Meeting Title: Brainforge <> Lilo Social: Confirm Priorities + Deliverables Date: 2026-02-24 Meeting participants: Pranav Narahari, Bobby Palmieri, zacfromson
WEBVTT
1 00:01:25.790 ⇒ 00:01:26.850 Bobby Palmieri: Hello, sir.
2 00:01:28.360 ⇒ 00:01:29.760 Pranav Narahari: Hey, hey, how’s it going?
3 00:01:30.120 ⇒ 00:01:30.939 Bobby Palmieri: How are you?
4 00:01:31.300 ⇒ 00:01:33.310 Pranav Narahari: Doing pretty good, pretty good.
5 00:01:34.240 ⇒ 00:01:36.120 Pranav Narahari: Did you get any time off yesterday?
6 00:01:36.790 ⇒ 00:01:44.000 Pranav Narahari: A little bit. I was, driving, from…
7 00:01:44.190 ⇒ 00:01:57.179 Pranav Narahari: where was I starting? Yeah, so yesterday I was in Illinois, and then now in Pittsburgh. So, like, on the East Coast right now, like, northeast area, there’s, like, a crazy snowstorm, and so just trying to, like, evade that, basically.
8 00:01:57.180 ⇒ 00:01:59.590 Bobby Palmieri: Where are you… where are you trying to get to?
9 00:01:59.950 ⇒ 00:02:01.179 Pranav Narahari: To Worcester, Mass.
10 00:02:01.470 ⇒ 00:02:06.060 Bobby Palmieri: Okay, my buddy was in the Dominican for a wedding.
11 00:02:07.360 ⇒ 00:02:14.640 Bobby Palmieri: And… supposed to fly back direct to Boston on Sunday, flight got canceled.
12 00:02:15.280 ⇒ 00:02:25.339 Bobby Palmieri: Then they were like, okay, we’re gonna fly you into Fort Lauderdale and then to Boston. That flight got canceled. So he ended up flying from the Dominican to Cleveland.
13 00:02:25.460 ⇒ 00:02:32.950 Bobby Palmieri: Spent a night in Cleveland, took a bus from Cleveland to Pittsburgh, rented a car, and then drove to Logan.
14 00:02:32.950 ⇒ 00:02:35.079 Pranav Narahari: Jeez. Oh my gosh.
15 00:02:35.930 ⇒ 00:02:37.740 zacfromson: Are you still in Worcester full-time?
16 00:02:37.950 ⇒ 00:02:41.499 Pranav Narahari: So, yeah, basically.
17 00:02:41.800 ⇒ 00:02:55.380 Pranav Narahari: Since, like, at Brainforge, everything’s remote. It’s, I spent some time in Austin, like, was, like, syncing with the boutam every, like, couple weeks. We would get, like, a WeWork, and… but yeah, full-time, basically, in Massachusetts.
18 00:02:55.380 ⇒ 00:02:55.970 Bobby Palmieri: Yeah.
19 00:02:55.970 ⇒ 00:02:58.200 Pranav Narahari: Yeah, yeah. Cool. Yup, yup.
20 00:02:58.730 ⇒ 00:03:00.249 zacfromson: I thought you were moving to Austin.
21 00:03:00.250 ⇒ 00:03:03.650 Pranav Narahari: Anyways, cool, let’s hop on in. One day, one day.
22 00:03:04.280 ⇒ 00:03:08.440 Bobby Palmieri: Are anyone else joining us, or just us today?
23 00:03:08.440 ⇒ 00:03:09.679 Pranav Narahari: Just 3 of us today.
24 00:03:09.750 ⇒ 00:03:18.420 Bobby Palmieri: Sweet. Cool. I think, appreciate you taking the time, or whoever. I know you sent the message, but hopefully you were…
25 00:03:18.420 ⇒ 00:03:21.379 Pranav Narahari: Driving and not working in spreadsheets to…
26 00:03:21.380 ⇒ 00:03:31.279 Bobby Palmieri: together. Yeah, I think, like I said, I think, you know, at the end of our conversation a week or two ago, like, I think we just want to be putting
27 00:03:31.680 ⇒ 00:03:36.269 Bobby Palmieri: more stuff in the done column, right? So,
28 00:03:36.840 ⇒ 00:03:42.979 Bobby Palmieri: For… for us, that’s, I guess, like, three core things, is…
29 00:03:43.290 ⇒ 00:03:49.040 Bobby Palmieri: MCP, like, just figuring out how we can get better data, and then figuring out the Google connect there.
