Meeting Title: Robert - Pranav - Sync on SOW Date: 2026-03-04 Meeting participants: Robert Tseng, Pranav Narahari
WEBVTT
1 00:02:35.020 ⇒ 00:02:36.060 Robert Tseng: Hey, Pranav.
2 00:02:36.330 ⇒ 00:02:38.699 Pranav Narahari: Hey, Robert, sorry about the confusion.
3 00:02:38.700 ⇒ 00:02:43.360 Robert Tseng: No, no, all good. I realized I probably didn’t accept the invite. I… yeah, that’s my bad.
4 00:02:43.360 ⇒ 00:02:44.760 Pranav Narahari: All good, all good.
5 00:02:45.200 ⇒ 00:02:50.160 Robert Tseng: Are you on the East Coast now, or are you back in Illinois? Yeah.
6 00:02:50.160 ⇒ 00:02:52.590 Pranav Narahari: No, I’m, back in Massachusetts, yeah.
7 00:02:52.590 ⇒ 00:02:53.180 Robert Tseng: Okay.
8 00:02:53.410 ⇒ 00:02:53.960 Pranav Narahari: Yeah.
9 00:02:54.260 ⇒ 00:02:56.049 Robert Tseng: Wait, are you usually based out of…
10 00:02:56.220 ⇒ 00:02:57.300 Robert Tseng: Past Jesus?
11 00:02:57.300 ⇒ 00:02:58.130 Pranav Narahari: Yeah, yes.
12 00:02:58.130 ⇒ 00:02:59.019 Robert Tseng: Oh, okay.
13 00:02:59.020 ⇒ 00:03:05.699 Pranav Narahari: All of, like, December, I was in Massachusetts, and then I, I used to live in Austin, so I have, like, a lot of, like, friends there.
14 00:03:06.810 ⇒ 00:03:09.529 Pranav Narahari: I’m just, like, living at home with, like, family right now, so…
15 00:03:09.530 ⇒ 00:03:10.830 Robert Tseng: Oh, okay, got it, got it.
16 00:03:10.830 ⇒ 00:03:13.319 Pranav Narahari: Get away from the snow a little bit, let’s get out of the state.
17 00:03:13.920 ⇒ 00:03:14.470 Robert Tseng: Yeah.
18 00:03:15.520 ⇒ 00:03:17.710 Robert Tseng: Family is in Massachusetts.
19 00:03:17.710 ⇒ 00:03:20.970 Pranav Narahari: Yeah, like, I grew up here, like, I’m living with my mom and my sister, yeah.
20 00:03:21.090 ⇒ 00:03:22.880 Robert Tseng: Cool. What part?
21 00:03:23.150 ⇒ 00:03:24.230 Pranav Narahari: Worcester?
22 00:03:26.270 ⇒ 00:03:26.789 Robert Tseng: I don’t know.
23 00:03:26.790 ⇒ 00:03:31.409 Pranav Narahari: Maybe you’ve heard of, like, the College Holy Cross, or WPI.
24 00:03:31.410 ⇒ 00:03:33.339 Robert Tseng: Oh, yeah, of course, yeah.
25 00:03:33.340 ⇒ 00:03:34.830 Pranav Narahari: Yeah, that’s in Worcester.
26 00:03:35.050 ⇒ 00:03:43.000 Robert Tseng: Okay. Yeah. Oh, are your parents in academia or something? Yeah, Holy Cross is like a… I mean, I know… I have a couple good friends that went there.
27 00:03:43.000 ⇒ 00:03:44.450 Pranav Narahari: Oh, no way, same.
28 00:03:45.880 ⇒ 00:03:46.680 Robert Tseng: Huh.
29 00:03:47.700 ⇒ 00:03:55.730 Robert Tseng: Nice. Well, I would like to share… I’ll share my screen and kind of give you some more context. I just thought it’d be…
30 00:03:55.770 ⇒ 00:04:02.759 Pranav Narahari: I haven’t sat around to kind of, like, write everything up yet. So, eventually, like, first I got pitched this thing from…
31 00:04:02.770 ⇒ 00:04:11.640 Robert Tseng: like, So, Eden’s executive team, they’re calling it Project Pulse, I think.
32 00:04:12.160 ⇒ 00:04:26.579 Robert Tseng: what it really feels like is they just want to have, like, an MCP that’s, like, connected across these different, like, BizOps systems, Calendar, Slack, Gmail, whatever, and they want to…
33 00:04:26.960 ⇒ 00:04:41.780 Robert Tseng: be able to better manage, like, their remote company as well. So, I thought it was an interesting kind of proposal, because it’s like, I feel like we do bits and pieces of this at Brainforge, like, we’ve… we’ve thrown an idea… I mean, I don’t know how much we’re really, like, logging in now, but we’re, like.
34 00:04:41.780 ⇒ 00:04:59.169 Robert Tseng: you know, technically we could see how many Slack messages people are sending, like, how many… I mean, these are projects that are kind of back on the roadmap that we’re having RICO kind of do internally as well. You know, how many people are using Cursor, trying to basically measure employee productivity, and, like.
