Meeting Title: AI Meeting Date: 2024-12-05 Meeting participants: Mariane Cequina, Luke Daque, Nicolas Sucari, Uttam Kumaran, Ryan Brosas, Miguel De Veyra, Casie Aviles


WEBVTT

1 00:00:02.570 00:00:04.329 Miguel de Veyra: Okay, yeah, there you go.

2 00:00:06.410 00:00:15.550 Miguel de Veyra: So yeah, so I think for this 1st session with them, I think, what I, what I want to do is basically just ask everyone you know on

3 00:00:15.680 00:00:17.759 Miguel de Veyra: what’s something they could.

4 00:00:18.235 00:00:31.060 Miguel de Veyra: They think that any agent could help. I could give you like some examples of what we use, what I use it for. Maybe that will, you know, bring up some bells here. You don’t necessarily have to come up with something right now.

5 00:00:31.270 00:00:35.659 Miguel de Veyra: It could be, you know, just be shared on the Brainforge team or the AI Channel.

6 00:00:37.038 00:00:43.370 Miguel de Veyra: But yeah, let me share my screen course relevance.

7 00:00:54.440 00:00:55.939 Miguel de Veyra: Yeah. So here you go.

8 00:01:00.410 00:01:02.040 Miguel de Veyra: Can everyone see my screen?

9 00:01:04.300 00:01:04.970 Nicolas Sucari: Yes.

10 00:01:04.970 00:01:09.800 Miguel de Veyra: Okay, yeah. So here’s like one of the

11 00:01:10.250 00:01:21.160 Miguel de Veyra: well, this is this meeting Summarizer. But this is one of the agents I recently built, because this was on my personal chat Gpt, so I just built it here. It’s basically like a coding agent for us.

12 00:01:22.010 00:01:28.379 Miguel de Veyra: because, for example, one of the things we were working on currently is with Vitaco. I’m handing it over to Casey.

13 00:01:28.510 00:01:37.780 Miguel de Veyra: So we created basically this agent. The problem is pretty short. You know, you’re a senior level developer with expertise in this areas.

14 00:01:38.270 00:01:43.309 Miguel de Veyra: And then basically, what I asked is to what I asked it to do is basically to Qa, it.

15 00:01:44.120 00:01:52.449 Miguel de Veyra: Wait. Let me just oh, yeah, basically here, go to review it, basically. And then, you know, came up with some stuff that we could improve.

16 00:01:53.800 00:02:11.770 Miguel de Veyra: should I separate to multiple files, basically just talking to it, and then the file structure? And then eventually I asked it to. You know, I’m handing it over to one of our developers. Could you please create for him to follow? This would have been otherwise. You know I have to create on my own. But then, as you can see, everything is already in here, and I’ve handed it over to Casey.

17 00:02:13.366 00:02:15.490 Miguel de Veyra: But yeah, this is one of the

18 00:02:15.690 00:02:17.419 Miguel de Veyra: things that I use it, for

19 00:02:17.730 00:02:21.899 Miguel de Veyra: on my personal chat I’m not sure if it’s logged in here, let me check.

20 00:02:23.070 00:02:24.359 Miguel de Veyra: Oh, yeah, it is.

21 00:02:24.400 00:02:27.879 Miguel de Veyra: Wait! I do also have wait. Let me see.

22 00:02:29.390 00:02:35.540 Miguel de Veyra: Yeah. So I have, like this voice agent, prompt maker and the improver and maker. So it’s basically.

23 00:02:35.580 00:02:42.740 Miguel de Veyra: I just put all the details of a website here, and then it automatically creates that. But again, I have to move this into relevance. So everyone can see

24 00:02:44.700 00:02:46.560 Uttam Kumaran: Can you zoom in a little bit, Miguel?

25 00:02:47.595 00:02:48.030 Miguel de Veyra: Sorry!

