⇒ 00:04:31.879 Uttam Kumaran: Sorry you can’t hit. You can’t hit an endpoint at all.
54 00:04:32.170 ⇒ 00:04:43.549 Sam Roberts: Well, I I bet there’s a way to, you know, based on what I’m seeing you. You could add certain like back end actions. And then, so you might be able to like, have it hit the back end, then hit an am. But that seems like it’s not
55 00:04:44.454 ⇒ 00:04:57.079 Sam Roberts: ideal looking at some of the other ones you mentioned. I hadn’t looked at the AI SDK from Vercell in a while. And then so it’s it’s updated. Since I’ve looked at it a little bit. You know, that has some of the
56 00:04:58.210 ⇒ 00:04:59.620 Sam Roberts: functions we’d need.
57 00:04:59.770 ⇒ 00:05:08.148 Sam Roberts: Kind of build it without having to add all the streaming and everything. Assistant ui looks pretty good as well for that, but those are both definitely a little more.
58 00:05:08.820 ⇒ 00:05:11.809 Sam Roberts: you know, might have to customize it a little bit more.
59 00:05:12.641 ⇒ 00:05:16.090 Sam Roberts: What I will say about copilot kit is
60 00:05:17.480 ⇒ 00:05:28.441 Sam Roberts: the the ui it has is nice. It’s it’s great for that it plugs into react. Well, if if there’s a time when we want to use some of those other frameworks like Lang graph or crew, AI,
61 00:05:29.330 ⇒ 00:05:31.060 Sam Roberts: it’s probably great for that.
62 00:05:31.510 ⇒ 00:05:38.420 Sam Roberts: Wow for integrating with the N. 8 N. Stuff. It doesn’t seem to play very nice together.
63 00:05:40.270 ⇒ 00:05:46.589 Sam Roberts: What it could do is, and this would probably involve a little bit more of like, you know. So
64 00:05:46.720 ⇒ 00:06:00.521 Sam Roberts: you know, based on what I’m seeing with the current chat like that hits that endpoint, and it end runs that whole thing, and then comes back. What you could do is if you wanted to make them like different tools available.
65 00:06:01.680 ⇒ 00:06:04.540 Sam Roberts: you could set up co-pilot kit
66 00:06:04.780 ⇒ 00:06:15.570 Sam Roberts: in such a way that you know it’s the basic chat functionality, the pinning Lm doing all the stuff that it needs to do. You could add custom actions.
67 00:06:16.208 ⇒ 00:06:25.899 Sam Roberts: Which are pretty easy to register, like I mentioned, like, yeah, use copilot access. So you can just kind of easily register and react. You have it on the back end. That then you could probably make calls to
68 00:06:26.400 ⇒ 00:06:27.817 Sam Roberts: and it, n
69 00:06:29.390 ⇒ 00:06:33.959 Sam Roberts: but if it’s just gonna be like going to any then and back. It’s probably not worth it.
70 00:06:34.250 ⇒ 00:06:36.619 Sam Roberts: as currently like, set up.
71 00:06:37.800 ⇒ 00:06:41.340 Uttam Kumaran: Yeah, I guess I’m just like curious. And like the I mean.
72 00:06:41.500 ⇒ 00:06:51.929 Uttam Kumaran: I will say, like the options that they have for the ui are like great like those exact things that I’d like to use. And for someone like them who’s gonna
73 00:06:52.140 ⇒ 00:06:54.760 Uttam Kumaran: they’re gonna continue to push the barriers like. And we.
74 00:06:54.760 ⇒ 00:06:55.800 Sam Roberts: Exactly.
75 00:06:55.800 ⇒ 00:07:02.929 Uttam Kumaran: Absorb whatever features they develop. So I don’t know. I feel like there’s not a better option right now.
76 00:07:03.080 ⇒ 00:07:19.978 Sam Roberts: Yeah, I mean, they. And you know the other ones that you mentioned, like I was looking at like they seem okay. But it’s definitely like a little more work setting it up kind of drops right in and and works except for this. So I don’t know if that means. Well, it could mean a few things.
77 00:07:20.710 ⇒ 00:07:27.750 Sam Roberts: One of the thoughts, and I think I mentioned it was like, maybe we try to go a little bit further with it. We may have to
78 00:07:28.721 ⇒ 00:07:35.650 Sam Roberts: either tweak how and it end works and make it like multiple pieces
79 00:07:35.860 ⇒ 00:07:50.000 Sam Roberts: that could plug into this or we just maybe we do what I said initially, which is like may add, like, set up our own back end and have it just point to am and kind of see how the latency is. See how it all goes. For implementing it into other.
80 00:07:50.390 ⇒ 00:07:52.540 Sam Roberts: Excuse me, other uis, if it’s like
81 00:07:53.000 ⇒ 00:08:05.577 Sam Roberts: for a client thing, you know that that’s might be. It might be a different consideration, because then, if you’re building new, any end flows, you could do it in such a way where it’s not one big flow, but lots of little tools that it could call
82 00:08:07.070 ⇒ 00:08:16.546 Sam Roberts: But as it’s currently set up, that might be the best way to do it is just let’s let’s see how bad the latency gets it could be nice to play around with it a little bit more
83 00:08:17.680 ⇒ 00:08:18.549 Sam Roberts: because it is. It is.
84 00:08:18.550 ⇒ 00:08:23.919 Uttam Kumaran: I’m open to it. I’m open to it because actually, a lot of our use cases are not like
85 00:08:24.440 ⇒ 00:08:30.870 Uttam Kumaran: out of our internal use cases, I would say, accuracy is probably more important than
86 00:08:31.040 ⇒ 00:08:33.299 Uttam Kumaran: having it come back immediately.
87 00:08:35.340 ⇒ 00:08:50.539 Uttam Kumaran: like, you know, generating, follow up emails generating. Let linear tickets like those types of things. We want to get right as much as possible versus like, I don’t mind if the user has to wait a little bit as long as we indicate that. So I mean my vote. My vote
88 00:08:51.146 ⇒ 00:08:54.510 Uttam Kumaran: is that we move forward with it, and then we
89 00:08:54.680 ⇒ 00:08:56.430 Uttam Kumaran: I would say, we try to take
90 00:08:56.840 ⇒ 00:08:59.651 Uttam Kumaran: you know any of our chat?
91 00:09:01.080 ⇒ 00:09:07.079 Uttam Kumaran: Any of our chat like interfaces. We we just replace with this as the as the next step.
92 00:09:09.030 ⇒ 00:09:13.019 Sam Roberts: Okay, yeah, I mean, I think, part.
93 00:09:13.020 ⇒ 00:09:16.079 Uttam Kumaran: Let me see. See? See how it works within it. Then, yeah.