35 00:05:01.480 ⇒ 00:05:12.539 Robert Tseng: Yeah, and, like, be able to figure out, like, how to support people based off of, like, what we know about their usage of, like, the tools that we give them. And so…
36 00:05:12.540 ⇒ 00:05:25.500 Robert Tseng: I tried to, like… I put together this SOW for them already, like, I think this was off of their first iteration, and I think there needs to be some work here, so… oh, I think you’ve looked through this already, see your… I see your name kind of hovering here.
37 00:05:25.500 ⇒ 00:05:36.659 Robert Tseng: I, like, abstracted it a bit more, and just tried to… gave him a few simple concepts to kind of, like, chew on of, like, how we would approach this. One is, like, yeah, we need to structure everything in this, like, kind of, like.
38 00:05:37.090 ⇒ 00:05:55.260 Robert Tseng: kind of like the way we do it for Brainforce platform, and then there’s, like, some kind of definitions of, like, okay, like, what are the different layers to this… to the structure, like, whether we call them skills as sub-agents, and whatever, like, basically synchronous, asynchronous kind of, kind of, like.
39 00:05:55.260 ⇒ 00:06:03.439 Robert Tseng: data movement, stuff that’s either just running in the background versus what are the things that people… you as, like, the leader in the organization need to manage.
40 00:06:03.440 ⇒ 00:06:21.460 Robert Tseng: And then I tried to, like, kind of give, like, an example from, like, the case study that I think you helped put together of, like, oh, like, an old versus new way thing. So, I think they kind of… they kind of latched onto that, and they thought it was a good direction to run in, and they came back with a second iteration of, like, what they were looking at.
41 00:06:21.530 ⇒ 00:06:33.099 Robert Tseng: I don’t like this as much, because he’s basically telling me how to build it now, and I’m like, I don’t really need you to tell me how to do it, but… so I think there’s a lot of noise here. Okay. I think it would be helpful to maybe just… I can’t actually share this.
42 00:06:33.210 ⇒ 00:06:42.130 Robert Tseng: directly, so I will, I can make a copy, and then… Share it, Bruce.
43 00:06:43.010 ⇒ 00:06:43.990 Robert Tseng: We’re out.
44 00:06:44.110 ⇒ 00:06:47.790 Robert Tseng: This is my Eden email, so…
45 00:06:54.540 ⇒ 00:07:05.359 Robert Tseng: Yeah, so I’m just sharing this with you. I was thinking that maybe we could spend a few minutes, like, kind of just reading through it on this call, and then trying to just, like.
46 00:07:06.240 ⇒ 00:07:11.669 Robert Tseng: I mean, I’d like to continue to build on this SOW, so however I want to, like, edit it based off of, like, any
47 00:07:11.670 ⇒ 00:07:27.259 Robert Tseng: new things that they’ve introduced, like, I want to start to price… I want to price this thing, for them. So, obviously get your take on, like, what do you, you know, do you think this is possible, like, and what other requirements do you feel like you need in order to put together a technical SOW?
48 00:07:27.690 ⇒ 00:07:29.390 Pranav Narahari: Okay, perfect. Yeah.
49 00:07:31.470 ⇒ 00:07:37.120 Pranav Narahari: Yeah, so I literally just, like I said, like, opened up that, like, thing right before the call, because I was.
50 00:07:37.120 ⇒ 00:07:37.930 Robert Tseng: Okay.
51 00:07:38.040 ⇒ 00:07:41.609 Pranav Narahari: so I didn’t get to read through the SOW. Okay.
52 00:07:41.610 ⇒ 00:07:44.420 Robert Tseng: All good. Let’s take a few minutes to read it.
53 00:07:44.420 ⇒ 00:07:45.200 Pranav Narahari: Yeah, perfect.
54 00:07:45.470 ⇒ 00:07:46.020 Robert Tseng: Yeah.
55 00:12:39.060 ⇒ 00:12:44.679 Robert Tseng: Okay, do you feel like you got a good chance to skim it? I’m happy to kind of help point us on what, like, there’s a lot.
56 00:12:44.680 ⇒ 00:12:45.080 Pranav Narahari: Yeah.
57 00:12:45.080 ⇒ 00:12:45.920 Robert Tseng: here, too.
58 00:12:46.230 ⇒ 00:12:49.880 Pranav Narahari: I think I just, yeah, I just finished reading the acceptance criteria.
59 00:12:50.020 ⇒ 00:12:55.910 Pranav Narahari: I feel like that’s, like, probably a good place for me to… like, I feel like I understand, like, what is the picture here, and I think this.
60 00:12:55.910 ⇒ 00:12:56.540 Robert Tseng: Yeah.
61 00:12:56.540 ⇒ 00:13:04.159 Pranav Narahari: Criteria, too, like… Helps me to kind of just, like, estimate things and, like, Phase by phase.
62 00:13:06.740 ⇒ 00:13:12.909 Pranav Narahari: So, yeah, how I’m thinking about doing this is just, like, thinking… giving, like, a high-level, like, hours estimate per phase.
63 00:13:12.910 ⇒ 00:13:13.290 Robert Tseng: Yeah.
64 00:13:13.290 ⇒ 00:13:15.569 Pranav Narahari: Do you think that would be helpful?