26 00:02:50.270 00:02:52.580 Uttam Kumaran: Or just make maybe make your window smaller. I don’t know.

27 00:02:54.340 00:02:55.489 Miguel de Veyra: Oh! Is this better?

28 00:02:56.420 00:02:57.449 Uttam Kumaran: Yeah, this is good.

29 00:02:58.090 00:02:59.999 Miguel de Veyra: So improver and maker.

30 00:03:01.490 00:03:06.660 Miguel de Veyra: So basically, that’s, you know, that’s what I use to like. If I want to create like an agent.

31 00:03:06.760 00:03:10.259 Miguel de Veyra: to have like a baseline of protocols and instructions.

32 00:03:12.650 00:03:17.340 Miguel de Veyra: But yeah, I think that’s pretty much it on this one.

33 00:03:20.510 00:03:24.240 Miguel de Veyra: Casey is you’re the guy from the AI side. Is there like any agent you’re using

34 00:03:24.500 00:03:25.860 Miguel de Veyra: or prompts you’re using.

35 00:03:26.970 00:03:31.170 Casie Aviles: Yeah. I mean, currently, right now, I I guess I can share a little bit about

36 00:03:31.676 00:03:35.040 Casie Aviles: the internal agents that we have at the company. So

37 00:03:35.700 00:03:38.359 Casie Aviles: yeah, some a few ones that I’ve set up

38 00:03:38.500 00:03:41.420 Casie Aviles: was for the lead researcher agent, like

39 00:03:42.180 00:03:45.790 Casie Aviles: we have one right there over there, as you can see in the screen. Yeah.

40 00:03:46.390 00:03:49.939 Casie Aviles: So basically, I guess the problem that

41 00:03:50.600 00:03:57.450 Casie Aviles: this agent is trying to solve is with when it comes to researching leads like, it’s

42 00:03:57.978 00:04:02.889 Casie Aviles: it’s a manual process, right? Like we have to. You know, do a bunch of Google searches and

43 00:04:03.230 00:04:10.369 Casie Aviles: some scrolling and some yeah, and basically, what this agent does is to streamline that process. So

44 00:04:10.690 00:04:14.140 Casie Aviles: we’ve equipped it with a bunch of tools to do that

45 00:04:14.830 00:04:17.740 Casie Aviles: like Google search and then read.

46 00:04:17.890 00:04:25.639 Casie Aviles: And then also another thing that we could also use is perplexity. Right? So we’ve also started using perplexity in some of our agents to do that.

47 00:04:27.170 00:04:34.617 Casie Aviles: So yeah, basically, the idea is to yeah, make the research process easier. Get the necessary data. And you know,

48 00:04:35.140 00:04:41.890 Casie Aviles: get that initial repetitive tasks out of the way and to focus on just the output.

49 00:04:44.880 00:04:49.179 Miguel de Veyra: So here’s 1 of the things that in case you worked on like the past few days.

50 00:04:49.280 00:04:52.810 Miguel de Veyra: So it’s for a client potential client we have.

51 00:04:54.237 00:05:01.530 Miguel de Veyra: It’s basically like a research agent like the big things here. I from what we can at least see from their documents.

52 00:05:01.710 00:05:08.660 Miguel de Veyra: is, where’s the investment, the stats, basically, how many employees they have, how many clients they have.

53 00:05:08.780 00:05:12.940 Miguel de Veyra: And then, you know, some Linkedin data. And then

54 00:05:15.560 00:05:18.210 Miguel de Veyra: where’s the way I can’t see it funding.

55 00:05:18.853 00:05:20.269 Miguel de Veyra: Here we go funding round.

56 00:05:21.010 00:05:26.950 Miguel de Veyra: So basically, it just researches. If they have, you know, funding or not from here, it doesn’t look like they have it. Do they do have.