94 00:09:16.080 ⇒ 00:09:19.279 Sam Roberts: Yeah, exactly. So. I think the the next step would be
95 00:09:19.720 ⇒ 00:09:37.149 Sam Roberts: setting that up like I, I can even do a quick test because I have it running locally with that I can get the back end going, and and just see how it, if it if it even works that way properly. I imagine it should, because I I think it’s pretty open ended once you like, give it an action. It’s just
96 00:09:37.290 ⇒ 00:09:49.948 Sam Roberts: typescript to to run that on the server so like if it if it’s just hitting an 8 n. And then sending it back as long as everything kind of plugs together and plays nicely. That shouldn’t be crazy.
97 00:09:50.670 ⇒ 00:09:52.240 Sam Roberts: But yeah, I think I can definitely
98 00:09:52.550 ⇒ 00:09:54.979 Sam Roberts: cash ball a little further forward to make sure that that’s good.
99 00:09:55.710 ⇒ 00:10:06.020 Uttam Kumaran: Okay, yeah, I don’t. I don’t mind if we have to. Self post stuff or whatever. Yeah, I think, I think, since you’re already kind of like, probably have everything open. Yeah, maybe
100 00:10:07.230 ⇒ 00:10:15.809 Uttam Kumaran: we just maybe we just close it out. And then if you can even develop a proof of concept or something, and then we could push that code. Then the rest of the team can use that as a baseline for stuff in terms of.
101 00:10:15.810 ⇒ 00:10:16.430 Sam Roberts: Like.
102 00:10:16.710 ⇒ 00:10:21.310 Uttam Kumaran: Yeah, in terms of our stuff. I mean, there are all the chat interfaces that we have are really the
103 00:10:21.856 ⇒ 00:10:40.000 Uttam Kumaran: that that the original chat that’s on the on the dashboard part of the platform. We also have the meeting specific chat for all of those like. There’s so much stuff on the Ui side that we want to do like, suggested answers we would. May want to do a canvas thing like.
104 00:10:40.140 ⇒ 00:10:47.660 Uttam Kumaran: And so I I do think that it’s it’s worth moving forward with. And then that’s our framework. Now versus having to
105 00:10:47.990 ⇒ 00:10:50.150 Uttam Kumaran: to generate our own experiences.
106 00:10:50.590 ⇒ 00:10:54.920 Sam Roberts: Yeah, I think that definitely, because, like, there’s, it’s it can become a big rabbit hole of.
107 00:10:54.920 ⇒ 00:10:55.260 Uttam Kumaran: Yeah.
108 00:10:55.260 ⇒ 00:10:57.589 Sam Roberts: All those little things that are, you know.
109 00:10:57.860 ⇒ 00:11:05.920 Sam Roberts: lot of work for, like just getting the streaming right, and a lot of work, for, you know, getting the spinner, and just like a lot of little things that hopefully, this is taken care of. And you know.
110 00:11:05.920 ⇒ 00:11:10.370 Sam Roberts: yes, experience across everything. I think that’s right. The only.
111 00:11:10.370 ⇒ 00:11:11.120 Uttam Kumaran: Yeah. And then for.
112 00:11:11.120 ⇒ 00:11:11.840 Sam Roberts: For, for.
113 00:11:11.840 ⇒ 00:11:12.430 Uttam Kumaran: For clients.
114 00:11:12.430 ⇒ 00:11:17.815 Uttam Kumaran: Okay, like, I, I think, for clients. I’ll let you know as soon as we get requests.
115 00:11:18.410 ⇒ 00:11:30.949 Uttam Kumaran: we will basically just move forward with with, this is our option. One. The only other thing I would also like to do is to get a demo of something that we can right now we have our demo site
116 00:11:31.930 ⇒ 00:11:36.559 Uttam Kumaran: where we have, like a few chat interface demos. I would love to just replace them with
117 00:11:36.988 ⇒ 00:11:41.001 Uttam Kumaran: branded copilot kit demo. So those are additional tickets that we can create.
118 00:11:41.810 ⇒ 00:11:46.049 Sam Roberts: Okay, yeah. I looked at those a little bit when I was kind of documenting some of the platform stuff.
119 00:11:46.050 ⇒ 00:11:46.400 Uttam Kumaran: Yeah.
120 00:11:46.400 ⇒ 00:11:47.580 Sam Roberts: Be prompts.
121 00:11:47.830 ⇒ 00:11:48.860 Sam Roberts: Yes, it’s not.
122 00:11:48.860 ⇒ 00:11:55.610 Sam Roberts: Yeah. In which case that’s that’s probably the best use case for copilot kit, because we can just feed it the right prompt. And it’s not doing.
123 00:11:55.610 ⇒ 00:12:02.479 Uttam Kumaran: Yeah. And for those yeah, for those, we don’t even need to hit any back end, really or like, it can all be synthetic, you know.
124 00:12:04.330 ⇒ 00:12:13.560 Uttam Kumaran: Like those are fake experiences like it could pretend to pull from documents or whatever. But again, we could host that all in in. We don’t have to actually
125 00:12:14.290 ⇒ 00:12:16.740 Uttam Kumaran: do anything outside of that. So perfect.
126 00:12:17.120 ⇒ 00:12:21.819 Sam Roberts: Okay, cool. Yeah, I I can. I can probably do that relatively easily.
127 00:12:22.120 ⇒ 00:12:30.379 Uttam Kumaran: Cool. Yeah, I think if you take, if you, if you after this meeting, if you take this transcript, you can use our transcript to ticket generator, and it’ll probably bring them.
128 00:12:30.790 ⇒ 00:12:33.209 Sam Roberts: That sounds perfect, cool.
129 00:12:35.350 ⇒ 00:12:36.380 Uttam Kumaran: But yeah, that’s.
130 00:12:36.380 ⇒ 00:12:38.271 Sam Roberts: Pretty much it for that.
131 00:12:39.140 ⇒ 00:12:49.570 Sam Roberts: The only other thing I’ll add is that setting up the back end for copilot kit might tie into some of the other stuff I’ve been thinking about in terms of combining the front end and back into the platform into one.
132 00:12:49.990 ⇒ 00:12:50.790 Mustafa Raja: Hmm.
133 00:12:50.790 ⇒ 00:12:56.770 Sam Roberts: Possibly, you know, moving everything to like a next app that it will be easier to just deploy at one. Go?
134 00:12:58.290 ⇒ 00:13:11.800 Sam Roberts: I’m not. I just. I’m just as we’re moving forward with some of the stuff. This is something to keep in mind. I just, you know, want to kind of get that out there a little bit, because that would eliminate some of the deployment issues we have where there’s a different front and different back end. We got to point the Urls to the right thing, and everything.
135 00:13:12.140 ⇒ 00:13:13.470 Uttam Kumaran: I agree, yeah.
136 00:13:13.470 ⇒ 00:13:20.850 Sam Roberts: So I don’t know. Obviously there’s no like tickets for that yet, but I’m just. I might put together a little plan of like migrating to that sounds cool.
137 00:13:21.000 ⇒ 00:13:21.580 Uttam Kumaran: Okay.
138 00:13:25.660 ⇒ 00:13:27.969 Mustafa Raja: Yeah, let’s talk about this.