65 00:13:15.800 ⇒ 00:13:23.700 Robert Tseng: Yeah, I think that would be helpful, and… well, yeah, just to kind of… I think we should be opinionated on, like, kind of what
66 00:13:25.310 ⇒ 00:13:39.209 Robert Tseng: I think there’s a lot of jargon and just, like, tools here that we don’t really need, so I’ve kind of going back to my shared screen. I was kind of working through the… what’s in scope. I think there’s really kind of… we call them work streams, we could call phases.
67 00:13:39.210 ⇒ 00:13:39.530 Pranav Narahari: Yeah.
68 00:13:40.280 ⇒ 00:13:44.830 Robert Tseng: Phase 2… Phase 3…
69 00:13:45.080 ⇒ 00:14:01.830 Robert Tseng: I think the first iteration was a little too narrow. He clearly expanded it since with his second one, because at first, it was just like, I just want a monthly… I just want a weekly report, but it seems like, based off of what I was reading in his… in his new doc, there’s actually more stuff than that. So, the way I see it is, okay, well, there’s, like, a…
70 00:14:01.870 ⇒ 00:14:09.199 Robert Tseng: you know, we just need to… there’s, like, a connections piece. We need to integrate with all of these different, like, you know, office functions.
71 00:14:09.200 ⇒ 00:14:09.879 Pranav Narahari: Oh, boy.
72 00:14:09.880 ⇒ 00:14:18.980 Robert Tseng: Maybe it’s just, like, a Google Drive MCP, similar to what we have, plus, like, they need to integrate with Slack, get a log of all the, kind of, like, the message history and everything that they see.
73 00:14:19.250 ⇒ 00:14:29.100 Robert Tseng: They want to be able to strip out some data from there. I don’t really know what a DLP integration is, but, yeah, they just don’t want any patient data coming through, because sometimes people are slacking
74 00:14:29.100 ⇒ 00:14:40.150 Robert Tseng: hey, this particular customer who lives here, who ordered this particular treatment, this is a telehealth company, by the way. Like, this is, like, that should not be…
75 00:14:40.410 ⇒ 00:14:48.789 Robert Tseng: I mean, they’re obviously worried about that being, kind of an exposure of HIPAA data, so, HIPAA data is, like,
76 00:14:48.790 ⇒ 00:15:01.769 Robert Tseng: just… it’s protected data. Pretty much anything around medical treatments. Name, email, number, totally fine, because that’s, like, just generic, consumer data that anybody can find. But anything around…
77 00:15:01.770 ⇒ 00:15:10.949 Robert Tseng: Which, they do have a lot more data about what medications are you on, how long have you been on this treatment, current weight, like, and, like, there’s, like, some…
78 00:15:10.950 ⇒ 00:15:29.349 Robert Tseng: health metrics, it’s not too many, but we already anonymized a lot of that in our inquiry that we set up with them, so I think this is totally doable, but it’s just that if we were to plug into Slack, and we were to download all the messages from Slack, for example, in their system, I’m sure that their team members are slacking each other, like.
79 00:15:31.350 ⇒ 00:15:34.850 Robert Tseng: Not compliant stuff all the time, so…
80 00:15:35.010 ⇒ 00:15:41.819 Robert Tseng: So yeah, we would probably just need to do some… do some cleaning there. So, I think that’s kind of how I view, the…
81 00:15:42.410 ⇒ 00:15:44.710 Robert Tseng: Like, the, the, kind of, the first phase.
82 00:15:46.040 ⇒ 00:15:53.830 Robert Tseng: Yeah, so that’s just, yeah, making sure that we have all the data sources, and then it’s all clean. And then, I guess this is probably where you’re,
83 00:15:54.160 ⇒ 00:16:06.749 Robert Tseng: gonna be a lot better than I am at kind of outlining what this actually looks like. To me, I think he’s really just building, like, this whole context graph, this concept that we’ve… that we’ve built for our,
84 00:16:06.890 ⇒ 00:16:15.959 Robert Tseng: our work internally, so whether we need to use Vertex AI, or, like, you know, do we care about, like, how we compute, like, theme weights, like, this is all, like.
85 00:16:15.980 ⇒ 00:16:22.790 Robert Tseng: computational stuff, I just… I think… I think this is probably way… way… way more detailed than he needed to see, like, I…
86 00:16:22.800 ⇒ 00:16:37.649 Robert Tseng: I think we could do a better job of framing this. I don’t know if you did something similar for Lilo, or if you were to think about, like, how do we describe what we built in BlaineForge platform to somebody? Like, I think this is really what they’re trying to do. Like,
87 00:16:37.730 ⇒ 00:16:41.139 Robert Tseng: And then, like, I think Phase 3 is really just making…
88 00:16:41.190 ⇒ 00:16:59.699 Robert Tseng: like, making it helpful for specific chief operating officer use cases. So, yeah, like, part of it is… is the kind of… I guess you saw in the acceptance criteria, it’s gonna be getting, like, the pulse report, wanting to know… I mean, these are pretty generic things, but…
89 00:16:59.820 ⇒ 00:17:06.829 Robert Tseng: where are people spending most of their time? Like, where are people getting stuck? Like, are there certain topics, like, in
90 00:17:06.829 ⇒ 00:17:21.389 Robert Tseng: email or, you know, Slack threads, pretty much, or people, you know, and things that are talked about in, in, in meetings. Like, do we know what those themes… can we identify what those themes are? And it’s like, hey, maybe,
91 00:17:22.030 ⇒ 00:17:26.350 Robert Tseng: Everybody’s asking the same question and really confused about,
92 00:17:27.089 ⇒ 00:17:37.079 Robert Tseng: a particular pharmacy, because they don’t actually… they don’t share enough data about, when their shipments go out. And so it’s like a… maybe a…
93 00:17:37.220 ⇒ 00:17:43.930 Robert Tseng: like, I’m using that as, like, a guess, like, as an example, but I actually think this is probably something that you would find.