30 00:03:49.290 ⇒ 00:03:52.660 Pranav Narahari: Yup. And, like, I think Meta, like.
31 00:03:52.680 ⇒ 00:03:56.040 Bobby Palmieri: Meta, Klaviyo, and Google all seem to be working at, like.
32 00:03:56.900 ⇒ 00:04:03.890 Bobby Palmieri: 85 to 95%, depending on what prompt you ask. Yeah. So I think just figuring out where that’s breaking.
33 00:04:04.180 ⇒ 00:04:06.599 Bobby Palmieri: Fixing that, and then, like, cool, check.
34 00:04:07.040 ⇒ 00:04:07.790 Pranav Narahari: Yeah.
35 00:04:08.600 ⇒ 00:04:22.899 Pranav Narahari: I totally agree with that. I think we kind of even went a little bit more specific on these specific things that we’ve talked about in the last couple weeks, and then mentioned that in the to-do column of that deliverables page that we sent over.
36 00:04:22.900 ⇒ 00:04:38.340 Pranav Narahari: So, we’ve scoped out, like, kind of 14 things that we’ve said, like, okay, let’s get these things done as priority, because they’re in the SOW, and you guys have also mentioned, like, these are kind of the things that we want to just drive from that 70% to 100%. Yep.
37 00:04:38.410 ⇒ 00:04:57.609 Pranav Narahari: And so, I’m completely, like, in line there, too. It sounds like, based on what you just said, like, that makes sense. And in these, like, 14 things, like, we’ve scoped out, like, specifically, like, the MCP connectors as well. And so, yeah, Google Ads, MCP UI enablement, like, that’s one thing that you guys have mentioned, how it’s, like.
38 00:04:57.810 ⇒ 00:05:16.900 Pranav Narahari: not fully there. You’d have to, like, talk about… and we’ve kind of already fixed that in dev. We just probably need to push that in, like, this week. I know I’m gonna be out Thursday, Friday. Sam is also gonna be out Friday as well. So… but in the… either this week or next week, that’ll be up in production.
39 00:05:16.930 ⇒ 00:05:24.090 Pranav Narahari: Let’s see, there’s like… and then forecasting is the big chunk of things, right? So…
40 00:05:24.090 ⇒ 00:05:43.989 Bobby Palmieri: Yeah, just to… just to iterate, like, I think there’s, like, 3 things, and then I agree, there’s probably, like, 14 subtasks within that. To me, it’s… it’s the Google connector so that you can select, like, the account, right? So, like, right now, it gets access to the parent account, and then we need child account to be selected. Yeah.
41 00:05:44.400 ⇒ 00:05:52.639 Bobby Palmieri: And that could just be a field where we paste in, and it’s, like, added to the system prompt, or something along those lines. I don’t know what you guys are thinking.
42 00:05:52.850 ⇒ 00:06:06.760 Bobby Palmieri: And then fine-tuning the model, probably upgrading to Sonnet 4.6 and some temperament changes that you were discussing. I think Sam had some additional thoughts on, like, code execution. I think that puts MCP in the done column.
43 00:06:07.240 ⇒ 00:06:09.010 Bobby Palmieri: I think Slack reports.
44 00:06:09.390 ⇒ 00:06:19.070 Bobby Palmieri: have been super helpful on the meta one. You know, I think we have, like, 3 more that we need to just roll out and probably have those thread in the daily KPIs.
45 00:06:19.680 ⇒ 00:06:20.100 Pranav Narahari: Yup.
46 00:06:20.100 ⇒ 00:06:31.149 Bobby Palmieri: Slack, and then that’s in the done column, and then obviously forecasting, you know, is the biggest chunk of that. That’s, you know, I think where we just want to
47 00:06:31.260 ⇒ 00:06:37.029 Bobby Palmieri: Drive all of our attention to getting, you know, those three things done, and
48 00:06:37.690 ⇒ 00:06:44.500 Bobby Palmieri: But I wanna, like, align on, like, just… pushing… Like, one at a time.
