Meeting Title: Brainforge Team Proof of Concept Sync Date: 2026-04-15 Meeting participants: Miranda Wen, Pranav Narahari
WEBVTT
1 00:01:54.890 ⇒ 00:01:55.890 Pranav Narahari: Hey, Miranda.
2 00:01:58.810 ⇒ 00:02:00.480 Miranda Wen: Hi! Hi, Brenda.
3 00:02:01.360 ⇒ 00:02:02.399 Pranav Narahari: How’s it going?
4 00:02:02.620 ⇒ 00:02:11.499 Miranda Wen: I’ve been good. Yeah, I was, like, looking into the premium model, is it because, like, your curse is, like, a premium, like, or higher tier?
5 00:02:11.850 ⇒ 00:02:15.250 Miranda Wen: So I don’t have, like, premium model in my setting.
6 00:02:15.860 ⇒ 00:02:18.190 Pranav Narahari: Really? Okay.
7 00:02:18.190 ⇒ 00:02:21.590 Miranda Wen: I can share a screen. I will ask them, yeah.
8 00:02:31.560 ⇒ 00:02:33.809 Pranav Narahari: I’m not seeing… are you sharing your screen right now?
9 00:02:33.810 ⇒ 00:02:35.089 Miranda Wen: I’m, I’m sharing, yeah.
10 00:02:35.090 ⇒ 00:02:39.390 Pranav Narahari: Oh, okay, okay. Yeah, because I’m having some issues with Zoom, that’s why. Okay, perfect, yeah, I can see now.
11 00:02:40.170 ⇒ 00:02:43.209 Miranda Wen: Yeah, so I… I went to the app models.
12 00:02:44.260 ⇒ 00:02:48.480 Miranda Wen: And it literally went through, like, all of them, I didn’t see it, and even it went.
13 00:02:49.870 ⇒ 00:02:51.930 Miranda Wen: It’s just not showing up.
14 00:02:53.100 ⇒ 00:03:01.600 Pranav Narahari: Interesting. Yeah, like, cursor’s kind of been buggy for me, too, like, I haven’t been able to even use the auto model.
15 00:03:01.600 ⇒ 00:03:02.760 Miranda Wen: Oh, really?
16 00:03:02.760 ⇒ 00:03:03.560 Pranav Narahari: Yeah.
17 00:03:03.740 ⇒ 00:03:11.329 Pranav Narahari: So, okay, I’m… I can definitely sit on my end. I can maybe share my screen.
18 00:03:11.780 ⇒ 00:03:12.440 Miranda Wen: Alright, cool.
19 00:03:14.660 ⇒ 00:03:19.319 Miranda Wen: So for this kind of issue, like, should I create, like, a linear ticket, or…
20 00:03:21.490 ⇒ 00:03:24.780 Pranav Narahari: For this type of issue, yeah, it’s a good question.
21 00:03:24.780 ⇒ 00:03:27.959 Miranda Wen: Or should I ask them platform or Springforce team?
22 00:03:29.010 ⇒ 00:03:45.209 Pranav Narahari: Yeah, I would just ask him platform, and then maybe even… that’s… it’s honestly a good question. I think there should probably be tickets assigned to that, because there’s just gonna be a lot more usage, there’s probably gonna be, like, more bugs that are being found, as, like, in the next coming weeks, so…
23 00:03:45.530 ⇒ 00:03:53.920 Pranav Narahari: Yeah, maybe just, like, as part of your message, you can just ask them, hey, is there a… should I create a request for this somewhere?
24 00:03:53.920 ⇒ 00:03:55.340 Miranda Wen: Mmm, I see, okay.
25 00:03:55.500 ⇒ 00:03:57.440 Pranav Narahari: I do think that’s a better process, yeah.
26 00:03:57.440 ⇒ 00:03:58.489 Miranda Wen: Cool, cool, cool.
27 00:04:07.330 ⇒ 00:04:09.679 Pranav Narahari: One sec… I just started using…
28 00:04:09.970 ⇒ 00:04:16.060 Pranav Narahari: Zoom desktop, so it’s, like, asking me to, like, update my permissions and everything.
29 00:04:16.380 ⇒ 00:04:17.610 Miranda Wen: No worries.
30 00:04:19.790 ⇒ 00:04:22.469 Pranav Narahari: Okay, it’s ask… okay, never mind.
31 00:04:22.610 ⇒ 00:04:25.210 Pranav Narahari: Let me just… it’s asking me to, like, restart.