139 00:13:29.730 ⇒ 00:13:35.360 Casie Aviles: Okay, yeah, I created this ticket. So in response to like the creation of the client hubs.
140 00:13:35.630 ⇒ 00:13:41.380 Casie Aviles: So the main reason why is I I found out that the existing
141 00:13:42.390 ⇒ 00:13:47.240 Casie Aviles: our our embedding step for the zoom is tied to the summarization workflow.
142 00:13:49.170 ⇒ 00:13:55.879 Casie Aviles: Some changes had made it happen inconsistently, so not all of the meetings are getting vectorized.
143 00:13:57.630 ⇒ 00:14:02.319 Casie Aviles: So my proposal here is to create, like a and a separate
144 00:14:03.428 ⇒ 00:14:06.239 Casie Aviles: vectorization which is not tied to
145 00:14:06.440 ⇒ 00:14:10.039 Casie Aviles: any existing workflows. So we could just trigger it
146 00:14:11.700 ⇒ 00:14:21.149 Casie Aviles: whenever we need. And yeah, and should be reusable. So I create. I’m currently working on the python script that will handle
147 00:14:22.120 ⇒ 00:14:28.929 Casie Aviles: that will take basically get the meetings from our master table, which is now the one we’re using for the platform
148 00:14:29.730 ⇒ 00:14:34.570 Casie Aviles: before we didn’t have something like that with the client hub. So
149 00:14:34.970 ⇒ 00:14:41.510 Casie Aviles: that’s my idea is to get from that master table and create the vector tables from there.
150 00:14:42.950 ⇒ 00:14:49.490 Casie Aviles: Yeah. So that’s that’s pretty much why I put the client hubs on blocked until I finish this one.
151 00:14:50.400 ⇒ 00:14:51.150 Uttam Kumaran: Okay. Okay.
152 00:14:51.150 ⇒ 00:14:51.690 Mustafa Raja: Okay.
153 00:14:52.520 ⇒ 00:15:00.220 Uttam Kumaran: Can you? Can you actually indicate that this one is blocking those other ones? Or this one is blocking the
154 00:15:01.060 ⇒ 00:15:03.650 Uttam Kumaran: the Zoom One for the deal.
155 00:15:03.980 ⇒ 00:15:04.310 Casie Aviles: Ones.
156 00:15:05.520 ⇒ 00:15:06.630 Uttam Kumaran: So you can
157 00:15:07.820 ⇒ 00:15:15.799 Uttam Kumaran: do if you if you exit X out of this like escape and then hit command. K, yeah, you can. You can just type in block blocking issue.
158 00:15:17.140 ⇒ 00:15:26.679 Uttam Kumaran: So go into that ticket. The standardized Zoom rag hit, command K, or control K hit blocking.
159 00:15:29.400 ⇒ 00:15:31.970 Uttam Kumaran: Yes, not that one.
160 00:15:31.970 ⇒ 00:15:32.550 Sam Roberts: Mark issue.
161 00:15:32.855 ⇒ 00:15:33.690 Sam Roberts: But 1st one.
162 00:15:33.690 ⇒ 00:15:36.960 Uttam Kumaran: Yeah, I think it’s that one.
163 00:15:37.350 ⇒ 00:15:38.080 Uttam Kumaran: Yeah.
164 00:15:38.330 ⇒ 00:15:40.119 Uttam Kumaran: And then you can type in default.
165 00:15:44.530 ⇒ 00:15:46.970 Uttam Kumaran: It’s like default client hub.
166 00:15:49.200 ⇒ 00:15:50.909 Casie Aviles: Yeah. Default. Client.
167 00:15:51.100 ⇒ 00:15:53.215 Uttam Kumaran: That’s the one. Yeah, you click on that. Yeah.
168 00:15:54.190 ⇒ 00:15:54.940 Uttam Kumaran: Great.
169 00:15:57.050 ⇒ 00:16:10.279 Mustafa Raja: Okay. So for this one, I need to start working on this today. What I’ll do today is I’ll create the pipeline to create a sync between
170 00:16:10.900 ⇒ 00:16:13.340 Mustafa Raja: github files in super base.
171 00:16:18.690 ⇒ 00:16:19.390 Mustafa Raja: Yeah.
172 00:16:19.877 ⇒ 00:16:26.529 Uttam Kumaran: Okay, I I honestly don’t think this is like super important dude like, how are you doing on time? I feel like we should
173 00:16:27.580 ⇒ 00:16:29.939 Uttam Kumaran: like. We can fund this the next week.
174 00:16:30.500 ⇒ 00:16:31.750 Mustafa Raja: I have plenty of time.
175 00:16:33.040 ⇒ 00:16:36.257 Uttam Kumaran: Okay, is there anything else in in the
176 00:16:37.850 ⇒ 00:16:45.819 Uttam Kumaran: in the sprint like, let’s just look at what’s in in to do. Yeah. The departments would be
177 00:16:46.500 ⇒ 00:16:49.230 Uttam Kumaran: much more important than this one.
178 00:16:49.970 ⇒ 00:16:52.989 Mustafa Raja: Oh, so I should do the department then right.
179 00:16:52.990 ⇒ 00:16:59.330 Uttam Kumaran: Yes, like, let’s let’s move the title and descriptions to like medium priority.
180 00:17:01.730 ⇒ 00:17:05.409 Uttam Kumaran: And then the video player, we can move to high priority. That’s
181 00:17:05.829 ⇒ 00:17:09.294 Uttam Kumaran: that’s also because basically just waiting to do that.
182 00:17:09.680 ⇒ 00:17:09.994 Uttam Kumaran: Yeah,
183 00:17:11.619 ⇒ 00:17:16.724 Uttam Kumaran: And then, yeah, Casey, there’s only there’s 1 other thing.
184 00:17:17.209 ⇒ 00:17:21.079 Uttam Kumaran: where we got some work in from insomnia cookies.
185 00:17:21.510 ⇒ 00:17:30.159 Uttam Kumaran: But I I’m kind of one thing I’m worried about is Mustafa. You’re on this stuff and you’re on default and interlude. I think jumping between 4
186 00:17:30.369 ⇒ 00:17:36.409 Uttam Kumaran: is gonna be a lot. I would rather just ha hand off the stuff to Casey.
187 00:17:36.803 ⇒ 00:17:40.219 Uttam Kumaran: How do you feel about that like. I’ll leave you in that channel, of course. But
188 00:17:40.768 ⇒ 00:17:45.779 Uttam Kumaran: since Casey is just has ABC. And internal stuff, I think it would be great for him to to see that.
189 00:17:46.240 ⇒ 00:17:48.700 Mustafa Raja: Yeah, yeah, I’m I’m okay with it.
190 00:17:49.190 ⇒ 00:17:49.720 Uttam Kumaran: Okay.
191 00:17:50.127 ⇒ 00:17:51.350 Casie Aviles: Yeah, that’s fine.