49 00:06:44.760 ⇒ 00:06:56.710 Bobby Palmieri: To make sure that, like, we don’t go from, like, 3 things at 90% to 3 things at 93%. It’s like, 2 things stay at 90, and we get 1 to 100. Perfect. In that regard.
50 00:06:56.710 ⇒ 00:07:05.570 Pranav Narahari: I have a couple thoughts on that, too, you know, like, forecasting is probably, like, it’s… we’ve got pretty far with that, but there’s still a few things, like…
51 00:07:05.670 ⇒ 00:07:12.480 Pranav Narahari: getting in the connections for more than just Shopify, we need to get the connections for Meta, Google, and so…
52 00:07:12.670 ⇒ 00:07:22.109 Pranav Narahari: But MCP is something we worked on in the beginning of January. Let’s kind of go back to that and just finalize that. Same thing with Slack reports, there were some issues with…
53 00:07:22.210 ⇒ 00:07:34.100 Pranav Narahari: yeah, there was issues with Google, with just kind of their timeline for getting us approved. So, now that that’s no longer an issue, now let’s just, like, finalize the Slack reports as well.
54 00:07:34.440 ⇒ 00:07:35.320 Pranav Narahari: I…
55 00:07:35.830 ⇒ 00:07:41.359 Pranav Narahari: I do need to talk to Sam about this a little bit, though, just in terms of, like, high-level architecture, because…
56 00:07:41.600 ⇒ 00:07:46.769 Pranav Narahari: The thing about all of this data is that we’re running things, kind of.
57 00:07:47.060 ⇒ 00:07:50.180 Pranav Narahari: Two separate things that are doing similar…
58 00:07:50.290 ⇒ 00:08:00.980 Pranav Narahari: similar tasks, and I think the double work is only gonna cause future issues for the platform, and what I’m referring to is the MCPs and the data warehouse data. And so…
59 00:08:02.440 ⇒ 00:08:11.559 Pranav Narahari: Pulling in data from the MCPs for the Slack reports is probably not the best way we want to… of doing things. They should really be pulling the data from the data warehouse.
60 00:08:11.910 ⇒ 00:08:22.700 Pranav Narahari: And the data ware… like, the MCPs are wired up, and they’re working. The data warehouse is still in progress, right? We’re only bringing the data from Shopify,
61 00:08:22.700 ⇒ 00:08:35.230 Pranav Narahari: even Meta, the Meta Ads, Slack report is pulling it from the MCP. So, what I’ll do is I’ll talk to Sam about just, like, okay, what is the lift for getting these Slack reports ready for…
62 00:08:35.400 ⇒ 00:08:39.019 Pranav Narahari: use, like, Slack reports ready using MCPs.
63 00:08:39.270 ⇒ 00:08:52.860 Pranav Narahari: I don’t think, like, if it’s gonna take, like, a week of development to get that done, it makes sense to do that, if it means we can just bang out the data warehouse right now, and then do the Slack reports right after, do the rest of the forecasting right after that.
64 00:08:53.220 ⇒ 00:09:07.239 Bobby Palmieri: Yeah, I think as much as we want to get these done, I don’t want to, like, double dip on your hours and our hours, and, like, if we agree that the data warehouse is where these Slack reports should be pulling in from, like, that’s the task at hand, and, like.
65 00:09:07.240 ⇒ 00:09:07.870 Pranav Narahari: Exactly.
66 00:09:07.870 ⇒ 00:09:08.470 Bobby Palmieri: Owned.
67 00:09:08.820 ⇒ 00:09:09.320 Pranav Narahari: Perfect.
68 00:09:09.320 ⇒ 00:09:09.750 Bobby Palmieri: Exactly.
69 00:09:09.750 ⇒ 00:09:11.480 Pranav Narahari: Okay, that’s what I’m…
70 00:09:12.060 ⇒ 00:09:23.690 zacfromson: Yeah, I think that makes sense. I mean, I think the MCP is just a call back and forth for, like, on-the-rip things. I think everything else, probably, from… as technically savvy as I am, which is low, is data warehouse.