94 00:17:43.980 ⇒ 00:17:58.759 Robert Tseng: Where, just to kind of contrast, like, a telehealth provider, multiple pharmacies, there are some pharmacies that have direct integration. We see the data flowing through via their API. Then there’s, like, one pharmacy that’s terrible, and they know they need to get off of it, but, like.
95 00:17:59.000 ⇒ 00:18:03.829 Robert Tseng: They’re just, like, sending batches of, like, order information.
96 00:18:03.860 ⇒ 00:18:17.290 Robert Tseng: whenever they want, and it’s just, like, in CSVs, ends up in email. That’s… that pharmacy, it has the most problems, it’s the hardest for the team to serve. I’m sure their team is spending a lot of time just, like, kind of in confusion.
97 00:18:17.290 ⇒ 00:18:30.980 Robert Tseng: talking… meeting with their… with the external vendor, talking in Slack, like, meetings, like, can they, like, identify, hey, like, this is… this is, like, a bottleneck that the team is, like, kind of constantly, like, stuck on operationally?
98 00:18:31.040 ⇒ 00:18:40.599 Robert Tseng: And then, like, I don’t really know what this efficiency trend thing is, but, like, we could kind of split that out with him. But I’m just trying to give you a flavor of, like, from a COO perspective, like, what he would be caring about.
99 00:18:41.170 ⇒ 00:18:42.880 Robert Tseng: He’s just trying to, like, you know.
100 00:18:43.030 ⇒ 00:19:02.579 Robert Tseng: he cares about utilization of his team, where the biggest bottlenecks are, and, like, he just wants a way to be able to, to get that type of report to him. So, I think it’d be cool to call it more… it’s kind of like an executive assistant or a chief of staff to him, so, right. Yeah.
101 00:19:02.580 ⇒ 00:19:12.559 Robert Tseng: But anyway, I think that’s kind of the way I am kind of packaging all this together. I guess, what questions do you have, kind of hearing me put it… lay it out that way?
102 00:19:12.560 ⇒ 00:19:24.099 Pranav Narahari: Yeah, no, this is super helpful. Also, what was super helpful is just, like, there’s a certain level of, like, granularity in here that, like, I wasn’t for sure is, like, is this a requirement, or is this kind of just, like…
103 00:19:24.100 ⇒ 00:19:26.550 Robert Tseng: What we kind of came up with, but it sounds like…
104 00:19:26.630 ⇒ 00:19:31.970 Pranav Narahari: it’s not all necessarily requirements. Like, one thing is, like.
105 00:19:32.440 ⇒ 00:19:40.620 Pranav Narahari: there was one thing that I was like, okay, I need to look into this a little bit more, like, unsupervised, like, clustering, like, cane games. Yeah. Okay. Yeah.
106 00:19:40.620 ⇒ 00:19:46.990 Robert Tseng: That’s all just jargon that they threw in that I don’t really think they need to tell us how to do it. Like, I think we’re probably over-complicating it, yeah.
107 00:19:46.990 ⇒ 00:19:54.280 Pranav Narahari: Exactly. At the end of the day, we just need to make sure we’re driving insights that are actually, like, accurate, and then probably…
108 00:19:54.730 ⇒ 00:20:07.600 Pranav Narahari: Yeah, I think that’s the main thing. So, how I think about this in, like, the simplest terms of, like, the three phases is, okay, we need to get the relevant data from each of these four sources, and…
109 00:20:07.730 ⇒ 00:20:14.849 Pranav Narahari: MCP is an option, but I think MCP is a better, medium for, like, a chat interface, where you need, like.
110 00:20:15.280 ⇒ 00:20:15.640 Robert Tseng: real.
111 00:20:15.640 ⇒ 00:20:24.580 Pranav Narahari: time, like, insights, and you need it straight from the connection. Since we’re giving them weekly insights, I think what’s better here is a data warehouse.
112 00:20:24.770 ⇒ 00:20:28.250 Pranav Narahari: I think that makes the most sense here. It’s going to be…
113 00:20:28.970 ⇒ 00:20:36.540 Pranav Narahari: Yeah, I just think it’s… it’s probably just a better application for… for what we’re trying to build here.
114 00:20:36.660 ⇒ 00:20:49.250 Pranav Narahari: And so, yeah, building the connections within some ETL tool, so, you know, probably will go with whatever the team has the most. Well, I think Polytomic, we have, like, the most amount of.
115 00:20:49.420 ⇒ 00:20:52.030 Pranav Narahari: like, I’m just thinking out loud, so we’ll probably use, like.
116 00:20:52.030 ⇒ 00:20:52.510 Robert Tseng: Yeah.