192 00:17:51.650 ⇒ 00:17:56.209 Uttam Kumaran: Okay, it’s a, it’s a pretty simple automation that they want built. So
193 00:17:56.933 ⇒ 00:18:01.530 Uttam Kumaran: we have a spike there. And yeah, we couldn’t
194 00:18:02.090 ⇒ 00:18:04.859 Uttam Kumaran: if you can, Mustafa, if you can tag him there and then.
195 00:18:05.260 ⇒ 00:18:06.020 Mustafa Raja: Yeah, yeah.
196 00:18:06.600 ⇒ 00:18:07.230 Uttam Kumaran: Okay. Cool.
197 00:18:07.840 ⇒ 00:18:12.909 Mustafa Raja: I’ll create the tickets and and then I’ll inform Casey about them.
198 00:18:13.380 ⇒ 00:18:15.763 Uttam Kumaran: Yeah, just send it in that channel. So everybody can see.
199 00:18:16.180 ⇒ 00:18:20.170 Mustafa Raja: Okay, okay, great, yeah. So we want to talk about this.
200 00:18:20.810 ⇒ 00:18:21.495 Uttam Kumaran: Yes,
201 00:18:22.180 ⇒ 00:18:22.630 Sam Roberts: Yeah.
202 00:18:22.630 ⇒ 00:18:23.420 Uttam Kumaran: Yeah. Go ahead.
203 00:18:23.690 ⇒ 00:18:28.479 Sam Roberts: I was just gonna say I. So I spent a little bit of time yesterday, just kind of documenting the current like
204 00:18:28.630 ⇒ 00:18:32.000 Sam Roberts: structure of the the platform. There’s.
205 00:18:32.520 ⇒ 00:18:36.600 Sam Roberts: you know, if it’s this is just gonna be a living document. I think we’re in a pretty good.
206 00:18:36.700 ⇒ 00:18:40.010 Uttam Kumaran: Do you wanna pull up the notion? Mustafa.
207 00:18:42.190 ⇒ 00:18:44.289 Sam Roberts: Yeah. So this is just got.
208 00:18:46.200 ⇒ 00:18:51.229 Sam Roberts: yeah, the goals, the Kpis stakeholders, all the general stuff, or all the
209 00:18:51.340 ⇒ 00:19:01.509 Sam Roberts: general and all the different tables that got added, and then now below it, it’s got the current structure. So I I’m not really sure the best way to do this in
210 00:19:01.630 ⇒ 00:19:10.020 Sam Roberts: notion, is there? I haven’t used notion like in depth in a while. I don’t know. There’s there’s not a way to link internally in between things in in one file. Is there.
211 00:19:12.235 ⇒ 00:19:16.599 Uttam Kumaran: There is like, you’re just trying to like link to another page.
212 00:19:16.850 ⇒ 00:19:23.430 Sam Roberts: I want to link to like some other things on this page. What I could probably do is just start to make pages for these things. That’s probably the notion way to do it, isn’t it?
213 00:19:23.430 ⇒ 00:19:29.789 Uttam Kumaran: Yeah, what? So what you could do? Yeah, you could just do slash link to page. Or if some of these become like pages themselves, then, yeah.
214 00:19:29.790 ⇒ 00:19:31.550 Sam Roberts: That’s what. Okay, yeah.
215 00:19:31.550 ⇒ 00:19:33.450 Uttam Kumaran: More like wiki wiki style.
216 00:19:33.450 ⇒ 00:19:40.383 Sam Roberts: That’s probably what I yeah, I’m just not thinking notion as much as I’m I’m used to. Just I because I was gonna start linking what
217 00:19:40.770 ⇒ 00:19:47.369 Sam Roberts: Some of the like kpis and and the processes and stuff to like where those things live in. If you scroll down a little bit
218 00:19:50.200 ⇒ 00:19:53.590 Sam Roberts: further further, further. So like right now, this kind of maps out
219 00:19:54.150 ⇒ 00:19:56.860 Sam Roberts: what is in the platform. And I started to want to start.
220 00:19:57.090 ⇒ 00:19:57.320 Uttam Kumaran: No.
221 00:19:57.320 ⇒ 00:19:58.979 Sam Roberts: Parallels to what is
222 00:19:59.360 ⇒ 00:20:08.459 Sam Roberts: kind of there already, and what needs like more like that, different metrics to display, and things like that? And like a few questions I had about different things.
223 00:20:08.460 ⇒ 00:20:24.239 Uttam Kumaran: Yeah, probably. Probably my open. My open question is like, since we’re since we’re not, you know, we don’t do like product. We don’t have like Prds for a lot of stuff. But maybe it is that like we should end up having a Prd for every core
224 00:20:24.670 ⇒ 00:20:30.289 Uttam Kumaran: like product area. And then, if it’s like more of a 1 off feature, we can do like Prd lights
225 00:20:30.490 ⇒ 00:20:36.499 Uttam Kumaran: like I’m happy to. I’m happy to to like, collaborate, and and write those.
226 00:20:36.620 ⇒ 00:20:39.174 Uttam Kumaran: But, like, I do agree that
227 00:20:40.660 ⇒ 00:20:51.729 Uttam Kumaran: we probably start now that these are getting a little bit bigger, we probably will need some living documents around like the pro, the products and features themselves. And this can turn into more of a
228 00:20:52.080 ⇒ 00:20:53.979 Uttam Kumaran: like a wiki over time. You know.
229 00:20:53.980 ⇒ 00:20:59.740 Sam Roberts: That’s probably yeah. Cause yeah, I was even just as I’m starting to break some of the stuff out. I realize I’m getting kind of nested, and I just wasn’t sure.
230 00:21:01.440 ⇒ 00:21:16.043 Sam Roberts: yeah, I haven’t I? My notion experience is a little older. I haven’t touched it kind of recently. So I was just like my brain wasn’t thinking like, Oh, yeah, these can be sub pages. And that’s kind of the way everything works. So yeah, I think that’s probably the best way to keep this going. But
231 00:21:17.240 ⇒ 00:21:17.940 Sam Roberts: yeah,
232 00:21:18.890 ⇒ 00:21:23.333 Sam Roberts: appeared is probably a good idea. Because this is kind of a yeah, more of a product thing.
233 00:21:24.900 ⇒ 00:21:29.160 Sam Roberts: yeah, I think it’s not not a bad, not a bad thought. We can kind of connect on that because I I definitely
234 00:21:29.500 ⇒ 00:21:35.909 Sam Roberts: wanna be moving this forward. But I also am not the like core stakeholder for some of the stuff. So I want to get a better sense of.
235 00:21:35.910 ⇒ 00:21:37.250 Uttam Kumaran: Okay, cool. So if we’re.
236 00:21:37.250 ⇒ 00:21:38.209 Sam Roberts: Some of that.
237 00:21:38.210 ⇒ 00:21:42.870 Uttam Kumaran: Yeah, if we feel good about. If we feel good about this, I can leave
238 00:21:45.000 ⇒ 00:21:52.450 Uttam Kumaran: comments there and then, like, I think this would be something great to also share out. Tomorrow’s in tomorrow’s meeting with everybody.