57 00:05:27.640 00:05:30.739 Miguel de Veyra: This is like the previous company automi that

58 00:05:30.810 00:05:43.310 Miguel de Veyra: me and Casey worked with. They do have like a funding, because, you know, when venture capitalists look for people are for companies to invest and they want to know if there’s fund basically, this is already involved.

59 00:05:43.580 00:05:46.889 Miguel de Veyra: So you know, it’s here. And yeah.

60 00:05:47.020 00:05:48.830 Miguel de Veyra: that’s what we get from it.

61 00:05:50.180 00:05:52.830 Miguel de Veyra: I know, Ryan, you’re working on some stuff

62 00:05:53.000 00:05:57.630 Miguel de Veyra: from the AI side of things. You’re using some agents. I’m not sure if it’s just

63 00:05:58.190 00:06:01.670 Miguel de Veyra: promptings on Gpt or Quad.

64 00:06:01.670 00:06:12.849 Luke Daque: Yeah, mostly, it’s just doing some prompts especially currently for the synthetic data creation. We did this with utham last, like, I think last week or the other week, I believe.

65 00:06:12.880 00:06:19.570 Luke Daque: where we did try to. Yeah, it’s basically just prompts to have a AI

66 00:06:19.600 00:06:21.700 Luke Daque: like generate a code for us,

67 00:06:23.480 00:06:33.829 Luke Daque: to generate synthetic data. Basically, we, we tried to do it like direct, like, making AI generate the synthetic data for us. But it

68 00:06:34.030 00:06:38.110 Luke Daque: didn’t do that. Well, so we

69 00:06:38.460 00:06:44.620 Luke Daque: we changed the process basically like to ask AI to generate a code that would

70 00:06:44.920 00:06:51.110 Luke Daque: essentially generate synthetic data as well. So yeah, that’s like one of the ways I’m using AI

71 00:06:51.650 00:06:56.059 Luke Daque: other ways I’ve been always using ever since we started

72 00:06:56.150 00:07:00.510 Luke Daque: in in brain. Forge was like having AI also like, Do

73 00:07:02.860 00:07:05.249 Luke Daque: like help improve code. Or like.

74 00:07:05.370 00:07:08.420 Luke Daque: when I’m creating a data model, for example, for

75 00:07:08.660 00:07:14.700 Luke Daque: a specific data model, I, I ask like AI to help us out in

76 00:07:14.710 00:07:18.500 Luke Daque: like making the code more effective, efficient. And stuff like that.

77 00:07:19.690 00:07:20.380 Luke Daque: Yeah,

78 00:07:21.360 00:07:29.978 Luke Daque: And even when I was using copilot in in Vs code before. Well, currently, I’m using cursor for us like my code.

79 00:07:30.680 00:07:31.979 Luke Daque: what do you call that?

80 00:07:32.570 00:07:37.309 Luke Daque: It’s it’s like it. Cursor is like Bs code, but it has AI with it.

81 00:07:37.310 00:07:37.720 Miguel de Veyra: Yep.

82 00:07:37.720 00:07:44.360 Luke Daque: So like, it can like basically complete the code even without me typing in it.

83 00:07:44.530 00:07:46.804 Luke Daque: So it makes a lot.

84 00:07:47.490 00:07:50.160 Luke Daque: it makes the process a lot faster. Basically

85 00:07:50.642 00:07:56.380 Luke Daque: in like creating data models, or even like dashboards, real dashboards and stuff like that. So yeah, it’s pretty

86 00:07:56.900 00:07:58.060 Luke Daque: pretty great.

87 00:08:00.600 00:08:04.645 Luke Daque: What I’ve also done. I tried doing in the past.

88 00:08:06.370 00:08:09.330 Luke Daque: was like, create. Try to have

89 00:08:09.960 00:08:15.940 Luke Daque: AI do code review in our Github repositories, although I haven’t like

90 00:08:16.050 00:08:25.969 Luke Daque: continued working on that one. But I did try that out before, and I did incorporate it in github actions. I was using open AI for that one.

