Meeting Title: Brainforge x D&G | Demo Call Date: 2026-03-03 Meeting participants: Luke Scorziell, IT Department, Luke’s Notetaker, peter.bassett, Pranav Narahari
WEBVTT
1 00:00:16.420 ⇒ 00:00:18.129 Joshua Dent: Hey, how’s it going, Luke?
2 00:00:18.960 ⇒ 00:00:20.229 Luke Scorziell: Josh, how are you?
3 00:00:20.230 ⇒ 00:00:20.920 Joshua Dent: Good.
4 00:00:21.340 ⇒ 00:00:23.049 Luke Scorziell: How’s your, your week?
5 00:00:23.190 ⇒ 00:00:24.330 Luke Scorziell: We’re getting started.
6 00:00:24.570 ⇒ 00:00:25.380 Joshua Dent: Yeah, it’s going nice.
7 00:00:26.680 ⇒ 00:00:32.930 Luke Scorziell: Cool. Well, I can… Yeah, I’ll just get started, and then as…
8 00:00:33.210 ⇒ 00:00:35.979 Luke Scorziell: Thanks for not finishing up a call, and then, Ethan…
9 00:00:36.590 ⇒ 00:00:39.329 Luke Scorziell: He can hop on, so,
10 00:00:39.550 ⇒ 00:00:46.579 Luke Scorziell: Yeah, I guess, like, just wanted to check before we started if you have any questions or anything that kind of came up in between this week and last week that we can…
11 00:00:47.320 ⇒ 00:00:48.920 Luke Scorziell: answer. Otherwise, I need to…
12 00:00:48.920 ⇒ 00:00:51.780 Joshua Dent: Yeah, yeah, yeah. You know, there is,
13 00:00:51.980 ⇒ 00:00:56.949 Joshua Dent: So, I was brought in to an internal AI group.
14 00:00:57.180 ⇒ 00:01:05.240 Joshua Dent: In the past week that’s working on some stuff, that’s more client-focused. You know, it… it…
15 00:01:05.340 ⇒ 00:01:09.359 Joshua Dent: It’s kind of hard to describe, it’s sort of about, like.
16 00:01:09.550 ⇒ 00:01:16.249 Joshua Dent: Harnessing our internal resources for, you know, client deployments and client development stuff.
17 00:01:16.730 ⇒ 00:01:24.830 Joshua Dent: And so I was hoping we could speak a little bit more about that, too. Peter is, one of the guys that’s, you know, leading that group, so…
18 00:01:25.350 ⇒ 00:01:26.990 Luke Scorziell: Yeah, good to meet you, Peter.
19 00:01:27.160 ⇒ 00:01:32.130 Luke Scorziell: Good. Yeah, excited to chat with you guys. Yeah, I mean, I think, like.
20 00:01:33.030 ⇒ 00:01:50.070 Luke Scorziell: I was just talking to our founder, and it’s like, every dollar that we put into… or you guys put into us, like, we want to be able to give you 5X back, and I think, like, one of the biggest ways that we see that is, once we start working together, with clients, like, we’re able to see,
21 00:01:50.440 ⇒ 00:01:54.609 Luke Scorziell: This is Fernand, who’s our… one of our lead AI engineers.
22 00:01:54.750 ⇒ 00:02:02.799 Luke Scorziell: But, yeah, we’re typically able to find other areas that we can really help you in the business, and so I would be like this with, like.
23 00:02:02.930 ⇒ 00:02:05.750 Luke Scorziell: This is, like, the problem that you’re probably feeling
24 00:02:06.080 ⇒ 00:02:11.889 Luke Scorziell: Right now, most pressing, and we can, you know, solve this for you, and then…
25 00:02:12.230 ⇒ 00:02:27.110 Luke Scorziell: Yeah, like, we’re just at the tip of the iceberg with the problems that we can solve, so if it’s building out client-facing resources, or, yeah, whatever that looks like at David and Goliath, we’d be, yeah, more than happy to chat about that.
26 00:02:27.240 ⇒ 00:02:28.200 Joshua Dent: Okay. Wow.
27 00:02:28.720 ⇒ 00:02:29.530 Luke Scorziell: Cool.
28 00:02:30.210 ⇒ 00:02:40.680 Joshua Dent: You know, before we get started, it’d be great if you could just give Peter a rundown of Brainforge and what you guys do, so he has a better idea of what could be done.
29 00:02:41.320 ⇒ 00:02:55.560 Luke Scorziell: Yeah, so, I’m Luke, and obviously Pranav, and I came from kind of a marketing brand background, went to USC, for journalism, and then
30 00:02:55.690 ⇒ 00:03:06.619 Luke Scorziell: kind of ran my own marketing agency for a little bit, worked in investment research company, kind of helping, yeah, just learn… learn a lot about the landscape, I guess, through that, and then…
31 00:03:07.290 ⇒ 00:03:21.719 Luke Scorziell: joined Brainforge, this past year, and we… I guess, essentially, like, the line is, like, we help Forge your company’s brain. So we’re helping companies really transform into more intelligent companies, and so…
32 00:03:21.720 ⇒ 00:03:40.290 Luke Scorziell: a lot of the pain points that, like, our clients typically struggle with are, like, manual information transfer, documentation, movement, so moving, like, information from one document to another, or having to go back into files and look it up. And so those are, like, chains that typically surface, but then we,
33 00:03:40.380 ⇒ 00:03:44.969 Luke Scorziell: Are also able to see, like, create a lot of return on… on…
34 00:03:45.120 ⇒ 00:03:50.729 Luke Scorziell: Like, kind of creating, like, centralized knowledge bases that your team can access.
35 00:03:50.950 ⇒ 00:03:55.880 Luke Scorziell: Like, type of a, you know, just swiping the keyboard and asking a question.