117 00:20:52.510 ⇒ 00:21:00.640 Pranav Narahari: that. For Lilo, we used, Mother Duck for, like, the data warehouse, and then…
118 00:21:01.670 ⇒ 00:21:16.369 Pranav Narahari: So that’s… that’s phase one, just getting all the data in, so there’s a couple things that we need to do for just, like, research purposes, which is assessing, okay, what is data that we actually want to track? What is just junk data that we were not going to be able to make?
119 00:21:16.500 ⇒ 00:21:18.570 Pranav Narahari: insights on.
120 00:21:18.670 ⇒ 00:21:32.230 Pranav Narahari: So, that’ll be a conversation with the team in Phase 1. But then what phase two is, is like, okay, how do we then use all this data so that we can actually, like, in kind of, like.
121 00:21:32.400 ⇒ 00:21:36.760 Pranav Narahari: It would be really cool if we did make, like, a… like a… like a graph.
122 00:21:37.230 ⇒ 00:21:40.270 Pranav Narahari: maybe not necessary.
123 00:21:40.550 ⇒ 00:21:52.719 Pranav Narahari: But how do we just, like, say, like, okay, these few fields relate to these, and we can use them together to give, like, better context for the insights that will drive to the CEO in Phase 3? Yeah.
124 00:21:52.890 ⇒ 00:21:56.340 Pranav Narahari: And then… Yeah, so…
125 00:21:56.980 ⇒ 00:22:11.029 Pranav Narahari: that is basically the entire problem of Phase 2, just like, okay, we have all the data, now how can we use that data to then use an LLM to drive insights? Yeah. And then Phase 3 is, I think, okay.
126 00:22:11.340 ⇒ 00:22:21.800 Pranav Narahari: really quickest thing is just formatting those insights into, like, you know, what we mentioned here, or whatever it ends up being, like, 5 bandwidth things, 3 friction points,
127 00:22:22.020 ⇒ 00:22:23.979 Pranav Narahari: But then it’s going to be just, like…
128 00:22:24.360 ⇒ 00:22:30.290 Pranav Narahari: there’s gonna be a QA, there’s gonna be a big part of that as well, too. It’s like, okay, what are the edge cases here? Like…
129 00:22:30.750 ⇒ 00:22:37.580 Pranav Narahari: we want to very much stress test it. We don’t want just, like, junk stuff going into, like, this report.
130 00:22:39.950 ⇒ 00:22:41.160 Pranav Narahari: What else?
131 00:22:42.370 ⇒ 00:22:54.149 Pranav Narahari: Yeah, that seems like… Phase 3 seems like probably the most straightforward, but it’s just going to be, like, probably the most rigorous in terms of testing. What’s not compassed, like, in what I just said, is like, okay, we need to be able to deploy this thing
132 00:22:54.380 ⇒ 00:23:02.980 Pranav Narahari: If GCP is a requirement, then okay, we can deploy it into GCP. What is easier is probably deploying it into Railway.
133 00:23:04.330 ⇒ 00:23:13.150 Pranav Narahari: I don’t know, I know Eden’s an existing client, so maybe they already have their infrastructure set up, and they don’t want two different infrastructures.
134 00:23:13.540 ⇒ 00:23:16.570 Robert Tseng: We basically maintain all their infrastructure, so,
135 00:23:16.610 ⇒ 00:23:35.240 Robert Tseng: Yeah, I mean, I would say this… we’re the most integrated with this client. Like, this is… this is our biggest client. This is, like, a third of our business. So, like, to me, I would… I would be looking to bring you and, I guess, whoever you feel like you need to bring in how many hours, like, I want to basically go to the CEO and be like, hey, we can do this for you, we’ll meet… like, I think the expectation is…
136 00:23:35.240 ⇒ 00:23:42.550 Robert Tseng: they’re not… they don’t… they’re not opinionated about the tools, like, they let us just kind of decide all the tools. So, yeah, I kind of just…
137 00:23:42.680 ⇒ 00:23:49.869 Robert Tseng: Obviously, cost is something that they care about, but they expect us to just kind of, decide on infrastructure decisions.
138 00:23:50.490 ⇒ 00:23:51.050 Robert Tseng: Yeah.
139 00:23:51.300 ⇒ 00:23:54.210 Pranav Narahari: And what’s, like, is there existing…
140 00:23:54.540 ⇒ 00:24:03.569 Pranav Narahari: things that we built for them that they have that’s going to assist on this project? Like, do they already have connections with some of these, platforms, like.
141 00:24:03.710 ⇒ 00:24:06.240 Pranav Narahari: Like, do we already have a data warehouse set up for…
142 00:24:06.240 ⇒ 00:24:11.720 Robert Tseng: Yeah, we already have BigQuery set up, so, yeah, as far as, like.
143 00:24:12.150 ⇒ 00:24:18.820 Robert Tseng: I don’t think I… I’d probably talk to a waste to see what other info that we have maintaining for them, but…
144 00:24:19.030 ⇒ 00:24:23.300 Robert Tseng: I know we don’t have these connections for them yet.
145 00:24:23.300 ⇒ 00:24:24.410 Pranav Narahari: I’m gonna go, okay.
146 00:24:24.410 ⇒ 00:24:24.850 Robert Tseng: Yeah.