91 00:08:26.030 00:08:34.959 Luke Daque: And like that’s I believe that’s why I asked you, Miguel, before, like, maybe we can do use notion for that instead of like.

92 00:08:34.960 00:08:35.570 Miguel de Veyra: But.

93 00:08:35.980 00:08:37.409 Luke Daque: Hard, coding it in a.

94 00:08:37.419 00:08:38.279 Miguel de Veyra: That they know.

95 00:08:38.280 00:08:40.491 Luke Daque: I haven’t. Unfortunately, I haven’t like

96 00:08:41.049 00:08:45.499 Luke Daque: Yeah, continued working on that one, but that would have been a a cool one to do.

97 00:08:45.760 00:08:48.919 Miguel de Veyra: Yeah, I remember we talked about it like 2 or 3 weeks ago. Now.

98 00:08:48.920 00:08:50.969 Luke Daque: Yeah, something like that. Yeah.

99 00:08:52.100 00:08:55.930 Luke Daque: So yeah, it’s been, that’s essentially like how I’ve been using AI.

100 00:08:55.930 00:09:00.450 Miguel de Veyra: The synthetic data you’ve mentioned. Is there? Is it more like dummy data? Sorry, not familiar.

101 00:09:00.450 00:09:04.470 Luke Daque: Yes, yes, it’s it’s dummy data, basically. So like,

102 00:09:05.280 00:09:09.989 Luke Daque: wait, I I can probably share my screen. Let me just pull pull that one up

103 00:09:10.590 00:09:14.570 Luke Daque: because it’s in a different. It’s in our brain forge repository.

104 00:09:21.850 00:09:23.539 Luke Daque: Just give me a second.

105 00:09:35.200 00:09:37.349 Luke Daque: So yeah, let me share my screen.

106 00:09:43.580 00:09:44.750 Luke Daque: Can you see my screen.

107 00:09:44.750 00:09:45.080 Miguel de Veyra: Yep.

108 00:09:46.000 00:09:47.690 Luke Daque: So this is cursor.

109 00:09:51.320 00:09:54.729 Luke Daque: Wait! Let me see if I can find the prompt that I did.

110 00:10:00.910 00:10:03.719 Miguel de Veyra: AI’s main job is basically to create data. Now.

111 00:10:04.270 00:10:08.430 Luke Daque: Yeah. So I’ll have. It might take a while for me to

112 00:10:09.200 00:10:20.989 Luke Daque: look for the the prompt but essentially what it did was, give us a a python script.

113 00:10:22.238 00:10:31.189 Luke Daque: Is it? Yeah. So we asked AI to generate a python script that would give us a synthetic data. I I can’t zoom in here, but this is like.

114 00:10:31.190 00:10:31.800 Miguel de Veyra: Yeah, yeah.

115 00:10:31.800 00:10:34.020 Luke Daque: Dummy data, basically. And it has

116 00:10:34.618 00:10:37.579 Luke Daque: all the fields that we need for this specific

117 00:10:38.439 00:10:42.629 Luke Daque: data. Right? So we, we asked it to generate

118 00:10:42.910 00:10:46.999 Luke Daque: data. That would be that would make sense for a manufacturing company.

119 00:10:47.330 00:10:50.239 Luke Daque: So we can create a real dashboard out of it.

120 00:10:50.360 00:10:56.020 Luke Daque: And basically, yeah, so this is just a hundred lines of data, I believe.

121 00:10:57.570 00:11:00.720 Luke Daque: For now, just we, we can, we can

122 00:11:01.090 00:11:04.670 Luke Daque: edit this as as we like. Like.

123 00:11:05.080 00:11:12.270 Luke Daque: So yeah, you can see that. Yeah, it has like randomness, it assigns random values to like each of the fields. And

124 00:11:13.470 00:11:17.850 Luke Daque: and yeah, it. It did that. So we didn’t have to like, think about

125 00:11:18.050 00:11:21.220 Luke Daque: how we want the randomness to be.