71 00:09:23.690 ⇒ 00:09:27.540 Pranav Narahari: Yeah, and I would say, like, MCPs are obviously getting a lot of…
72 00:09:27.810 ⇒ 00:09:31.910 Pranav Narahari: Praise, but for y’all’s specific,
73 00:09:32.540 ⇒ 00:09:46.430 Pranav Narahari: for y’all’s, like, specific use case, like, all the data is coming into the data warehouse, we’re getting a fresh sync every single day. The data that you’re querying, I don’t think is gonna be on data from just, like, an hour ago, right? You’re probably gonna be querying on the data from
74 00:09:46.810 ⇒ 00:09:58.569 Pranav Narahari: days ago, or the day before. And so, that’s something I’m gonna talk to Sam about, too, is, like, what is the actual benefit we’re getting from these MCPs? And so there may be, like, these…
75 00:09:58.780 ⇒ 00:10:08.009 Pranav Narahari: for these, like, ad hoc, like, questions that we’re asking it, it may do a better job of pulling in the data from the data warehouse.
76 00:10:08.520 ⇒ 00:10:14.720 Pranav Narahari: But for things like Slack reports, for forecasting, we’ve already decided, you know, data warehouse is where we need to be.
77 00:10:14.720 ⇒ 00:10:31.549 zacfromson: Yeah, I think the interesting piece for us are, like, contextual questions, like helping us with strategy, or, like, random ad hoc stuff that we don’t have reports built out for. Like, I ask it, like, what are all the active flows? What are their purposes? Like, what are all the segmentation we’ve used in the last 30 days? Like, why do you think CPM is up last.
78 00:10:31.550 ⇒ 00:10:32.010 Pranav Narahari: Right.
79 00:10:32.010 ⇒ 00:10:56.809 zacfromson: 7 days before that, so it’s, like, more than just, like, here is the data, where, like, the reports for Slack are just, like, here is the data, there’s no context or insight, versus, I think, the MCPs, where there’s, like, a contextual question being asked that isn’t necessarily set in an automated report, which that’s helpful. It’s like, hey, like, help me strategize, like, a plan for my email marketing calendar for the month of March, and it’s like, here are some things that we did last March that worked really well, here are some holidays, like.
80 00:10:56.900 ⇒ 00:11:08.189 zacfromson: Right, so I think when it’s, like, more, like, conversational-type stuff to be had, that’s, like, MCP for us. I think anything static, I don’t… I mean, I’ll leave it up to you guys what you guys think is best, but,
81 00:11:08.280 ⇒ 00:11:24.920 zacfromson: But yeah, I think that’s kind of, like, the differentiator between, like, the static stuff, or the stuff that’s updated daily, which is, like, forecasting a Slack report versus MCP, which are these, like, random ad hoc asks, or, like, more contextual strategy things that, like, are trying to help our team plan or, like, answer questions.
82 00:11:25.060 ⇒ 00:11:27.440 Pranav Narahari: I think maybe the better way for me to, like…
83 00:11:27.610 ⇒ 00:11:41.310 Pranav Narahari: like, illustrate, like, what I’m trying to figure out is the best way going forward, is should we be pulling the data ad hoc every time we’re asking a query from, let’s say, Shopify, Meta, Klaviyo? Or should it just be
84 00:11:41.480 ⇒ 00:11:54.570 Pranav Narahari: querying our data warehouse, because our data warehouse has the data, and so it’s basically all-encompassing of all the data that we’re asked… that people could be asking for, is my…
85 00:11:54.620 ⇒ 00:12:03.380 Pranav Narahari: like, this is my understanding right now. Sam has been working with the MCPs much more closely, and so that’s just something I’ll double-check with him, but…
86 00:12:03.380 ⇒ 00:12:16.380 Bobby Palmieri: My thought is Slack reports and forecasting, definitely data warehouse. I think in terms of, like, next steps, like, let’s just keep the MCP for chat, just because I just want to put that in the done column, and then…
87 00:12:16.380 ⇒ 00:12:25.550 Bobby Palmieri: we want to revisit it down the line, we can. My guess is that we’re not going to pull everything into the data warehouse, so, like, the MCP will still be needed. Like, if I want to
88 00:12:25.550 ⇒ 00:12:36.680 Bobby Palmieri: Yeah. We spend on, like, Instagram Stories placements yesterday, like, I don’t know if we’ll be pulling in placement data into the data warehouse, right? Because we… we pay on, like, a per…
89 00:12:38.120 ⇒ 00:12:40.410 Bobby Palmieri: the row of data, right? So…
90 00:12:40.410 ⇒ 00:12:41.050 Pranav Narahari: Right.