239 00:21:52.580 ⇒ 00:21:53.250 Sam Roberts: Totally.
240 00:21:53.620 ⇒ 00:21:57.549 Sam Roberts: That’s that’s a good point. Yeah, there were definitely things where I’m like, what do people actually like
241 00:21:57.680 ⇒ 00:22:09.490 Sam Roberts: using currently, what is what is being adopted. What is not adopted like what needs to you know what are the things that are missing that people would want like. I know you’ve mentioned a few things, but it’d be great to hear from people that are
242 00:22:09.770 ⇒ 00:22:13.640 Sam Roberts: in it if they’re if they’re in there doing stuff. It’s like, Oh, this
243 00:22:14.010 ⇒ 00:22:19.140 Sam Roberts: rough edge exists, or this is just missing. Yeah.
244 00:22:19.140 ⇒ 00:22:28.679 Uttam Kumaran: Yeah. And I think in our meeting with with a wish later, let’s also plan on, maybe walking him through this. And then, yeah, I’m I’m sort of starting to read more about like.
245 00:22:28.960 ⇒ 00:22:34.030 Uttam Kumaran: you know, how do? How do you start to encourage AI adoption?
246 00:22:34.150 ⇒ 00:22:50.739 Uttam Kumaran: You know we’re we’re we’re doing this for clients and ourselves. You know, I think, what what one thing is. We started to have these like cross team slack channels like AI and the design team AI and the Pm team. I think we will probably need to start something
247 00:22:50.910 ⇒ 00:22:56.009 Uttam Kumaran: like on a monthly basis. That’s more of like, almost like an internal council or something.
248 00:22:56.482 ⇒ 00:22:58.659 Uttam Kumaran: It doesn’t have to be like on a
249 00:22:58.910 ⇒ 00:23:03.700 Uttam Kumaran: on like a weekly thing. But ideally, it’s like one person from every
250 00:23:04.030 ⇒ 00:23:17.480 Uttam Kumaran: like core department or like that, can cover an area that can come in and basically report on how their divisions are using AI ideally for those meetings. It would be great to start to look at data like, okay.
251 00:23:17.670 ⇒ 00:23:27.657 Uttam Kumaran: your processes or your teams are using this many tokens. Here’s the leaderboard of people are using it. And then what are things that you want our AI team to build right?
252 00:23:27.970 ⇒ 00:23:28.570 Sam Roberts: Okay.
253 00:23:28.780 ⇒ 00:23:31.117 Uttam Kumaran: So May. Maybe that’s something that I think.
254 00:23:31.700 ⇒ 00:23:42.369 Uttam Kumaran: I think we can. We’ll talk about later today as well as I’m reading this book about it, and it’s it’s it’s kind of giving us a good layout. I don’t want to be too big company about it, but I also
255 00:23:42.860 ⇒ 00:23:52.059 Uttam Kumaran: again. I just wanna get me out of the loop. Because I’m I’m sort of the the one core person that’s bridging the gap a lot. And I think the idea is
256 00:23:52.200 ⇒ 00:24:00.059 Uttam Kumaran: the people already exist with ideas, and then we already now are now gonna exist with, like the tools and the structure to execute. And then it’s just like.
257 00:24:00.390 ⇒ 00:24:08.100 Uttam Kumaran: okay, how can we measure that that these are getting adopted? And there’s several people that are like on the hook for it not just me or you, Sam, you know.
258 00:24:08.410 ⇒ 00:24:10.969 Sam Roberts: Totally. Yeah. I think that that sounds smart.
259 00:24:11.430 ⇒ 00:24:12.040 Uttam Kumaran: Okay.
260 00:24:14.820 ⇒ 00:24:15.520 Mustafa Raja: Okay.
261 00:24:15.520 ⇒ 00:24:15.855 Uttam Kumaran: Okay.
262 00:24:20.340 ⇒ 00:24:27.140 Mustafa Raja: So for this. We did talk about adding these departments in the teams column.
263 00:24:28.761 ⇒ 00:24:32.874 Mustafa Raja: and a process where we
264 00:24:34.220 ⇒ 00:24:42.949 Mustafa Raja: filter the meetings, or, yeah, filter the meetings and fill up that column to add one of these departments to those meetings.
265 00:24:43.130 ⇒ 00:24:47.500 Mustafa Raja: so the meetings will show up in these department, right?
266 00:24:49.443 ⇒ 00:24:50.270 Uttam Kumaran: Yes, that’s
267 00:24:53.292 ⇒ 00:25:02.599 Uttam Kumaran: I think probably my only my only piece of feedback is that for the departments? I mean, I think we have to. We’ll have to meeting. Categorization
268 00:25:04.580 ⇒ 00:25:12.689 Uttam Kumaran: is like its own feature, right? Like right now, we’re only categorizing based on the the title but I think we will need to improve that.
269 00:25:12.990 ⇒ 00:25:13.750 Mustafa Raja: Yeah, yeah.
270 00:25:14.460 ⇒ 00:25:20.080 Uttam Kumaran: Like, cause, yeah. So what we can get basically improve like
271 00:25:20.490 ⇒ 00:25:26.160 Uttam Kumaran: any meeting that comes in, where does it go? And then if meetings aren’t able to get categorized. Where do they go?
272 00:25:26.350 ⇒ 00:25:29.369 Uttam Kumaran: We will also need to start screening one thing that I
273 00:25:29.550 ⇒ 00:25:38.649 Uttam Kumaran: I haven’t brought up, but we’ll start to. We’ll need to start screening meetings for like sensitive information, and probably just like auto, delete them or flag for deletion.
274 00:25:39.410 ⇒ 00:25:40.410 Sam Roberts: Hmm.
275 00:25:40.410 ⇒ 00:25:47.040 Uttam Kumaran: One of the pitfalls of reporting. Every meeting is that if we are to get into litigation
276 00:25:47.640 ⇒ 00:25:51.070 Uttam Kumaran: they can basically subpoena every single thing.
277 00:25:53.300 ⇒ 00:25:57.160 Uttam Kumaran: And so I just want to be careful that we can start to
278 00:25:57.330 ⇒ 00:26:10.399 Uttam Kumaran: like if there’s meetings where we talk about like if if there just are meetings that are more sensitive than others. Whether it’s about people plans like, I just want to be able to delete it, but not just delete it from super base like
279 00:26:10.600 ⇒ 00:26:19.319 Uttam Kumaran: we need to go scrub the recordings like just removing from S. 3 trigger deletion in zoom. So not. It’s not like a
280 00:26:19.590 ⇒ 00:26:23.940 Uttam Kumaran: again. Not like it’s like a medium priority. But that’s the
281 00:26:24.420 ⇒ 00:26:31.100 Uttam Kumaran: as a like, the the work that I’ve done to make sure that we can do this up recording stuff. That’s the real risk here.