36 00:03:56.020 ⇒ 00:04:00.360 Luke Scorziell: So, BrandForge itself, been around for…
37 00:04:00.600 ⇒ 00:04:06.860 Luke Scorziell: 3 years, and we… we’ve got a whole AI team that all we’re thinking about is
38 00:04:06.980 ⇒ 00:04:15.239 Luke Scorziell: AI all the time, and then we have a kind of a data background and, like, foundation to the business, and so that…
39 00:04:15.380 ⇒ 00:04:22.379 Luke Scorziell: Gives us, like, you know, we’re able to deal a lot with privacy issues, and making sure that the data stays internal to
40 00:04:22.580 ⇒ 00:04:29.520 Luke Scorziell: our clients’ companies, and then bring the AI into that, as opposed to, like, what a lot of
41 00:04:29.620 ⇒ 00:04:45.690 Luke Scorziell: AI for the last few years has been is bringing your data into a server that you have no idea where the information, is really going. And so that’s kind of a Brainforge intro. I guess, Pranav, maybe you could speak to some of the work that I know you’ve done with,
42 00:04:46.130 ⇒ 00:04:49.710 Luke Scorziell: Yeah, with similar clients, and just your, kind of, experience with AI.
43 00:04:50.260 ⇒ 00:05:06.019 Pranav Narahari: Totally, yeah. Nice to meet you, Peter. Nice to see you again, Josh. Yeah, so some of my experience, I’ll be a little bit brief about just, like, my background, too. Yeah, so software engineering background, worked in banking for 3 years on their
44 00:05:06.020 ⇒ 00:05:11.299 Pranav Narahari: Cloud engineering effort, kind of just pushing applications from on-prem servers into the cloud.
45 00:05:11.300 ⇒ 00:05:18.009 Pranav Narahari: Not super exciting. After that, got a lot more into, AI engineering back in,
46 00:05:18.070 ⇒ 00:05:29.490 Pranav Narahari: 2023 was, part of, like, you know, kind of when that craze happened with, like, when ChatGPT dropped in December 2022, I believe it was. I had a…
47 00:05:29.490 ⇒ 00:05:38.540 Pranav Narahari: friend from back then that was actually, like, working on a lot of these, like, natural language models, and so he really, like, was hyping it up to me.
48 00:05:38.540 ⇒ 00:05:52.969 Pranav Narahari: soon thereafter, I started using it for coding, so with, like, GIP, Copilot, things of that nature, and that’s what kind of got me into, like, this career where, working at Brainforge, I feel like they’re kind of doing it at the highest level, which is, kind of onboarding
49 00:05:52.970 ⇒ 00:05:55.059 Pranav Narahari: Clients into, like, the…
50 00:05:55.060 ⇒ 00:06:10.069 Pranav Narahari: the AI world. So, a lot of these… our clients are already using, like, ChatGPT, Gemini, Anthropic for just, like, copy-pasting stuff, maybe emails already, but really, like, that last-mile integration of things is, like.
51 00:06:10.070 ⇒ 00:06:16.369 Pranav Narahari: where I feel like we really excel, because we’re really on top of, like, what are the latest, tools out there, what are…
52 00:06:16.370 ⇒ 00:06:30.090 Pranav Narahari: the best products to build behind these. Also, what are the limitations behind these, is super important as well. And so, yeah, working here has been perfect for me, because I already had that interest, and now I’m just working alongside people that
53 00:06:30.090 ⇒ 00:06:34.829 Pranav Narahari: share that same interest. And so, yeah, similar clients that we worked with,
54 00:06:35.350 ⇒ 00:06:44.789 Pranav Narahari: in terms of, like, the products we’ve built for them, like Luke mentioned, like, building them knowledge bases, and so kind of allowing them to,
55 00:06:45.040 ⇒ 00:06:59.949 Pranav Narahari: query their own information that they have defined as the source of truth. They don’t really care, essentially, like, what ChatGPT says as the source of truth based on the crawling of the internet. And so building an AI system that uses that as, like, the grounding,
56 00:07:00.040 ⇒ 00:07:14.130 Pranav Narahari: truth of, like, what is backing all of their insights, that’s something we’ve built. Also, integration with, like, a bunch of other, applications as well, using MCP servers is, other things that we’ve also built.
57 00:07:15.340 ⇒ 00:07:16.000 Luke Scorziell: Yeah.
58 00:07:18.160 ⇒ 00:07:18.820 Luke Scorziell: Born.
59 00:07:19.000 ⇒ 00:07:21.270 Luke Scorziell: Pretty lost.
60 00:07:21.270 ⇒ 00:07:24.280 peter.bassett: Very cool, thanks so much for the background info.
61 00:07:25.300 ⇒ 00:07:30.190 Luke Scorziell: Yeah, of course. And Peter, what’s your role again at David and Goliath? I don’t know.
62 00:07:30.320 ⇒ 00:07:32.950 peter.bassett: So I head up the…
63 00:07:33.420 ⇒ 00:07:49.260 peter.bassett: the, like, integrated production and technology department, so we, generally are the ones who are doing anything in, like, the digital space, social, content creation, technology,
64 00:07:49.260 ⇒ 00:07:49.980 Luke Scorziell: Yeah.