147 00:24:26.030 ⇒ 00:24:26.780 Robert Tseng: And…
148 00:24:26.780 ⇒ 00:24:32.380 Pranav Narahari: Even though we have BigQuery set up, because then we’ll just have everything else, probably…
149 00:24:32.520 ⇒ 00:24:37.509 Pranav Narahari: whatever ETL tool, we’ll just have it all… all the data set up in… into there. Yeah.
150 00:24:37.870 ⇒ 00:24:38.830 Pranav Narahari: Okay.
151 00:24:40.360 ⇒ 00:24:43.329 Pranav Narahari: Alright, so, yeah, now let me think about just…
152 00:24:47.280 ⇒ 00:24:49.810 Pranav Narahari: Let me… let me just look at…
153 00:24:50.070 ⇒ 00:24:53.820 Pranav Narahari: Just for reference, like, what I said for Lilo for, like, kind of similar items.
154 00:24:54.270 ⇒ 00:24:55.060 Robert Tseng: Yeah.
155 00:24:55.240 ⇒ 00:25:06.490 Robert Tseng: And what I’m gonna do is, I’m gonna strike out more stuff that’s just, like, I think it’s just too much jargon here, so, like, hopefully it’ll be… it’ll be a little simpler, and we can kind of build back up from there. Yeah.
156 00:25:07.000 ⇒ 00:25:11.659 Robert Tseng: Yeah, so I’m gonna, I’m gonna go through this document, kind of delete, delete a bunch of stuff here.
157 00:25:16.790 ⇒ 00:25:17.670 Pranav Narahari: Okay.
158 00:25:38.320 ⇒ 00:25:41.500 Pranav Narahari: Okay, yeah, so for Lilo, we said…
159 00:25:46.860 ⇒ 00:25:50.290 Pranav Narahari: You said data warehouse would take… what?
160 00:25:51.390 ⇒ 00:25:54.700 Pranav Narahari: 50… 60 hours?
161 00:25:56.270 ⇒ 00:26:02.219 Pranav Narahari: Okay, so 60 hours, but they were starting from scratch, so it’s like…
162 00:26:06.280 ⇒ 00:26:08.900 Pranav Narahari: Yeah, I’ll just say 50 hours for that.
163 00:26:09.910 ⇒ 00:26:11.080 Pranav Narahari: So…
164 00:26:11.280 ⇒ 00:26:11.640 Robert Tseng: Okay.
165 00:26:11.640 ⇒ 00:26:13.419 Pranav Narahari: That’s essentially phase…
166 00:26:14.770 ⇒ 00:26:20.520 Pranav Narahari: That’s phase one. That’s not fully phase one, though, because we need to also do some research to figure out, like.
167 00:26:21.660 ⇒ 00:26:30.519 Pranav Narahari: Not a ton of research, to be honest. Okay, yeah, so… I would say, yeah, 60 hours for a data warehouse, just phase one, 60 hours, seems…
168 00:26:31.810 ⇒ 00:26:34.810 Pranav Narahari: Seems like it makes sense.
169 00:26:35.100 ⇒ 00:26:41.580 Pranav Narahari: But Phase 2 is probably going to be, like, the… the bulk of it.
170 00:26:42.830 ⇒ 00:26:45.460 Pranav Narahari: Let me think… Phase 2…
171 00:26:46.340 ⇒ 00:26:51.870 Pranav Narahari: Now, this is something that we just didn’t do for Lilo at all, so it’s kind of like, I have no reference. Yeah.
172 00:26:52.570 ⇒ 00:26:54.629 Pranav Narahari: I mean, I have some references from, like.
173 00:26:55.190 ⇒ 00:26:58.250 Pranav Narahari: things I’ve done in this field, but just, like, not from Lilo.
174 00:27:02.950 ⇒ 00:27:04.360 Pranav Narahari: Okay, so…
175 00:27:40.540 ⇒ 00:27:50.320 Pranav Narahari: Like, this is just, like, a part where I’m just like, I know it’s going to be difficult, like, we’re gonna just be like, okay, the insights we’re getting are junk at first, and then we’re just gonna have to continuously refine it.
176 00:27:50.580 ⇒ 00:27:51.510 Pranav Narahari: Yeah.
177 00:28:00.340 ⇒ 00:28:05.089 Pranav Narahari: How many weeks would that take? Let me think… Just one person’s on it.
178 00:28:06.380 ⇒ 00:28:09.659 Pranav Narahari: So, maybe 2… Two and a half?
179 00:28:09.810 ⇒ 00:28:14.700 Pranav Narahari: Two to two and a half, so, like, 80 to… Yeah
180 00:28:14.890 ⇒ 00:28:20.390 Pranav Narahari: like, it’s pretty straightforward, like, I would say 80 hours,
181 00:28:23.780 ⇒ 00:28:30.330 Pranav Narahari: It could be quicker, But… I’m… I’m just…
182 00:28:30.890 ⇒ 00:28:38.110 Pranav Narahari: I want to be kind of more conservative, too, and then of course, like, in negotiations, you guys can talk about reducing timelines and whatever.
183 00:28:38.110 ⇒ 00:28:38.530 Robert Tseng: Yeah.
184 00:28:38.530 ⇒ 00:28:44.750 Pranav Narahari: Or potentially extending, which would be great, of course. Yeah.