32 00:04:25.830 ⇒ 00:04:27.529 Pranav Narahari: I’ll just send you a screenshot.
33 00:04:36.450 ⇒ 00:04:38.740 Pranav Narahari: Yeah, I just sent you a screenshot on Slack.
34 00:04:47.820 ⇒ 00:04:50.440 Miranda Wen: Oh, I see it, it’s, like, on top for you.
35 00:04:50.970 ⇒ 00:05:01.179 Pranav Narahari: Yeah, not sure why, yeah, for you, on, like, the top toggles, like, you have two other ones, and I only have max mode.
36 00:05:02.140 ⇒ 00:05:03.050 Miranda Wen: Yeah.
37 00:05:03.050 ⇒ 00:05:03.700 Pranav Narahari: Yeah.
38 00:05:04.970 ⇒ 00:05:10.429 Pranav Narahari: Anyways, I’m also probably gonna start using open code for most of my stuff.
39 00:05:10.570 ⇒ 00:05:11.230 Pranav Narahari: So…
40 00:05:11.230 ⇒ 00:05:12.270 Miranda Wen: Oh, okay.
41 00:05:12.270 ⇒ 00:05:15.150 Pranav Narahari: Yeah, and yeah, I mean…
42 00:05:15.400 ⇒ 00:05:22.770 Pranav Narahari: I think you… like, that, research document that you put together was, like, honestly super informative, like, I learned a lot of stuff there.
43 00:05:24.080 ⇒ 00:05:33.520 Pranav Narahari: I think, if you just want to do, like, the same… my only concern, right, is, like, with premium, it’s saying it’s doing a lot of things under the hood,
44 00:05:33.970 ⇒ 00:05:39.959 Pranav Narahari: you know, as you saw with the composer, like, that’s kind of cap, right?
45 00:05:39.960 ⇒ 00:05:40.280 Miranda Wen: Yeah.
46 00:05:40.280 ⇒ 00:05:45.919 Pranav Narahari: They’re just kind of saying that as a marketing thing, and then it’s really just, using an open weight model.
47 00:05:46.240 ⇒ 00:05:50.800 Miranda Wen: Yeah. I think they’ll be lost to Serana, for sure.
48 00:05:50.800 ⇒ 00:05:53.430 Pranav Narahari: Yeah, yeah. So…
49 00:05:53.580 ⇒ 00:06:06.439 Pranav Narahari: I think what… I think what we should just do, just to do our due diligence, and I talked to Bea about this for, like, our client projects as well, is when we’re coming up with…
50 00:06:07.350 ⇒ 00:06:17.929 Pranav Narahari: we should just do more proof of concepts and evaluations, just to, like, prove our point. And I think this is, like, a great example of, like, where we could do that.
51 00:06:18.170 ⇒ 00:06:19.680 Pranav Narahari: there’s certain…
52 00:06:19.850 ⇒ 00:06:28.640 Pranav Narahari: skills, right, for example, that are getting a lot of usage. Let’s probably just, like, go with the top 10 of those, run them with both the…
53 00:06:28.930 ⇒ 00:06:36.350 Pranav Narahari: API, as well as just Premium versus Composer 2.
54 00:06:36.650 ⇒ 00:06:42.720 Pranav Narahari: And then you can use, like, LLM as a judge, just to, like, confirm that all three of them are basically the same.
55 00:06:43.270 ⇒ 00:06:44.320 Miranda Wen: Mmm.
56 00:06:44.320 ⇒ 00:06:44.990 Pranav Narahari: Yeah.
57 00:06:44.990 ⇒ 00:06:51.790 Miranda Wen: I’m curious, like, if you had experience on, like, when we use OM to judge, like, which OM should we use?
58 00:06:53.070 ⇒ 00:06:55.329 Pranav Narahari: I would say probably just, like.
59 00:06:58.180 ⇒ 00:07:02.929 Pranav Narahari: That’s a good question. I think for this, since it’s just, like, a one-time use, you wanna…
60 00:07:02.940 ⇒ 00:07:10.290 Pranav Narahari: you don’t really care about execution time, like, if it takes longer, but it’s gonna be more thorough, that’s better.
61 00:07:10.300 ⇒ 00:07:24.939 Pranav Narahari: Also, since you’re comparing the output of… from 3 different models, you may want to have… you need to make sure you have, like, a large enough context window, which shouldn’t be an issue, honestly.