126 00:11:21.360 00:11:25.820 Luke Daque: And yeah, I can show you real quick, like how the dashboard looks like.

127 00:11:27.010 00:11:32.509 Nicolas Sucari: And the goal here is to create different data sets regarding different markets so that we can share

128 00:11:32.520 00:11:55.820 Nicolas Sucari: with possible leads when we are trying to get a new company, a new client and we are we we haven’t worked with a company in, or a client in that kind of area. The idea is to have, like these example data sets that we are creating so that we can share how the dashboards will look when they share with us like their data. And we start working.

129 00:11:56.740 00:12:03.520 Luke Daque: Yeah, so this is the example of like, when we already made this. So basically this data.

130 00:12:03.550 00:12:06.739 Luke Daque: the output of this script is a Csv file.

131 00:12:07.310 00:12:10.920 Luke Daque: So yeah, it’s something like this. It’s just a Csv file.

132 00:12:11.110 00:12:14.309 Luke Daque: And then we loaded this Csv file into

133 00:12:14.700 00:12:20.862 Luke Daque: a dashboard. And it looks something like this. It’s pretty much like very random. So

134 00:12:22.070 00:12:33.221 Luke Daque: it’s not very it. It’s not like very linear or something like that. So it makes sense like total calls made, for example, would go up and down per day, depending on the day

135 00:12:33.620 00:12:38.219 Luke Daque: yeah, stuff like that. So yeah, it’s pretty sick. Actually.

136 00:12:38.220 00:12:41.589 Miguel de Veyra: How how many, how many? Sorry? Yeah, it’s same.

137 00:12:41.620 00:12:47.319 Miguel de Veyra: How many data do you guys usually like, basically, for example, how many data are you guys looking to make

138 00:12:48.070 00:12:48.840 Miguel de Veyra: a thousand.

139 00:12:48.840 00:12:52.379 Nicolas Sucari: Like, how many data sets different data sets, or how many rows of.

140 00:12:52.380 00:12:53.740 Miguel de Veyra: Records basically yeah.

141 00:12:53.740 00:12:54.770 Luke Daque: Records.

142 00:12:54.860 00:12:58.510 Luke Daque: Yeah, currently, this is just a hundred records. We can up

143 00:12:59.122 00:13:00.579 Luke Daque: like, make this like a thousand.

144 00:13:00.580 00:13:04.289 Uttam Kumaran: It depends on like the it depends on the data set. I think, like.

145 00:13:04.290 00:13:04.960 Nicolas Sucari: Sad to hear.

146 00:13:04.960 00:13:12.970 Uttam Kumaran: Sales data. Then we probably want to do like, we’re trying to create data that looks like the companies we’re going after. So

147 00:13:13.340 00:13:21.589 Uttam Kumaran: you know hundreds of sales, thousands of sales per day. It kind of depends. The data set. The biggest thing is that it looks real. Right. So

148 00:13:21.710 00:13:24.230 Uttam Kumaran: if it’s a has seasonality

149 00:13:24.410 00:13:38.089 Uttam Kumaran: data goes down on weekends like there’s some realistic nature to it. And then, ideally, again, for me, the biggest thing was, we just have the the data generate new data every month.

150 00:13:38.230 00:13:48.340 Uttam Kumaran: right or or or basically what we do is actually, this is, I think we talked about this right. We just generate future data. And then that way, when the day hits, there’s always fresh data.

151 00:13:48.380 00:13:50.470 Uttam Kumaran: But the biggest thing here is

152 00:13:50.970 00:14:02.379 Uttam Kumaran: companies don’t. We can’t. When we do a demo of like what we can do and data, we can’t show other people’s data. So it’s kind of hard to give a sense of the types of things we can accomplish.

153 00:14:02.720 00:14:14.569 Uttam Kumaran: This is like a great way for us to show, like a fake company of like, hey? Your company’s data. Imagine it here and imagine you can do these things with it. So that’s basically the pitch.