185 00:28:46.050 ⇒ 00:28:49.520 Pranav Narahari: But, okay, so now Phase 3. Phase 3 is probably, like…
186 00:28:50.030 ⇒ 00:28:52.099 Robert Tseng: Quickest, but then also…
187 00:28:52.120 ⇒ 00:28:56.720 Pranav Narahari: This whole thing, we gotta think about deploying from dev to then production.
188 00:28:57.150 ⇒ 00:29:03.379 Pranav Narahari: So I’d want, like, kind of a week to just, like, work out stuff for that.
189 00:29:03.630 ⇒ 00:29:05.980 Robert Tseng: Okay. 40, like…
190 00:29:07.380 ⇒ 00:29:12.490 Pranav Narahari: Yeah, I would say maybe another 60 for this. I think that’s fair. Like…
191 00:29:13.270 ⇒ 00:29:15.440 Pranav Narahari: Like, half a week for just…
192 00:29:16.400 ⇒ 00:29:22.890 Pranav Narahari: Yeah, half a week for just, like, generating, like, just that final… What does this…
193 00:29:23.040 ⇒ 00:29:28.370 Pranav Narahari: what do these insights look like? Which is…
194 00:29:28.760 ⇒ 00:29:47.880 Pranav Narahari: we’re already gonna kind of create, like, the backend for driving those insights in Phase 2, it’s just about, like, the formatting for Phase 3, and then just adhering to a certain structure. And then, yeah, another just week to just kind of stress test it, add guardrails, deploy to production.
195 00:29:50.380 ⇒ 00:29:52.630 Robert Tseng: Okay, yeah, that sounds fair.
196 00:29:52.980 ⇒ 00:29:59.410 Robert Tseng: Yeah, I mean, like, do you… do you think it would be you, just you, or do you… who do you… who do you think we would staff on this?
197 00:30:03.430 ⇒ 00:30:09.709 Robert Tseng: I mean, I would probably be the one kind of figuring out what these requirements are with you for Phase 3.
198 00:30:10.160 ⇒ 00:30:12.180 Robert Tseng: Yeah. Yeah.
199 00:30:15.110 ⇒ 00:30:20.630 Robert Tseng: And then I’m assuming you would probably work with, like, Awash on our team and the data engineers to help with this part?
200 00:30:20.860 ⇒ 00:30:24.069 Pranav Narahari: Yeah. That’s the one thing, probably just, like…
201 00:30:24.170 ⇒ 00:30:30.609 Pranav Narahari: getting… because I’ve never used BigQuery, just getting acclimated to BigQuery. Yeah. It’s probably pretty straightforward, but…
202 00:30:30.750 ⇒ 00:30:39.700 Pranav Narahari: I don’t even think we need to have, like, always staffed on the project, per se. Maybe he’ll put in a few hours here and there just to, like.
203 00:30:40.150 ⇒ 00:30:45.109 Pranav Narahari: get me up to speed on, like, where Eden currently is, like, how BigQuery is currently being used.
204 00:30:46.090 ⇒ 00:30:56.950 Pranav Narahari: One thing is just the… that I didn’t mention, that, okay, maybe… just the un…
205 00:30:57.280 ⇒ 00:31:01.690 Pranav Narahari: making this data anonymized, I’ve never done that before.
206 00:31:02.300 ⇒ 00:31:08.360 Pranav Narahari: And so… seems like a pretty straightforward task. Like, it doesn’t… there’s probably…
207 00:31:08.780 ⇒ 00:31:12.830 Pranav Narahari: Is the… is the Eden team, like, at Brainforge, like, already doing that?
208 00:31:13.510 ⇒ 00:31:29.979 Robert Tseng: We do that with all, like, kind of patient data. Like I said, in the transactions, like, when people place orders, like, we’re… we have it pretty separated out, so anything that goes into reporting, you can’t drill down to the patient level.
209 00:31:31.010 ⇒ 00:31:36.240 Robert Tseng: Yeah, when… and… but, like I said, once we start connecting these other sources in.
210 00:31:36.720 ⇒ 00:31:46.000 Robert Tseng: Like, I’m sure people are… breaking all the rules in talking internally in Slack and Gmail about, like.
211 00:31:46.140 ⇒ 00:31:51.220 Robert Tseng: About patients, and we would have to figure out, basically, how to
212 00:31:52.050 ⇒ 00:31:55.980 Robert Tseng: how to clean… clean that data before it ends in BigQuery as well.
213 00:31:56.630 ⇒ 00:31:58.000 Pranav Narahari: Yeah, 100%.
214 00:31:58.210 ⇒ 00:31:59.240 Pranav Narahari: Okay.
215 00:32:00.160 ⇒ 00:32:02.579 Pranav Narahari: I kind of want to…
216 00:32:05.840 ⇒ 00:32:06.710 Pranav Narahari: just…
217 00:32:06.860 ⇒ 00:32:11.380 Robert Tseng: Take that into consideration a little bit more, and maybe just extend.
218 00:32:11.380 ⇒ 00:32:13.700 Pranav Narahari: Phase 1 by, like, a few hours.
219 00:32:17.530 ⇒ 00:32:22.020 Pranav Narahari: Probably just, like… Like, another day, so, like, maybe let’s, like.