62 00:07:26.670 ⇒ 00:07:34.950 Pranav Narahari: Yeah, there’s… if you do, like, a quick search, I bet it’ll give you, like, some advice on which model to choose. Probably, like, some…
63 00:07:36.170 ⇒ 00:07:41.679 Pranav Narahari: like, either, like, Claude, Sonnet, or, or Opus, potentially.
64 00:07:41.860 ⇒ 00:07:44.160 Miranda Wen: Mmm, okay, okay, yeah, we’ll look into that.
65 00:07:44.350 ⇒ 00:07:45.459 Miranda Wen: Cool, yeah.
66 00:07:49.070 ⇒ 00:07:54.769 Miranda Wen: Yeah, I feel that… yeah, I feel that’s definitely, like, a proof of concept we can easily implement.
67 00:07:55.360 ⇒ 00:08:04.010 Pranav Narahari: Yeah, and I think with Cursor, you should be able to build that script with no problem. Let me know if you have any issues with that, but it should be pretty smooth.
68 00:08:04.520 ⇒ 00:08:05.990 Miranda Wen: Okay, gotcha.
69 00:08:06.170 ⇒ 00:08:07.480 Pranav Narahari: Yeah.
70 00:08:07.630 ⇒ 00:08:08.340 Pranav Narahari: Yeah.
71 00:08:08.340 ⇒ 00:08:09.909 Miranda Wen: Yeah, you go ahead, you go ahead.
72 00:08:10.320 ⇒ 00:08:25.700 Pranav Narahari: I just… I noticed, too, I didn’t get to them yet, but you sent, like, two other links for, like, the refined project plan, and then just, like, I think the other doc was just, like, all the comments that, you’ve, you’ve addressed, so… yeah, I’ll take a look at that as well, and maybe…
73 00:08:25.740 ⇒ 00:08:31.059 Pranav Narahari: If you want, maybe tomorrow we can have, like, another, like, quick call to just go over anything, and…
74 00:08:31.240 ⇒ 00:08:36.600 Pranav Narahari: if we even need the call, I’ll probably just try to, like, give you that, feedback async as well.
75 00:08:36.789 ⇒ 00:08:40.679 Miranda Wen: Okay, okay, yeah, yeah, yeah. Toshin Common works, for sure, yeah.
76 00:08:41.000 ⇒ 00:08:42.209 Pranav Narahari: Cool. Cool, cool.
77 00:08:43.150 ⇒ 00:08:50.349 Miranda Wen: Okay, cool, thank you so much for now, like, this is really helpful. Oh, sorry, another question is, like, right now I was thinking, like.
78 00:08:50.720 ⇒ 00:09:00.179 Miranda Wen: Like, first, do you have, like, do you think, like, how many… number of times we should run, or how many skills we should run when we are doing this, like, proof of concept would be, like…
79 00:09:00.350 ⇒ 00:09:04.670 Miranda Wen: Like, legit, like, we can justify it, we’re enough to justify that.
80 00:09:05.250 ⇒ 00:09:07.819 Pranav Narahari: Yeah, that’s a good question.
81 00:09:12.190 ⇒ 00:09:13.349 Pranav Narahari: I, like…
82 00:09:14.320 ⇒ 00:09:21.219 Pranav Narahari: to be honest, maybe, like, 2 or 3 per skill, right? And I feel like that’s really thorough, like…
83 00:09:21.220 ⇒ 00:09:21.600 Miranda Wen: Yeah.
84 00:09:21.600 ⇒ 00:09:26.249 Pranav Narahari: even if you just did it once per skill, like, that’s really good data.
85 00:09:28.160 ⇒ 00:09:40.500 Pranav Narahari: what you could… what I think is also interesting is you can ask Cursor, or… yeah, probably cursor, which skills, like, rank these skills based on, like,
86 00:09:41.250 ⇒ 00:09:42.660 Pranav Narahari: complexity.
87 00:09:43.160 ⇒ 00:09:44.240 Miranda Wen: Mmm, no.
88 00:09:44.420 ⇒ 00:09:45.180 Pranav Narahari: So, like…
89 00:09:45.420 ⇒ 00:09:53.400 Pranav Narahari: You’re, you know, you can get a couple, like, you can use a couple of the skills from the low complexity, bucket, a few from the medium, a few from the high.
90 00:09:54.680 ⇒ 00:09:57.759 Pranav Narahari: I think that’ll be really good to see as well.
91 00:09:58.320 ⇒ 00:10:01.460 Miranda Wen: Gotcha, got you. Cool, cool, cool. Yeah.