282 00:26:32.690 ⇒ 00:26:41.520 Uttam Kumaran: is that you kind of have to. You know, if someone sues us, we have to, and we get need to produce evidence like we have to send every single thing because it’s reporting.
283 00:26:41.750 ⇒ 00:26:45.609 Uttam Kumaran: So I would rather us just be careful over time. So.
284 00:26:47.380 ⇒ 00:26:48.050 Mustafa Raja: Okay.
285 00:26:50.040 ⇒ 00:26:58.010 Sam Roberts: Yeah, along with that might even be as we start to add more features to the platform, like, does everyone need access to all meetings are there going to be meetings that are like.
286 00:26:58.010 ⇒ 00:26:58.820 Uttam Kumaran: Yes.
287 00:26:59.366 ⇒ 00:27:05.299 Sam Roberts: You know, restricted to departments, or, you know, to just the people in it. If it’s something about, you know.
288 00:27:05.450 ⇒ 00:27:07.940 Sam Roberts: Hr. Or something, you know something where you don’t need
289 00:27:08.100 ⇒ 00:27:18.015 Sam Roberts: everything everywhere. There’s definitely gonna have to be some, or I imagine there should be some amount of like permissions and and roles or things eventually. But
290 00:27:19.120 ⇒ 00:27:25.706 Sam Roberts: that’s a you know. Another thing that that along with this like liability stuff, you want to make sure that there was something there, I imagine.
291 00:27:26.290 ⇒ 00:27:26.660 Uttam Kumaran: Yes.
292 00:27:26.660 ⇒ 00:27:31.960 Sam Roberts: That’s a that’s a whole other like, you know, control flow thing that we’d have to sort out. So so
293 00:27:32.210 ⇒ 00:27:33.010 Sam Roberts: to do.
294 00:27:35.770 ⇒ 00:27:37.500 Sam Roberts: Yep, just on on that topic.
295 00:27:39.120 ⇒ 00:27:40.089 Mustafa Raja: I agree, yeah.
296 00:27:40.090 ⇒ 00:27:46.329 Uttam Kumaran: Yeah, I think we’ll we’ll end up with some like user personalization. And I think a lot of that should probably get sourced from
297 00:27:48.350 ⇒ 00:28:03.210 Uttam Kumaran: Google, like operations will own, like this person is in this department, and then there will be a slew of permissions that can get granted, you know. I would rather not have us manage that. And it just get managed as part of our Google Iam stuff.
298 00:28:03.400 ⇒ 00:28:04.719 Uttam Kumaran: I think. But again, it’s.
299 00:28:04.720 ⇒ 00:28:06.539 Sam Roberts: Total sense. Yeah, yeah.
300 00:28:06.540 ⇒ 00:28:07.050 Uttam Kumaran: That’s the source.
301 00:28:07.050 ⇒ 00:28:08.100 Sam Roberts: Shoot for that. That’s perfect.
302 00:28:08.100 ⇒ 00:28:17.550 Uttam Kumaran: Yeah, exactly. And perfect opportunity. Like, if we had a cross functional meeting, you could tell Rico, okay, you need to make sure that everybody’s in the right departments, and that we have a clarity on
303 00:28:17.750 ⇒ 00:28:22.020 Uttam Kumaran: when people move teams like what they access to, because then we can infer that. And then
304 00:28:22.300 ⇒ 00:28:27.039 Uttam Kumaran: it’s not on our team to manage that, but I don’t wanna do off like at all.
305 00:28:27.040 ⇒ 00:28:36.130 Sam Roberts: No, no, but it would definitely be like pulling that. And then, you know, whatever is important would get. Yeah, there’d be have to be a little bit in the platform that’s just reading from there. But yeah, not not
306 00:28:36.730 ⇒ 00:28:38.579 Sam Roberts: controlling it from there necessarily.
307 00:28:39.190 ⇒ 00:28:39.810 Uttam Kumaran: Yeah.
308 00:28:40.330 ⇒ 00:28:51.210 Uttam Kumaran: And then I guess my question, Sam, just because I’m starting to like, I, I feel like we have a big backlog of ideas, but just curious, like, what? What kind of a back end work.
309 00:28:51.370 ⇒ 00:28:57.850 Uttam Kumaran: or any like full stack work did you guys do at your last shop like what were, what was the stack that you guys were on? And
310 00:28:58.080 ⇒ 00:29:05.290 Uttam Kumaran: just like, you know, interested to think about that even also like, what? What if there’s any interesting like front end
311 00:29:05.747 ⇒ 00:29:11.209 Uttam Kumaran: patterns that you also develop just like. So I can get my mind going. So I’m thinking about ideas.
312 00:29:11.410 ⇒ 00:29:13.170 Sam Roberts: Yeah. I think we.
313 00:29:13.380 ⇒ 00:29:23.177 Sam Roberts: This might have come up, I think, both when you were late to the call the other day I was talking a little bit about the platform moving. My my last stack was basically a next app with the super based back end
314 00:29:23.580 ⇒ 00:29:40.906 Sam Roberts: which was perfect for kind of moving fast where, you know, you can add backend features with the next app, because you can have Api calls and things. But you’re not necessarily worrying about all of the logic super base is handling a decent chunk of that
315 00:29:41.640 ⇒ 00:29:48.004 Sam Roberts: and that was for kind of me as like the solo dev with my designer partner. So it was. It was a really good system for that.
316 00:29:49.310 ⇒ 00:30:01.739 Sam Roberts: that being said like, I was also doing a lot of like custom run in stuff because he was a kind of intense designer and so really wanted things like a very particular way. But there are. We mentioned a couple of tools on the call the other day, like Chat Cn.
317 00:30:02.533 ⇒ 00:30:14.810 Sam Roberts: that are kind of like good things that we can just drop in we can obviously customize. But we don’t necessarily need to rebuild every component all the time. Yeah, I think we’re using
318 00:30:15.490 ⇒ 00:30:16.820 Sam Roberts: mui right now.
319 00:30:16.820 ⇒ 00:30:17.350 Mustafa Raja: Yeah.
320 00:30:17.350 ⇒ 00:30:18.722 Sam Roberts: Or some things.
321 00:30:19.410 ⇒ 00:30:26.219 Sam Roberts: which is, is, you know, as long as we have something that’s good. But, like we might want to think a little bit about standardizing some of that stuff as we add more things.
322 00:30:27.380 ⇒ 00:30:34.589 Sam Roberts: the other thing I was thinking about. Yeah, it was just moving, because, like, I said. Right now. I I hopped in this a few times. Now the separate back end front end thing
323 00:30:35.410 ⇒ 00:30:55.210 Sam Roberts: is gonna slow down, just deploying and testing and all that other stuff. So like, I said, I was probably gonna put together a little bit of a plan for like migrating to at least one repo, if not a next step. Specifically, if not, maybe a few other things that might be helpful for us to just move a little faster and not have to worry about
324 00:30:55.953 ⇒ 00:30:57.140 Sam Roberts: you know
325 00:30:57.640 ⇒ 00:31:11.385 Sam Roberts: the kind of connection, and making sure we’re hitting the right endpoint. There’s there’s a few things we can do that might kind of streamline, especially for an internal tool where it’s not necessarily you know, gonna be serving thousands of users or something. We can do things a little bit
326 00:31:12.880 ⇒ 00:31:15.430 Sam Roberts: and I don’t wanna say quick and dirty, but you know
327 00:31:15.600 ⇒ 00:31:20.287 Sam Roberts: a a quick, maybe not like infinitely scalable way, but good enough for us right now.