65 00:07:49.980 ⇒ 00:07:56.729 peter.bassett: From, like, a production standpoint, so bringing the ideas to life, and… Also, Josh and I are…
66 00:07:57.640 ⇒ 00:08:06.220 peter.bassett: for lack of a better term, sort of, like, part of the AI steering committee at David and Goliath, like, trying to really figure out
67 00:08:06.410 ⇒ 00:08:08.950 peter.bassett: ways that we can use AI in…
68 00:08:09.210 ⇒ 00:08:13.340 peter.bassett: Our everyday… like, we know,
69 00:08:13.550 ⇒ 00:08:22.929 peter.bassett: Like you guys both mentioned, people are using chatbots and stuff like that, so we’re trying to really take it to the next step, where we can actually start
70 00:08:23.160 ⇒ 00:08:28.729 peter.bassett: going beyond the hype of what AI can do, and actually starting to see what
71 00:08:29.200 ⇒ 00:08:39.199 peter.bassett: it… it, it will do, I guess, to help us in our day-to-day lives. So, like, Josh and I were talking yesterday about some of the…
72 00:08:39.539 ⇒ 00:08:48.390 peter.bassett: Implementations that were… Working on, whether it’s like, building… simple…
73 00:08:48.790 ⇒ 00:09:08.480 peter.bassett: plugins or something, like, code-assisted plugins where I wouldn’t need to hire an external developer, or a team of external developers to, like, develop a Figma plugin. You know, like, I know enough about coding myself, and have used coding agents over the past year as they’ve become actually useful.
74 00:09:08.680 ⇒ 00:09:12.819 peter.bassett: Like, last night, Josh, just so you know, I did…
75 00:09:13.380 ⇒ 00:09:23.589 peter.bassett: I did spin up a plugin for Figma that did what we were talking about yesterday. It’s not perfect, but it took me…
76 00:09:23.590 ⇒ 00:09:25.869 Joshua Dent: Several hours to get…
77 00:09:25.870 ⇒ 00:09:32.680 peter.bassett: an MVP that I can then show to the designers and be like, hey, Would this tool save you…
78 00:09:32.780 ⇒ 00:09:37.519 peter.bassett: Hours and hours of design time by generating all of these, like.
79 00:09:37.950 ⇒ 00:09:41.159 peter.bassett: Different sizes of banner ads, or whatever it ends up being.
80 00:09:41.430 ⇒ 00:09:46.510 peter.bassett: So that’s… that’s the kind of stuff that we’re really interested in doing. I think, you know, we’re…
81 00:09:46.630 ⇒ 00:09:59.259 peter.bassett: when you talk to folks like Josh and I, you’re talking to folks who have a pretty deep understanding of what’s going on in AI and what tools are out there, I think our hardest job, and Josh, jump in here.
82 00:09:59.440 ⇒ 00:10:01.099 peter.bassett: is,
83 00:10:01.230 ⇒ 00:10:09.899 peter.bassett: Is, like, merchandising what we can do internally and getting support, especially from, like, senior leadership about what to do, so…
84 00:10:09.900 ⇒ 00:10:10.550 Luke Scorziell: Yeah.
85 00:10:10.910 ⇒ 00:10:12.350 peter.bassett: Anyway, that’s…
86 00:10:12.350 ⇒ 00:10:13.390 Luke Scorziell: Yeah, it’s my little…
87 00:10:14.050 ⇒ 00:10:16.140 Joshua Dent: Getting their buy-in, for sure, they’re difficult.
88 00:10:16.530 ⇒ 00:10:23.260 Luke Scorziell: We can… we can get to that. I want to show you what we built, just because it’s exciting, and I think I’d love to get into some of those.
89 00:10:23.550 ⇒ 00:10:27.119 Luke Scorziell: conversations as well, because I think,
90 00:10:27.220 ⇒ 00:10:32.279 Luke Scorziell: Yeah, like, I’m from a marketing kind of brand background, and it’s like.
91 00:10:32.800 ⇒ 00:10:40.879 Luke Scorziell: Yeah, it’s exciting. It’s an exciting time to get to be here, and even just, like, this is a demo that I spun up myself, too. It’s no,
92 00:10:41.030 ⇒ 00:10:48.670 Luke Scorziell: I don’t have, like, the coding background, so I would not be the one working on your projects, because I don’t have the development expertise that,
93 00:10:48.940 ⇒ 00:10:50.509 Luke Scorziell: Pranab does, but
94 00:10:50.870 ⇒ 00:11:01.689 Luke Scorziell: Yeah, so just kind of sharing the screen, we built this, just to kind of look kind of like it’s in Teams, so I know this isn’t, like.
95 00:11:02.340 ⇒ 00:11:05.539 Luke Scorziell: Exactly, James, but I know, like, the biggest problem that
96 00:11:06.000 ⇒ 00:11:18.150 Luke Scorziell: Yeah, you had mentioned last week, Josh, the tools fall off quickly, and that it’s like, hey, here’s this, like, beautiful new, like, AI thing that promises to solve all your problems, and then
97 00:11:18.230 ⇒ 00:11:38.149 Luke Scorziell: you know, people log onto it, they see that it doesn’t really work as fast as they want, or they have to learn a new platform, and they just ditch it, and then it kind of goes, like, to put. So… so that was the pain point that we were listening to, and then another pain point was just being able to connect softwares like Kanto, FileStage, Frame.io, and your entire Adobe tenant within this, so…
98 00:11:38.330 ⇒ 00:11:40.579 Luke Scorziell: Want to show both of those,
99 00:11:40.970 ⇒ 00:11:52.960 Luke Scorziell: And yeah, I mean, I think the biggest thing we would like to show is just that this can save massive amounts of time and energy, and help you focus more effort onto strategic activities.
100 00:11:53.340 ⇒ 00:11:58.110 Luke Scorziell: And so, yeah, if we kind of open up here, this is, like, the mimic for…
101 00:11:58.320 ⇒ 00:12:01.780 Luke Scorziell: what Copilot would be, and so…
102 00:12:02.060 ⇒ 00:12:04.560 Luke Scorziell: You can see up here, these are the different MCP
103 00:12:04.700 ⇒ 00:12:06.690 Luke Scorziell: Servers that would be checked, so…
104 00:12:06.840 ⇒ 00:12:11.940 Luke Scorziell: If we just kind of start, like, right off the bat with just Adobe.