220 00:32:22.330 ⇒ 00:32:25.600 Pranav Narahari: 10 hours to just, like, figure out how do I…
221 00:32:25.740 ⇒ 00:32:30.769 Pranav Narahari: how do I do that in the best way? Considering, like, that there’s all these different avenues, like…
222 00:32:31.170 ⇒ 00:32:36.540 Pranav Narahari: Slack, Gmail are the main two. Oh, actually, we’ll probably drive, too. There could be, like, certain documents.
223 00:32:36.540 ⇒ 00:32:36.870 Robert Tseng: Yeah.
224 00:32:37.380 ⇒ 00:32:38.340 Pranav Narahari: Yeah.
225 00:32:39.060 ⇒ 00:32:42.459 Pranav Narahari: So, I don’t think I necessarily need to…
226 00:32:43.570 ⇒ 00:32:47.130 Pranav Narahari: Yeah, to answer your previous question, like, who else do I need staffed on this? Like, I could…
227 00:32:47.700 ⇒ 00:32:50.999 Pranav Narahari: Seems like something I could do on my own.
228 00:32:52.000 ⇒ 00:32:55.710 Pranav Narahari: like, as an engineer, like, I could build everything,
229 00:32:57.470 ⇒ 00:33:04.459 Pranav Narahari: Yeah, I feel pretty confident in that. Like, nothing… this all makes sense to me, I already have, like, a picture of how I would start building it, like…
230 00:33:05.100 ⇒ 00:33:09.360 Pranav Narahari: Nothing really seems like… Totally up in the air to me.
231 00:33:09.910 ⇒ 00:33:10.520 Robert Tseng: Cool.
232 00:33:10.960 ⇒ 00:33:20.990 Robert Tseng: Okay, well then what I’m gonna do is I’m gonna, like I said, clean up the dock, I’m gonna put pricing on here, and then I’m gonna pretty much schedule one more call with the COO, and I’ll probably bring you onto that call.
233 00:33:21.110 ⇒ 00:33:22.450 Robert Tseng: Cool.
234 00:33:22.750 ⇒ 00:33:33.579 Robert Tseng: Yeah. I think, like, some concessions that I’m willing to make here, like, whether we call 60 hours or not, I think he wants to… I want to, like, let him know you should bring in somebody else, because, like.
235 00:33:33.650 ⇒ 00:33:57.210 Robert Tseng: you’re gonna… we’re gonna be able to share resources with the existing team, it’s not gonna slow down. We were able to… you know, Pranav’s already very integrated with the Brainforge team, so him plugging into the data… to the data team is gonna be a lot smoother than you going for another vendor, or, like, another person to… to build this. So, you know, if we can… if that helps speed up time, or just, like, reduce costs in the Phase 1, like, I’m willing to make a concession there.
236 00:33:57.210 ⇒ 00:34:12.309 Robert Tseng: Yeah. Here, totally understand, this is, like, very much, like, we have not touched this at all. So, like, all this stuff is kind of where your… where your expertise is going to really have to come into play. So, I… he’s not technical, but, like, I would tell him… I would be prepared to tell him why
237 00:34:12.929 ⇒ 00:34:22.699 Robert Tseng: his choices are not necessarily the way that we would go, and, like, what your… what your proposal would be, I guess. I don’t know if it’s too heavyweight for you to kind of put together
238 00:34:22.770 ⇒ 00:34:30.089 Robert Tseng: something, like, I don’t know, some chart, like, something pretty high level, just to explain to him, just to give him confidence that you, like, kind of know
239 00:34:30.090 ⇒ 00:34:47.610 Robert Tseng: after you’ve built this thing before, you know what to do. Like, obviously he just kind of ran everything through GPT or Gemini, and he just threw together some stack that they recommended to him. So, like, I think, you know, he overcomplicated it, this is the way, like, you’re kind of making sense of, like, what actually goes into this.
240 00:34:47.610 ⇒ 00:34:54.720 Robert Tseng: And then I’ll let him know that, like, I’ll help him kind of flush this stuff out. So, I think this is… that’s kind of the approach that I want to take with him.
241 00:34:55.159 ⇒ 00:35:01.919 Pranav Narahari: Okay, cool. And in terms of, like, timeline, too, like, bringing on another engineer, I guess, of course, like, it would just…
242 00:35:02.309 ⇒ 00:35:16.719 Pranav Narahari: like, reduce the overall time to, like, going to production. So… Yeah. What… what did… so we said 60, 80, 60, 140, 200, that’s…
243 00:35:17.559 ⇒ 00:35:21.229 Pranav Narahari: 40… yeah, so it’s, like, 5 weeks?
244 00:35:21.640 ⇒ 00:35:22.230 Robert Tseng: Yep.
245 00:35:22.230 ⇒ 00:35:23.950 Pranav Narahari: Yeah, okay, that sounds good.
246 00:35:24.640 ⇒ 00:35:26.120 Robert Tseng: Okay. Yeah. Cool.
247 00:35:27.290 ⇒ 00:35:31.289 Robert Tseng: Alright, then, I’m gonna put this together. Let’s go, let’s go get it.
248 00:35:31.290 ⇒ 00:35:32.749 Pranav Narahari: Let’s do it. Cool.
249 00:35:32.750 ⇒ 00:35:35.190 Robert Tseng: Thanks. Yeah, thank you. Alright, talk to you later.
250 00:35:35.190 ⇒ 00:35:36.180 Pranav Narahari: Og players, yep.