328 00:31:20.740 ⇒ 00:31:25.269 Sam Roberts: which is often my like 1st pass at something, anyway.
329 00:31:25.860 ⇒ 00:31:29.069 Sam Roberts: But I think something like a next step with maybe
330 00:31:29.170 ⇒ 00:31:33.915 Sam Roberts: I mean, we’re already using super base and for a few different parts.
331 00:31:35.660 ⇒ 00:31:39.170 Sam Roberts: I think we’re pretty set up to just migrate over to something like that.
332 00:31:39.170 ⇒ 00:31:39.810 Mustafa Raja: Yeah.
333 00:31:42.380 ⇒ 00:32:01.690 Sam Roberts: I I haven’t done a ton of digging through the back end. I did a little bit. Just kind of bouncing back and forth between, like what’s getting calls from the front end to see the back end. But I think it it will. It won’t be very hard to actually make that happen. It’ll it’ll be work. It won’t be crazy. But there’s also like a balance between, like
334 00:32:02.180 ⇒ 00:32:09.820 Sam Roberts: how vibe coded versus how, you know, non coded things are and
335 00:32:10.060 ⇒ 00:32:14.560 Sam Roberts: I’m still kind of wrestling with that myself, just in terms of like long term technical debt.
336 00:32:14.560 ⇒ 00:32:15.080 Uttam Kumaran: Yeah.
337 00:32:15.080 ⇒ 00:32:20.280 Sam Roberts: Of things like. I’ve put in some side projects that are great, and I’ve done some that I’ve had to abandon.
338 00:32:20.280 ⇒ 00:32:31.500 Uttam Kumaran: Well, but here, but here’s like here’s where I would. Here, here’s here, where where, I would say we can find a balance is one like we should develop like a cursor. Rules for each core repo for us.
339 00:32:32.465 ⇒ 00:32:33.070 Sam Roberts: Yep.
340 00:32:33.230 ⇒ 00:32:33.750 Sam Roberts: I added that.
341 00:32:33.750 ⇒ 00:32:34.550 Uttam Kumaran: Yeah.
342 00:32:34.550 ⇒ 00:32:42.529 Sam Roberts: Some. I think the interesting admin thing the other earlier, because I was there. Cursor rules maybe starting to use some like memory bank.
343 00:32:42.970 ⇒ 00:32:47.689 Uttam Kumaran: Yes, plan because then that way, if people are jumping in and out of repos.
344 00:32:48.249 ⇒ 00:32:52.040 Sam Roberts: You know. Cursor will know what’s going on, at least
345 00:32:52.870 ⇒ 00:33:03.830 Uttam Kumaran: Yeah. And second, also, like, I think you could, we could also start to set up like notion Mcp, or something like when or linear Mcp, like when you’re when we’re one thing I want this team to basically do is like.
346 00:33:04.170 ⇒ 00:33:16.120 Uttam Kumaran: be on the edge of like, what’s the best way to Vibe code because the other teams are gonna all all want. This similar stuff is like, Okay, you have it. Let’s say you have a linear ticket associated with you. How do you.
347 00:33:16.460 ⇒ 00:33:23.799 Uttam Kumaran: when you open up cursor? How do you make sure that ticket gets into contact? Any relevant docs get into context and that it can take the 1st draft
348 00:33:25.230 ⇒ 00:33:38.460 Uttam Kumaran: and then the last piece is like, how do we do aipr review? So I there’s a backlog ticket about choosing a vendor, for that, like reptile is one. There’s graphite. There’s
349 00:33:38.630 ⇒ 00:33:52.329 Uttam Kumaran: get up copilot cursor now has one we would just do a spike on it. And then this team can decide. It’s they’re anywhere from like 20 to 40 bucks a person per month. So it’s not like cheap.
350 00:33:52.620 ⇒ 00:33:57.049 Uttam Kumaran: But I also think that if we pick one that is like very
351 00:33:58.190 ⇒ 00:34:10.079 Uttam Kumaran: be tunable, then that’s another step where we could start to enforce standards, and like, you can have the Pr like, you basically can vibe code something. And then the AI Pr view can like compete against that. You know.
352 00:34:10.310 ⇒ 00:34:11.420 Sam Roberts: But there’s like, yeah.
353 00:34:11.590 ⇒ 00:34:13.019 Uttam Kumaran: There’s some balance.
354 00:34:13.440 ⇒ 00:34:21.510 Sam Roberts: Okay, no, that’s that’s that’s some good. I gotta think a little broader with some of the stuff, because I I’m still like in like product mode sometimes, and like building a thing. And like you’re.
355 00:34:21.510 ⇒ 00:34:37.460 Uttam Kumaran: Yeah. And and also these are these are things that like, we’re getting a little bit Meta, right? It’s like, we’re the guinea pigs. But also any opportunity we have to leverage. AI. We should consider so the entire one thing that I think, Sam, me and you will do start to decompose.
356 00:34:37.540 ⇒ 00:34:55.940 Uttam Kumaran: like one of the things that I’ve always wanted to do is look at like, go through linear basically find use. A tell us, like, what are all the common development, like patterns like, okay, create a new data table, create a new create a new like endpoint. We break those down and we can start to have sops on like
357 00:34:56.090 ⇒ 00:35:04.549 Uttam Kumaran: the best way to create a new endpoint, using cursor or the best way to write a new Dbp model using cursor. And so
358 00:35:04.660 ⇒ 00:35:07.359 Uttam Kumaran: each workflow can have like.
359 00:35:07.790 ⇒ 00:35:21.120 Uttam Kumaran: how is it, AI augmented? Because ultimately, you know, if I was to tell you where the future is for us is, there’s probably 20 to 40% of tickets that an autonomous agent can just do on its own with with very little help.
360 00:35:21.420 ⇒ 00:35:22.260 Uttam Kumaran: Right?
361 00:35:22.520 ⇒ 00:35:25.390 Uttam Kumaran: So my my goal is that at some point
362 00:35:25.860 ⇒ 00:35:28.929 Uttam Kumaran: the limiting factor is actually how good the ticket is.
363 00:35:29.515 ⇒ 00:35:41.229 Uttam Kumaran: And then it gets handed to AI to take the 1st crack versus, and then the human can come in like second to review and take it from there. But a couple of things have to happen. We have to be
364 00:35:41.420 ⇒ 00:35:50.180 Uttam Kumaran: pretty we have to be. We have to make sure we have great ticket quality, which is sort of the work that I I asked Miguel to start, which was the grooming stuff.