105 00:12:12.110 ⇒ 00:12:18.680 Luke Scorziell: I can say, like, I find all of the video… assets…
106 00:12:19.080 ⇒ 00:12:26.900 Luke Scorziell: Well, from… and I know you guys work with Kia, the Kia… Summer 2025 campaign.
107 00:12:28.030 ⇒ 00:12:31.650 Luke Scorziell: And I ask that, and then it just spits out right off the bat.
108 00:12:31.910 ⇒ 00:12:38.070 Luke Scorziell: I mean, it’d be a little… little slower, just because it’s querying all of your
109 00:12:38.500 ⇒ 00:12:43.129 Luke Scorziell: This, but then if… you know, so then you, you get right there, the two.
110 00:12:43.340 ⇒ 00:12:50.919 Luke Scorziell: brand assets, but then if you want to ask the same thing, but you want to also add in Frame.io and Kanto, you can say.
111 00:12:51.730 ⇒ 00:12:55.300 Luke Scorziell: Fine, can you show me all…
112 00:12:56.130 ⇒ 00:13:00.690 Luke Scorziell: of the video assets from the Kia summer.
113 00:13:01.920 ⇒ 00:13:02.920 Luke Scorziell: I’m saying.
114 00:13:06.810 ⇒ 00:13:14.910 Luke Scorziell: And now you’ve got, your files from Adobe, from Frame.io, and… and from Kanto.
115 00:13:15.070 ⇒ 00:13:24.520 Luke Scorziell: So that’s kind of the first thing I wanted to show. I was curious, just from your perspective, like, how much time you guys see your team currently spending signing those assets.
116 00:13:24.850 ⇒ 00:13:34.439 Joshua Dent: I mean, I really like that. That’s really great to be able to search across all of our different, you know, sort of data lakes that live independently.
117 00:13:36.090 ⇒ 00:13:36.670 Luke Scorziell: Yeah.
118 00:13:38.070 ⇒ 00:13:42.240 Luke Scorziell: Best, cool. And then,
119 00:13:42.890 ⇒ 00:13:47.179 Luke Scorziell: Yeah, so we’ve got that, and this is something that I’ll… yeah, I kinda gotta…
120 00:13:47.910 ⇒ 00:14:04.989 Joshua Dent: And let me ask you this, Luke, so when it comes to… obviously, not everyone has access to everything, for security reasons. Yeah. So we could individualize, like, what areas people are able to search into, using this… something like this?
121 00:14:04.990 ⇒ 00:14:19.090 Luke Scorziell: Yeah, yeah, you can do role-based permissions, and it… we could totally design you a system where, like, IT has access to IT data and information, legal has access to their stuff, and then even across the different teams, like your,
122 00:14:19.250 ⇒ 00:14:23.080 Luke Scorziell: Yeah, so that can be part of the work, too, that we scope out.
123 00:14:23.220 ⇒ 00:14:27.270 Luke Scorziell: Within that. So, so that’s, like, too, like, if you look here.
124 00:14:27.420 ⇒ 00:14:44.200 Luke Scorziell: you know, the first one I did, we’re only searching Adobe Creative Cloud. You could, like… and then this one, we added in the other two. So if you wanted someone only to have access to, like, certain softwares, you can kind of set those permissions. So, yeah, does that kind of answer that question for you?
125 00:14:44.360 ⇒ 00:14:45.050 Joshua Dent: Yeah.
126 00:14:45.410 ⇒ 00:14:52.470 Luke Scorziell: Cool. Yeah, okay, and then the other, you know.
127 00:14:52.470 ⇒ 00:14:53.610 peter.bassett: Like, before we…
128 00:14:53.610 ⇒ 00:14:54.800 Luke Scorziell: Yeah, yeah, yeah.
129 00:14:54.800 ⇒ 00:15:00.790 peter.bassett: Before we move on, how… can you just give me a 2-second…
130 00:15:00.900 ⇒ 00:15:10.219 peter.bassett: explanation of how that… how this system works? Like, what’s the architecture behind it? Just briefly. We don’t need to go crazy.
131 00:15:10.220 ⇒ 00:15:12.850 Luke Scorziell: Pranav, maybe if you can give a quick explanation.
132 00:15:13.010 ⇒ 00:15:27.699 Pranav Narahari: Yeah, yeah. So, there’s two things kind of underpinning, like, this whole thing, which is basically pulling in all this knowledge. We’ve mentioned it before, MCP servers and knowledge bases. So for static information, maybe data that you guys have curated.
133 00:15:27.700 ⇒ 00:15:39.159 Pranav Narahari: And you just have it in, like, some directory or some file store within your company, that can be something that’s integrated into a knowledge base. And so, yeah, super high level, we don’t need to go into super depth.
134 00:15:39.160 ⇒ 00:15:47.769 Pranav Narahari: Basically, you create certain indexing on those files, depending on the file type, and then the LLM is then able to
135 00:15:47.940 ⇒ 00:16:04.990 Pranav Narahari: assess which ones are most similar to the query that you have provided into this chatbot. And so that’s how, in, like, the demo that Luke is showing, it’s showing these ones referring to Kia Summer 2025. That would be based on the indexing that you did, which we would do.
136 00:16:04.990 ⇒ 00:16:15.210 peter.bassett: Okay, so that, like, all of that… that tagging and indexing and all that kind of stuff, that would happen as part of the setup of a tool like this?
137 00:16:15.550 ⇒ 00:16:22.459 peter.bassett: Or does that happen… Natively, like, as new assets are added and things like that.