92 00:10:01.570 ⇒ 00:10:08.290 Miranda Wen: Cool, I think that’s, like, really, really helpful. Yeah, I should… yeah, I will learn from Cursor on, like, how to implement that.
93 00:10:08.920 ⇒ 00:10:09.620 Miranda Wen: Cool.
94 00:10:09.620 ⇒ 00:10:10.260 Pranav Narahari: Yeah.
95 00:10:10.720 ⇒ 00:10:15.160 Miranda Wen: Yeah, then… Yeah.
96 00:10:15.160 ⇒ 00:10:24.619 Pranav Narahari: Also, talking to B about which skills are getting the most amount of usage is gonna be helpful as well, because I think at the end of the day, right, like, Utam…
97 00:10:24.850 ⇒ 00:10:29.889 Miranda Wen: The thing, I asked B this question before, and B told me, like, they couldn’t track, like, which.
98 00:10:29.890 ⇒ 00:10:30.740 Pranav Narahari: Oh…
99 00:10:30.740 ⇒ 00:10:36.220 Miranda Wen: They can only track, like, which model being used the most, but that’s all they have, the data.
100 00:10:36.630 ⇒ 00:10:37.829 Pranav Narahari: Oh, that’s annoying.
101 00:10:37.830 ⇒ 00:10:38.470 Miranda Wen: Yeah.
102 00:10:38.470 ⇒ 00:10:42.000 Pranav Narahari: Telling me about how that doesn’t track that much, so… okay.
103 00:10:44.680 ⇒ 00:10:56.769 Miranda Wen: Okay, cool, yeah, but I think this is, like, really helpful. Cool. And, yeah, I would just, like, put a placeholder, maybe the same time tomorrow this time, but if you think non-aster, just let me know, yeah.
104 00:10:57.010 ⇒ 00:11:11.709 Pranav Narahari: Yeah, we can have that meeting. Maybe also on the usage thing, if you want to just, like, create a quick, like, poll in Slack, or something for just, like, all the skills, I don’t know, actually, there may be too many skills, I don’t know.
105 00:11:12.140 ⇒ 00:11:14.630 Miranda Wen: Oh, yeah, just I have, like, a list of skills.
106 00:11:15.040 ⇒ 00:11:27.800 Pranav Narahari: Yeah, if there’s… and then you can just, like, have people react to the ones that they’re using. If there’s too many skills, there may be, like, you know, over, like, 100 or something. If that’s the case, then maybe you can just, like, start a thread, and then, you know…
107 00:11:28.080 ⇒ 00:11:34.709 Pranav Narahari: People can just, like, write the skills that they’re using, and then in that thread, and then you can do evals specifically on those as well.
108 00:11:35.230 ⇒ 00:11:39.380 Miranda Wen: Gotcha, gotcha. Oh, okay, that sounds like a good idea, yeah, I haven’t thought about that, yeah.
109 00:11:39.510 ⇒ 00:11:40.260 Pranav Narahari: Yeah.
110 00:11:41.030 ⇒ 00:11:45.329 Miranda Wen: Yeah, for that, it should be done in the team, right? The team channel.
111 00:11:46.560 ⇒ 00:11:49.340 Pranav Narahari: Yeah, probably Brainforge Team is the best place for that.
112 00:11:49.340 ⇒ 00:11:52.539 Miranda Wen: Okay, cool, cool, cool. Okay, thank you so much for now, it’s really helpful.
113 00:11:52.540 ⇒ 00:12:00.319 Pranav Narahari: Yeah, totally. Cool, we can, yeah, let me know if you need anything else, but I also wanted to ask you, well, this is your third week now, right?
114 00:12:00.320 ⇒ 00:12:01.630 Miranda Wen: Yes, yes.
115 00:12:01.630 ⇒ 00:12:04.370 Pranav Narahari: Cool. And how are things going overall?
116 00:12:04.770 ⇒ 00:12:23.389 Miranda Wen: Yeah, I think it’s been going pretty well. I think I do, like, collective feedback from, Uten before, like, he was not really happy with my async communication skills, so I’m, like, trying to work on that and be more active in platform in general.
117 00:12:23.390 ⇒ 00:12:24.090 Pranav Narahari: Okay.
118 00:12:24.280 ⇒ 00:12:36.669 Pranav Narahari: Cool. Yeah, it takes a minute, I don’t… you know, like… but, you know, with that project plan, I think I was, like, super impressed. I think also, like, with the research that you just did, too, like.