365 00:35:50.340 ⇒ 00:36:02.720 Uttam Kumaran: So we need to make sure that there is a basically a minimum threshold for tickets. Second, is, we need some mechanism of identifying the type of work and then kicking off a series of steps.
366 00:36:02.840 ⇒ 00:36:04.900 Uttam Kumaran: The thing for us is like we
367 00:36:05.600 ⇒ 00:36:18.820 Uttam Kumaran: we. I don’t want to use like a I’m not interested in sort of just like throwing this at an agent and saying doing it like I think we should do. The the legwork of here are the 5 to 10 workflows that we trust AI to do first, st
368 00:36:18.930 ⇒ 00:36:27.789 Uttam Kumaran: and then we sort of work from there. But this is, you know this is sort of the long term vision of like. How right now we see in our company, AI is sort of penetrated
369 00:36:27.930 ⇒ 00:36:31.849 Uttam Kumaran: sales operations, but engineering is the
370 00:36:32.100 ⇒ 00:36:38.114 Uttam Kumaran: kind of the last bastion, but also the biggest cost center, the biggest opportunity for us.
371 00:36:38.490 ⇒ 00:36:39.030 Sam Roberts: Yeah.
372 00:36:39.200 ⇒ 00:36:40.349 Uttam Kumaran: Right? So that’s
373 00:36:41.700 ⇒ 00:36:53.760 Uttam Kumaran: but all of these platform things help to go solve that that problem over time. But it will require some deeper thinking. And and for us to plan out like, what are the top 20% of like things that common
374 00:36:54.070 ⇒ 00:36:57.659 Uttam Kumaran: engineering tasks that we have. And out of these like, what can we
375 00:36:57.900 ⇒ 00:37:00.980 Uttam Kumaran: begin to do? But training is also going to be a big piece.
376 00:37:01.710 ⇒ 00:37:04.409 Uttam Kumaran: you know, how do we get everyone to be familiar with cursor.
377 00:37:04.530 ⇒ 00:37:06.540 Uttam Kumaran: Yeah. And things like that.
378 00:37:07.520 ⇒ 00:37:11.479 Uttam Kumaran: Yeah, I think we’ll probably be assessing that later, with the way, give us all that on the.
379 00:37:11.630 ⇒ 00:37:16.959 Sam Roberts: You guys talked about the other day. I think I’m curious to a little more about how other teams are using it, too. So yeah, that’s great.
380 00:37:19.023 ⇒ 00:37:25.340 Sam Roberts: Yeah, no. I like this kind of meta, thinking, it’s it’s helpful for me to wrap my head around kind of where things are going, just to.
381 00:37:27.070 ⇒ 00:37:28.130 Uttam Kumaran: Perfect, great.
382 00:37:28.610 ⇒ 00:37:29.300 Mustafa Raja: Yeah.
383 00:37:32.850 ⇒ 00:37:43.939 Uttam Kumaran: For the other tickets. I think we’re gonna talk through default today and then interlude. I created some. Follow up tickets, Mustafa, so you could take a look at those
384 00:37:44.496 ⇒ 00:37:47.949 Uttam Kumaran: I’m gonna add notes. I think we should just probably do a working session.
385 00:37:49.150 ⇒ 00:37:52.220 Uttam Kumaran: Yeah, maybe an hour or 2. Yeah, tomorrow or later today.
386 00:37:52.220 ⇒ 00:37:53.730 Mustafa Raja: Yeah, yeah, I love that.
387 00:37:55.030 ⇒ 00:37:58.289 Uttam Kumaran: I think it’ll we can walk through a bunch of stuff a lot here.
388 00:37:58.790 ⇒ 00:38:05.470 Mustafa Raja: Yeah, yeah, I guess. I guess that’s all. Then for this.
389 00:38:05.640 ⇒ 00:38:13.649 Uttam Kumaran: Yeah, I think the I think the only thing, Casey. What is there anything left on the ABC. I think you went through it a little bit, but I just want to confirm like
390 00:38:14.130 ⇒ 00:38:19.744 Uttam Kumaran: what I I’m gonna just send them a note with the fixes that we made
391 00:38:20.410 ⇒ 00:38:29.619 Uttam Kumaran: And then the stuff we need to do with the sheets still stands like, are you gonna be? Are you just applying that that same logic column by column? Or is there anything else.
392 00:38:31.200 ⇒ 00:38:37.539 Casie Aviles: Yeah, pretty much. I’m just pasting all the formulas and just making sure that, you know, since the sources are
393 00:38:38.250 ⇒ 00:38:42.370 Casie Aviles: kind of separate, I’m just making sure that I’m drawing from the right tables
394 00:38:42.740 ⇒ 00:38:45.170 Casie Aviles: and that the the data is updated.
395 00:38:47.650 ⇒ 00:38:53.510 Casie Aviles: But yeah, I I did the tests already. And I also filled in the filled out the document.
396 00:38:56.180 ⇒ 00:38:58.030 Uttam Kumaran: Okay, okay, great.
397 00:39:00.030 ⇒ 00:39:02.320 Uttam Kumaran: Okay, perfect. Alright guys.
398 00:39:02.610 ⇒ 00:39:09.499 Uttam Kumaran: thank you. We also have this. Engineering AI thing meeting later. Is a way, is a wish leading that?
399 00:39:09.700 ⇒ 00:39:10.470 Uttam Kumaran: Or do you know the.
400 00:39:10.974 ⇒ 00:39:14.000 Casie Aviles: Aisha asked me to host it.
401 00:39:16.070 ⇒ 00:39:16.690 Uttam Kumaran: Great. Okay.
402 00:39:17.380 ⇒ 00:39:20.170 Sam Roberts: Yeah, actually be interesting.
403 00:39:20.890 ⇒ 00:39:22.420 Sam Roberts: Yeah, as long as you have welcome.
404 00:39:22.640 ⇒ 00:39:28.939 Uttam Kumaran: Yeah. And as long as you have a plan, Casey, or if you want to create a plan before that meeting, and we can all add comments
405 00:39:29.320 ⇒ 00:39:31.150 Uttam Kumaran: that may be the best.
406 00:39:34.480 ⇒ 00:39:38.780 Casie Aviles: It’s going to be in an hour. I do have like a rough plan.
407 00:39:40.510 ⇒ 00:39:41.450 Uttam Kumaran: Okay, okay.
408 00:39:41.770 ⇒ 00:39:47.900 Uttam Kumaran: I think it’s something we’re gonna do more often. So if you’d want to throw that in notion, because then we can kind of create a running meeting, log.
409 00:39:50.420 ⇒ 00:39:51.100 Casie Aviles: Sure.
410 00:39:55.670 ⇒ 00:39:57.230 Uttam Kumaran: Okay, perfect. Thanks.
411 00:39:57.230 ⇒ 00:39:58.830 Mustafa Raja: Thank you. Thank you.
412 00:39:58.830 ⇒ 00:39:59.530 Sam Roberts: Yeah, later.