138 00:16:22.840 ⇒ 00:16:31.669 Pranav Narahari: Yeah, so as new assets are added, too, we would create, like, some pipeline that basically, creates the metadata for it so that it can be indexed.
139 00:16:31.890 ⇒ 00:16:36.389 Pranav Narahari: If the more technical kind of implementation, like, we create embeddings, if you’ve…
140 00:16:36.410 ⇒ 00:16:49.880 Pranav Narahari: If you’re familiar with that, like, kind of creating embeddings for each of these files. And then based on those embeddings, that’s how we would be able to assess which ones are relevant to the query that was…
141 00:16:49.880 ⇒ 00:17:02.469 Pranav Narahari: ask to the chatbot. And so that’s knowledge bases. MCP servers, how that… how those work is that it actually is, like, you can think of it as, like, a wrapper around an API so that an LLM can…
142 00:17:02.470 ⇒ 00:17:20.120 Pranav Narahari: interact. So, we basically assess, like, for some of these, I looked into FileStage specifically, and I saw FileStage has an API, and so we would basically create an MCP server for you guys, deploy that, and then connect it to the chatbot.
143 00:17:21.520 ⇒ 00:17:21.859 peter.bassett: Cool.
144 00:17:21.869 ⇒ 00:17:22.519 Luke Scorziell: Nice.
145 00:17:22.989 ⇒ 00:17:23.759 Luke Scorziell: Bye.
146 00:17:24.469 ⇒ 00:17:31.819 Luke Scorziell: So, yeah, I can kind of… So, let’s see… Have we…
147 00:17:32.659 ⇒ 00:17:48.159 Luke Scorziell: check on these other two, so yeah, these would be the MCP servers, and this is kind of also what we would build, for you guys, because this may or may not currently exist. And so, yeah, if I check on, like, File Stage, and it’s like, I want to get feedback.
148 00:17:48.279 ⇒ 00:17:53.999 Luke Scorziell: And I’ll kind of show you a cool little feature that I built just from a brand.
149 00:17:54.209 ⇒ 00:17:57.049 Luke Scorziell: perspectives. Obviously, David and Goliath, you guys have…
150 00:17:57.189 ⇒ 00:18:00.809 Luke Scorziell: You know, the slingshot is how David killed Goliath.
151 00:18:01.089 ⇒ 00:18:05.429 Luke Scorziell: And so, we go into this DM,
152 00:18:05.699 ⇒ 00:18:12.239 Luke Scorziell: you know, Sarah asked, hey, Maria, just ping me, she wants to know where we stand on the Kia approvals before the 11 a.m. sync.
153 00:18:12.499 ⇒ 00:18:13.809 Luke Scorziell: Do you have eyes on that?
154 00:18:14.229 ⇒ 00:18:22.249 Luke Scorziell: We would… this could work in channels, I think, as well. We’d have to kind of see, but yeah, if I go, say, at Slingshot.
155 00:18:22.809 ⇒ 00:18:29.219 Luke Scorziell: And then say, what’s the latest client feedback?
156 00:18:29.409 ⇒ 00:18:33.179 Luke Scorziell: on TIOs, EV spot deliverables.
157 00:18:34.459 ⇒ 00:18:36.519 Luke Scorziell: In bile stage.
158 00:18:37.709 ⇒ 00:18:45.059 Luke Scorziell: And I asked that… Then, you know, right here in the chat, slingshot
159 00:18:45.299 ⇒ 00:18:48.139 Luke Scorziell: The bot, you know, comes and… and…
160 00:18:48.359 ⇒ 00:18:54.939 Luke Scorziell: gives you the information that you need. So this connects with FileSage, you can see the approvals, kind of see,
161 00:18:55.459 ⇒ 00:19:00.219 Luke Scorziell: Yeah, what you’re getting there. And so that’s, again, kind of to that main pain point that you hear…
162 00:19:00.219 ⇒ 00:19:02.469 Joshua Dent: you were concerned about, Josh, of like…
163 00:19:02.489 ⇒ 00:19:17.469 Luke Scorziell: would people use this? It’s like, well, you can ask, I guess, like, would you rather just add Slingshot or, or go and, like, search through the emails and notifications and all that stuff? And so, yeah, hopefully this is pretty native.
164 00:19:17.519 ⇒ 00:19:23.439 Luke Scorziell: I guess, to… to the processes that… that you guys are seeing. And this is something, too, that, like.
165 00:19:24.199 ⇒ 00:19:26.739 Luke Scorziell: Again, I don’t come from a technical background, really.
166 00:19:26.969 ⇒ 00:19:44.679 Luke Scorziell: And I use… we have all this stuff built out at Brainforge, for ourselves, and it’s like, I use our knowledge base every day. I use, like, Cursor to do all the stuff. We have a Brainforge bot that we can trigger and ask questions to, and so…
167 00:19:44.929 ⇒ 00:19:55.949 Luke Scorziell: As, like, a non-technical user, probably closer to the average person at, Ibn and Goliath, like, I, yeah, I would be, kind of your…
168 00:19:56.489 ⇒ 00:19:59.099 Luke Scorziell: Yeah, just very excited to use a lot of this stuff.
169 00:20:00.660 ⇒ 00:20:12.080 Joshua Dent: So could you… could you page this chatbot into meetings that are going on? You know, someone has a question, and everyone’s on like this, and you tag it and ask a question, it’ll respond inside the meeting?
170 00:20:13.240 ⇒ 00:20:14.150 Luke Scorziell: Yeah.
171 00:20:14.810 ⇒ 00:20:15.460 Pranav Narahari: Meeting chat.
172 00:20:16.930 ⇒ 00:20:18.320 Joshua Dent: Yeah, meeting chat.