91 00:12:41.060 ⇒ 00:12:44.729 Bobby Palmieri: I think the MCPs will be useful for chat. Yeah.
92 00:12:44.920 ⇒ 00:12:47.939 Pranav Narahari: I think you’re probably… there’s probably something…
93 00:12:48.240 ⇒ 00:13:01.510 Pranav Narahari: that the MCPs are pulling in, that the data warehouse isn’t… doesn’t necessarily need to have saved within it. So, yeah, that sounds good. Slack reports, likely, though, are going to be data warehouse, so in terms of order of operations.
94 00:13:01.510 ⇒ 00:13:12.199 Bobby Palmieri: consistent report each day, like, we know what data that is, and really, it’s, like, spend and, like, maybe 4 metrics per channel. Like, we’ll definitely want that in the data warehouse.
95 00:13:12.330 ⇒ 00:13:13.420 Pranav Narahari: Yeah, yep.
96 00:13:13.680 ⇒ 00:13:23.469 Pranav Narahari: Okay, so I think high level, we kind of understand what that is going to look like in terms of week-to-week. Data warehouse first, then,
97 00:13:23.620 ⇒ 00:13:25.160 Bobby Palmieri: And then maybe…
98 00:13:25.160 ⇒ 00:13:30.839 Pranav Narahari: in parallel, we’ll see, like, what the MCP lift is to kind of get us to 100%. So…
99 00:13:30.970 ⇒ 00:13:43.309 Pranav Narahari: this is a good conversation to have, just to make sure we’re aligned there. I’m gonna just work with Sam and Casey, just to, like… probably just, like, what I’ll do is recreate a Gantt chart for y’all, just to see, like, okay, week to week, what are we gonna be working on? .
100 00:13:43.520 ⇒ 00:13:53.100 Bobby Palmieri: And I was able to get Meta connected to Polytomic for one account, too. Okay. You have Newton data in there. I need… I think I need you to unsync
101 00:13:53.500 ⇒ 00:14:02.269 Bobby Palmieri: Google Ads. There’s, like, a Brainforge test in there. I think you need to disconnect, and then I can connect.
102 00:14:02.270 ⇒ 00:14:03.180 Pranav Narahari: Okay.
103 00:14:03.520 ⇒ 00:14:11.179 Pranav Narahari: Perfect, yeah, I think Sam was probably messing around with that to just get some data pulled in, so I’ll, I’ll uncheck that.
104 00:14:11.360 ⇒ 00:14:12.310 Bobby Palmieri: Okay, cool.
105 00:14:12.310 ⇒ 00:14:28.449 Pranav Narahari: Cool. That was mostly what I wanted to talk about in terms of priorities. So, what I’ll do next, though, is just, yeah, I’ll talk to Sam and Casey, get that Gantt chart out, and then make sure everything looks good, and then we can just, like, yeah, like you guys said, just start pushing things into the done column.
106 00:14:28.780 ⇒ 00:14:46.609 Bobby Palmieri: Cool. Yeah, I mean, I think from… I mean, yeah, connect with Sam and let me know what you guys think is next steps. For me, it’s, like, those quick shifts to Google Ads and the rest of the MCPs, and then, like, I think we need to spend the next however long getting the data warehouse set up, because that’s going to do forecasting and everything else.
107 00:14:46.840 ⇒ 00:14:48.010 Pranav Narahari: Yeah, exactly.
108 00:14:49.150 ⇒ 00:14:49.670 Bobby Palmieri: Perfect.
109 00:14:49.670 ⇒ 00:14:50.250 Pranav Narahari: Nice.
110 00:14:50.620 ⇒ 00:14:51.509 Bobby Palmieri: Appreciate the time.
111 00:14:51.610 ⇒ 00:14:52.160 Pranav Narahari: Yeah.
112 00:14:52.160 ⇒ 00:14:54.250 Bobby Palmieri: Thanks, guys. Talk soon.
113 00:14:54.290 ⇒ 00:14:55.050 Pranav Narahari: See ya.