119 00:12:36.730 ⇒ 00:12:45.750 Pranav Narahari: I think that’s the type of work that Utam really appreciates, everybody else really appreciates. And then, yeah, just…
120 00:12:45.860 ⇒ 00:12:51.110 Pranav Narahari: probably, like, with acing skills, it’s kind of just, like, speed to response, probably.
121 00:12:51.380 ⇒ 00:12:56.729 Pranav Narahari: And then also, like, effectively just describing everything that you’re working on.
122 00:12:57.090 ⇒ 00:12:57.750 Miranda Wen: Yeah.
123 00:12:58.000 ⇒ 00:13:05.290 Pranav Narahari: Yeah, but, you know, so first couple weeks, you’re just kind of learning the ropes, like, about how we do things here, right, specifically, so…
124 00:13:05.510 ⇒ 00:13:09.469 Pranav Narahari: the… I think that’s… that’s actually good feedback that he’s… that he’s giving you.
125 00:13:09.900 ⇒ 00:13:12.580 Miranda Wen: Yes, yes, definitely. Yeah.
126 00:13:12.970 ⇒ 00:13:26.419 Miranda Wen: Yeah, I kind of assume, like, by default, like, everything I post on the platform, like, even if he didn’t react home or anything, like, he’d probably, like, take a look of everything, or should I, like, promote, like, my visible… my work visibility better?
127 00:13:28.030 ⇒ 00:13:33.569 Pranav Narahari: Yeah, so… Mmm, I noticed, like, you’re… you tagged me.
128 00:13:34.230 ⇒ 00:13:41.540 Pranav Narahari: And I don’t know in the past if you had been tagging people, but I always just tag people, even if it’s just, like, overkill.
129 00:13:42.390 ⇒ 00:13:44.129 Pranav Narahari: Yeah, I try to do that.
130 00:13:44.480 ⇒ 00:13:52.559 Pranav Narahari: Even… so, I know Utam, he mutes chats, right? Because there’s no way for him to, like…
131 00:13:52.760 ⇒ 00:14:08.229 Pranav Narahari: because he’s in every single channel, there’s no way for him to actually effectively, like, look into every single small thing to, like, get his own updates. And so he probably has it on a certain, like, notification level, where, like, if he is added on a certain thing, then…
132 00:14:08.640 ⇒ 00:14:11.179 Pranav Narahari: he will look into it.
133 00:14:11.520 ⇒ 00:14:27.870 Pranav Narahari: Now, he also will unmute chats, maybe, like, say if things aren’t… like, he wants to keep, like, closer… closer visibility, like, he’ll just unmute it. But I think if you’re like, Uten needs to see this, don’t just put it in a channel that he’s in, also tag him.
134 00:14:28.130 ⇒ 00:14:30.129 Miranda Wen: Okay, okay.
135 00:14:30.350 ⇒ 00:14:30.880 Miranda Wen: Yeah.
136 00:14:30.880 ⇒ 00:14:32.479 Pranav Narahari: Does that kind of answer your question, or are you, like.
137 00:14:32.480 ⇒ 00:14:34.840 Miranda Wen: Yeah, yeah, that answers my question, yeah.
138 00:14:34.990 ⇒ 00:14:35.910 Pranav Narahari: Okay, cool.
139 00:14:36.060 ⇒ 00:14:40.030 Miranda Wen: Okay, cool. Yeah, thank you, Springo, that’s… that’s helpful, yeah.
140 00:14:40.030 ⇒ 00:14:41.330 Pranav Narahari: Yeah, totally, no problem.
141 00:14:41.710 ⇒ 00:14:47.840 Miranda Wen: Well, yeah, I will send you a calendar… Invite right now, and…
142 00:14:48.360 ⇒ 00:14:50.830 Miranda Wen: Let’s see if it’s necessary for tomorrow, yeah.
143 00:14:52.890 ⇒ 00:14:57.949 Pranav Narahari: A what invite? Oh, another calendar invite? Yeah, yeah. Perfect, that works.
144 00:14:58.180 ⇒ 00:15:00.379 Miranda Wen: Okay, thank you so much for now, yeah.
145 00:15:00.570 ⇒ 00:15:02.239 Pranav Narahari: Cool. Thanks, Miranda. Talk soon.
146 00:15:02.240 ⇒ 00:15:03.990 Miranda Wen: Perfect. See you, have a great day.
147 00:15:03.990 ⇒ 00:15:04.990 Pranav Narahari: See ya, you too.