173 00:20:18.320 ⇒ 00:20:26.260 Pranav Narahari: Yeah, specifically, I just need to look into the… the team’s, like, SDK.
174 00:20:27.320 ⇒ 00:20:44.739 Pranav Narahari: But I believe so. So, we’ve built a ton of Slack bots at Brainforge, and so that is something that we can definitely do. Like, Slack right now has, like, an AI transcription, feature, within channel, and those are usually, like.
175 00:20:44.940 ⇒ 00:20:50.979 Pranav Narahari: the chat’s… tied to the… the main thread as well, and so…
176 00:20:51.330 ⇒ 00:20:56.190 Pranav Narahari: I totally see we can do something, if not exactly that, something similar to that.
177 00:20:56.960 ⇒ 00:20:57.790 Joshua Dent: Okay.
178 00:20:58.270 ⇒ 00:21:06.670 Luke Scorziell: And even with this stuff, it’s like, you can just start your own chat with yourself, and then just be messaging back and forth, and… so this is… this would be in Teams, so.
179 00:21:06.930 ⇒ 00:21:07.790 Joshua Dent: Yeah.
180 00:21:07.790 ⇒ 00:21:19.869 Luke Scorziell: And then, just… this is kind of the last one that, you know, I’ll show, is, like, you mentioned the archives, and again, like, I think you guys are actually in a really unique position, because companies that have been around for longer.
181 00:21:20.040 ⇒ 00:21:33.959 Luke Scorziell: and have more data and archive information, have a bigger moat around them as we go into the AI age, because you… like, right now, all the LLMs and whatnot have been trained on, like, mainly Reddit, honestly.
182 00:21:34.130 ⇒ 00:21:36.279 Luke Scorziell: Especially with ChatGPT, and…
183 00:21:36.460 ⇒ 00:21:52.230 Luke Scorziell: But what’s coming is that companies are going to be using their own information and data, and so David and Goliath is actually uniquely positioned to be able to leverage, like, the years of experience and decades of experience that an archive that you guys have into your own database, and so…
184 00:21:52.470 ⇒ 00:21:57.010 Luke Scorziell: Or into your own, like, LLM or Slingshot here, so…
185 00:21:57.310 ⇒ 00:22:03.140 Luke Scorziell: like, this is an example of, Priya asks, like, hey, quick question for the summer.
186 00:22:03.270 ⇒ 00:22:09.229 Luke Scorziell: strategy deck, which brand pillar are we leading with for Kia? I want to make sure we’re aligned with our 2023 annual.
187 00:22:09.360 ⇒ 00:22:13.930 Luke Scorziell: I go here and I say, hey, at Slingshot, search.
188 00:22:14.460 ⇒ 00:22:20.869 Luke Scorziell: our archives for the Kia Brand Strategy 2023 Annual.
189 00:22:23.630 ⇒ 00:22:33.049 Luke Scorziell: And then, again, obviously, this would take a little bit longer as it’s indexing, with real files, but it, you know, and again, I don’t know how many files you guys have, but…
190 00:22:33.080 ⇒ 00:22:48.840 Luke Scorziell: like, the example here is 847 documents that I went through to find this, and then you’ve got the Kia Launch Creative Brief, you know, each of these briefs that you’ve kind of done. And the cool thing, too, that I don’t have built out in this, but
191 00:22:48.860 ⇒ 00:22:54.130 Luke Scorziell: You can ask it to, like, analyze trends, or analyze, kind of…
192 00:22:54.230 ⇒ 00:23:05.880 Luke Scorziell: what is… what’s going on between each of these briefs and patterns and whatnot, or you could ask, like, hey, what was the positioning that we used? And it can pull out that information, so instead of having to go back,
193 00:23:06.090 ⇒ 00:23:07.689 Luke Scorziell: And look for it yourself.
194 00:23:07.810 ⇒ 00:23:15.410 Luke Scorziell: this… this can use that. So, it’s essentially like having Copilot, but… Instead of Reddit.
195 00:23:15.610 ⇒ 00:23:18.450 Luke Scorziell: As the… the knowledge base, it’s…
196 00:23:18.610 ⇒ 00:23:22.230 Luke Scorziell: you know, the decades of experience that you guys have. And we can put in
197 00:23:22.690 ⇒ 00:23:29.140 Luke Scorziell: Campaign briefs, meeting transcripts, we can put, Performance reports, like.
198 00:23:29.330 ⇒ 00:23:36.230 Luke Scorziell: Yeah, past client approvals. You can create custom profiles of each client to see, like, what are their preferences.
199 00:23:36.290 ⇒ 00:23:52.389 Luke Scorziell: I could say, hey, what are Keo’s typical preferences when it comes to something like this? So it’s, yeah, it’s pretty wide, wide-ranging in terms of the use cases. So I’ll kind of pause there. I know that’s… that’s kind of the three things, is showing you guys the…
200 00:23:52.580 ⇒ 00:23:54.540 Luke Scorziell: Access to,
201 00:23:54.660 ⇒ 00:24:07.010 Luke Scorziell: the, like, Adobe tenant, client feedback and file stage, and then access to the archived search. So, yeah, I’ll pause there, and if you guys have any questions or feedback or thoughts, would love to…
202 00:24:07.970 ⇒ 00:24:08.790 Luke Scorziell: Here.
203 00:24:09.890 ⇒ 00:24:16.879 Joshua Dent: In terms of the architecture, again, with the MCP servers, I know last time we talked.
204 00:24:16.960 ⇒ 00:24:19.010 Luke Scorziell: An example that you gave.
205 00:24:19.240 ⇒ 00:24:31.060 Joshua Dent: Pranav, was that we would have a server on site here that would be in charge of, you know, handling the queries for each individual service that we want to query.
206 00:24:31.160 ⇒ 00:24:34.930 Joshua Dent: If that is the case, is there…
207 00:24:35.080 ⇒ 00:24:39.270 Joshua Dent: You know, how do we manage it if, you know, there’s…
208 00:24:39.980 ⇒ 00:24:47.529 Joshua Dent: An internet outage here, or a power outage at the building while everyone’s working from home, and, you know, the bot can’t reach the servers.
209 00:24:48.710 ⇒ 00:24:55.770 Pranav Narahari: Gotcha. So, when you say you have them on, on… do you mean, like, on-premises servers? Is that what you
210 00:24:55.910 ⇒ 00:24:56.860 Pranav Narahari: working with.
211 00:24:56.860 ⇒ 00:24:58.960 Joshua Dent: Yeah, that’s what we were talking about last time, yeah.
212 00:24:59.270 ⇒ 00:25:11.490 Pranav Narahari: Yeah, yeah, so if you’re not using, like, you know, like, GCP, AWS, things of that nature, we would definitely just have to have, like, some type of,
213 00:25:11.990 ⇒ 00:25:12.930 Pranav Narahari: kind of…
214 00:25:13.050 ⇒ 00:25:25.350 Pranav Narahari: it’s basically guardrails that we create to say, like, okay, this MCP server is online versus offline, and then that needs to be reflected in the user interface, right? So…
215 00:25:25.350 ⇒ 00:25:33.730 Pranav Narahari: like, how you can see what, like, Luke is building right here, like, when he’s, like, he’s having the ability to click and not click, like, certain sources for information.
216 00:25:33.930 ⇒ 00:25:37.239 Pranav Narahari: Another status would be, like, offline.
217 00:25:37.460 ⇒ 00:25:38.480 Pranav Narahari: Okay.
218 00:25:38.650 ⇒ 00:25:39.410 Pranav Narahari: Yeah.
219 00:25:39.860 ⇒ 00:25:42.229 Joshua Dent: Okay, that’s the easy answer. Alright, thanks.
220 00:25:45.380 ⇒ 00:25:48.459 Luke Scorziell: Peter, any kind of thoughts or questions come up for you?
221 00:25:50.420 ⇒ 00:25:54.039 peter.bassett: No, no, it looks… Looks really cool. Thank you.
222 00:25:54.040 ⇒ 00:25:58.250 Luke Scorziell: Yeah, I’m curious, if you guys, like, are there, like.
223 00:25:58.870 ⇒ 00:26:04.370 Luke Scorziell: Power users that come to mind as you’re looking at this, that you’re like, oh, this person would just…
224 00:26:04.480 ⇒ 00:26:06.050 Luke Scorziell: Eat this up and be like.
225 00:26:06.440 ⇒ 00:26:10.690 Luke Scorziell: You know, we have a ranking within Brainforge, actually, of, like, who’s using
226 00:26:10.830 ⇒ 00:26:16.629 Luke Scorziell: our agents the most. So I’m curious if anyone or, like, a team comes to mind.
227 00:26:16.860 ⇒ 00:26:17.930 Luke Scorziell: For you.
228 00:26:18.490 ⇒ 00:26:25.399 Joshua Dent: I think, our project management team would find something like this incredibly useful, because they have to go digging for files all the time.
229 00:26:26.620 ⇒ 00:26:27.300 Luke Scorziell: Yeah.
230 00:26:33.840 ⇒ 00:26:47.169 peter.bassett: Can… actually, in this search, would you be able to, like, if you knew a little bit more information, like, if I knew Josh had uploaded a specific image or bunches of image to…
231 00:26:47.550 ⇒ 00:26:48.770 peter.bassett: Kanto.
232 00:26:50.090 ⇒ 00:27:01.340 peter.bassett: I could be like, hey, can you let me know the Telluride image that Josh uploaded from the Spring 2024 campaign, or something like that? It could…
233 00:27:01.840 ⇒ 00:27:08.340 peter.bassett: find that data if… only if Kanto has that data available in their API.
234 00:27:10.260 ⇒ 00:27:18.609 Pranav Narahari: Yeah, so if Kanto, like, if you’re able to query based off of, like, let’s say, like, a user’s name,
235 00:27:18.760 ⇒ 00:27:22.360 Pranav Narahari: these… these MCP servers, they’re… how they…
236 00:27:22.730 ⇒ 00:27:32.709 Pranav Narahari: they’re… they allow you to utilize all the context that you give it, and it’s based off of, like, whatever APIs, like, routes that they expose, you’ll use as many as it needs.
237 00:27:32.710 ⇒ 00:27:44.729 Pranav Narahari: To kind of pull in that information. There’s other ways to go about it as well, to, like, if it’s, say, like, the API isn’t as, doesn’t expose as much of the data,
238 00:27:44.960 ⇒ 00:27:57.849 Pranav Narahari: you can create, like, scraping, bots that can maybe work on, like, an interval. And then that… if, say, if you guys have a specific use case that isn’t being, exposed by the API, then we could build that as well.
239 00:27:58.920 ⇒ 00:27:59.590 peter.bassett: I see.
240 00:28:01.550 ⇒ 00:28:02.080 peter.bassett: Cool.
241 00:28:03.910 ⇒ 00:28:05.390 Joshua Dent: Yeah.
242 00:28:06.840 ⇒ 00:28:08.160 Luke Scorziell: Josh, do you have a…
243 00:28:08.760 ⇒ 00:28:13.860 Joshua Dent: No, I was just saying, this is great, I like that. We’ve got all the courses there, they’re all indexed.
244 00:28:14.290 ⇒ 00:28:30.110 Joshua Dent: Yeah. Does it… can it provide any insights, or is it just reporting data? Like, if we had questions about, for instance, a new business campaign from, like, 2 years ago, could we ask it, like, what was our approach?
245 00:28:30.360 ⇒ 00:28:37.369 Joshua Dent: You know, how did we handle this similar type company to what we’re pitching now? Could we ask it questions like that?
246 00:28:37.950 ⇒ 00:28:47.500 Luke Scorziell: Yeah, especially in the knowledge base, and with the archive, like, we have… you know, I do stuff like that all the time, where it’s like, hey, how did we pitch this client in the past, or…
247 00:28:47.610 ⇒ 00:28:58.419 Luke Scorziell: What did we build out for this client? And we build case studies all the time using that. I did a brand strategy process for us a couple weeks ago that
248 00:28:58.600 ⇒ 00:29:02.610 Luke Scorziell: I just searched through all of our different,
249 00:29:03.600 ⇒ 00:29:12.010 Luke Scorziell: client transcripts and asked, like, what are some of the most common pain points that are described, and then it gave me an analysis, and I can kind of use that, so…
250 00:29:12.120 ⇒ 00:29:18.589 Luke Scorziell: Yeah, I guess we could get to the technical side, and probably the future, if you guys want to
251 00:29:18.760 ⇒ 00:29:24.869 Luke Scorziell: continue chatting about this, of, like, where that would fit in. I think we clearly have it in Teams.
252 00:29:25.060 ⇒ 00:29:30.579 Luke Scorziell: But, yeah, 100%, it’s, like, it’s… it’s like, chatty, all the benefits that you get with,
253 00:29:30.690 ⇒ 00:29:34.489 Luke Scorziell: A large language model, but the,
254 00:29:35.110 ⇒ 00:29:42.000 Luke Scorziell: The knowledge base is… or then you also have your knowledge base, so… Yeah.
255 00:29:43.350 ⇒ 00:29:46.420 Pranav Narahari: Yeah, to add to that too, like.
256 00:29:46.570 ⇒ 00:29:51.969 Pranav Narahari: We actually have more freedom, because when we’re using the API for creating a chat like this, we can…
257 00:29:51.990 ⇒ 00:30:11.950 Pranav Narahari: modulate, like, all these different parameters that ChatGPT just kind of packages for, like, the consumer. For your specific use case, we can figure out, okay, what temperature do you guys want? What, like, budget for thinking do you guys want in terms of tokens? And there’s a list of different parameters that we can, like, figure out there, a list of models that we can fit to your guys’ use case as well.
258 00:30:15.780 ⇒ 00:30:16.410 Joshua Dent: Right.
259 00:30:17.680 ⇒ 00:30:24.040 Luke Scorziell: I know we’re right up at 10.30. Do you guys have another… do you have a couple minutes, or what’s your schedule look like?
260 00:30:24.040 ⇒ 00:30:26.270 Joshua Dent: I, I do need to jump, I have a heart out.
261 00:30:26.270 ⇒ 00:30:28.449 peter.bassett: Yeah, I… I do as well.
262 00:30:28.740 ⇒ 00:30:33.430 Luke Scorziell: Okay. Well, what would a good next step look like? Is there another, like.
263 00:30:34.060 ⇒ 00:30:40.020 Luke Scorziell: someone else that you’d want to see this, or we could set up another meeting on Tuesday again?
264 00:30:40.020 ⇒ 00:30:47.909 Joshua Dent: You know, let me and Peter convene internally and get back to you, you know, potentially by the end of the week, and
265 00:30:48.290 ⇒ 00:30:49.330 Joshua Dent: We’ll let you know.
266 00:30:49.330 ⇒ 00:30:49.900 peter.bassett: Yeah.
267 00:30:50.070 ⇒ 00:30:50.560 Luke Scorziell: Okay.
268 00:30:50.560 ⇒ 00:30:51.579 peter.bassett: That sounds great.
269 00:30:52.710 ⇒ 00:30:57.530 Luke Scorziell: Cool. Yeah, well, if there’s any questions, feel free to shoot me an email, and then,
270 00:30:57.690 ⇒ 00:31:02.130 Luke Scorziell: Yeah, would love to, love to work with you guys, so, really excited.
271 00:31:02.510 ⇒ 00:31:14.610 Pranav Narahari: Quick thing too, Josh, I just looked it up for, the Teams meeting, like, if you could add the bot, and they have, this graph API that Microsoft exposes, and they can do that too. So, just thought I’d let you know.
272 00:31:14.610 ⇒ 00:31:15.919 Joshua Dent: Okay, thanks a lot, friend.
273 00:31:16.760 ⇒ 00:31:17.400 Pranav Narahari: Totally.
274 00:31:17.690 ⇒ 00:31:18.640 peter.bassett: Excellent.
275 00:31:19.050 ⇒ 00:31:20.040 Joshua Dent: I’ll catch you so much.
276 00:31:20.040 ⇒ 00:31:20.670 peter.bassett: period of time.
277 00:31:21.030 ⇒ 00:31:22.010 Pranav Narahari: Thank you.
278 00:31:22.720 ⇒ 00:31:23.650 Luke Scorziell: But…
279 00:31:26.710 ⇒ 00:31:27.960 Pranav Narahari: Hey, Luke.
280 00:31:27.960 ⇒ 00:31:28.739 Luke Scorziell: off. Let’s do it.
281 00:31:28.740 ⇒ 00:31:30.030 Pranav Narahari: Alright, after this.
282 00:31:30.370 ⇒ 00:31:31.460 Luke Scorziell: Oh, you do? Okay.
283 00:31:31.760 ⇒ 00:31:32.720 Pranav Narahari: We can sync up.
284 00:31:33.230 ⇒ 00:31:33.780 Luke Scorziell: Stink, yeah.