154 00:14:15.100 00:14:15.630 Luke Daque: Yeah.

155 00:14:17.360 00:14:17.950 Miguel de Veyra: Okay.

156 00:14:19.130 00:14:27.090 Miguel de Veyra: okay, yeah, I can think of something. Let’s hop on a different call, maybe not this one. So we don’t want to take everyone’s time, but I think I can think of something.

157 00:14:28.140 00:14:29.600 Luke Daque: I’ll share it to you.

158 00:14:30.450 00:14:33.929 Miguel de Veyra: Okay, yeah. So a synthetic data agent. Let’s put that.

159 00:14:34.570 00:14:35.950 Miguel de Veyra: Okay. Yeah.

160 00:14:38.300 00:14:39.220 Miguel de Veyra: Wait. Sorry.

161 00:14:39.350 00:14:40.130 Miguel de Veyra: Ryan.

162 00:14:40.820 00:14:44.880 Miguel de Veyra: Okay, so the other thanks for sharing Luke.

163 00:14:45.180 00:14:45.599 Luke Daque: Thanks.

164 00:14:55.160 00:14:56.010 Miguel de Veyra: All right.

165 00:14:56.940 00:14:57.590 Ryan Brosas: Sorry.

166 00:14:58.250 00:15:00.019 Miguel de Veyra: Is there like anything you’re, you know.

167 00:15:00.190 00:15:03.410 Miguel de Veyra: using an AI agent for? So I know you’re in the content side right.

168 00:15:04.195 00:15:04.700 Ryan Brosas: Yeah.

169 00:15:05.930 00:15:17.670 Ryan Brosas: Yeah. For analyzing. The structure of template of specific individual, like Justin Welsh or like

170 00:15:18.558 00:15:24.779 Ryan Brosas: alex Hermosi. I used like a a like a a prompt

171 00:15:24.830 00:15:30.830 Ryan Brosas: that could, you know, destruct the structure. And I can follow that structure so we can make

172 00:15:31.090 00:15:38.139 Ryan Brosas: like follow their their how the content works and how we can, you know, use an app.

173 00:15:38.290 00:15:40.349 Ryan Brosas: apply it to our content as well.

174 00:15:41.073 00:15:47.579 Ryan Brosas: I use, and also, I think, for the research. I use like

175 00:15:48.090 00:15:51.539 Ryan Brosas: perplexity for our blog posts.

176 00:15:52.970 00:15:59.639 Ryan Brosas: to have, like a, you know, a detailed and have, like an analytic side of stuff in our blog posts.

177 00:15:59.880 00:16:08.970 Ryan Brosas: I think that’s all in my in my side, on on AI stuff. But you know I’m looking forward to like build you know. A content

178 00:16:10.113 00:16:15.839 Ryan Brosas: second brain content. I’m still building it. It’s quite a while right so.

179 00:16:15.840 00:16:16.370 Miguel de Veyra: Once.

180 00:16:17.127 00:16:23.210 Ryan Brosas: Yeah, I’m in relevance. But I’m looking forward to you know the new stuff in in in Claude.

181 00:16:24.069 00:16:33.309 Ryan Brosas: I’m looking like building like a a content that could use by Nicholas and Tom. So we can, you know, push contents on Linkedin.

182 00:16:33.712 00:16:44.940 Ryan Brosas: I will like feed it with information on how we you know how, how other specific individuals that is, you know, have having, like a good structure on content.

183 00:16:45.542 00:16:51.989 Ryan Brosas: like creating it more data on that stuff. And also, you know, as as also the

184 00:16:54.960 00:17:00.930 Ryan Brosas: yeah, I think that’s that’s 1 of what I want to do on on AI also.

185 00:17:01.373 00:17:04.890 Ryan Brosas: I think that’s all for for my side in the content side.

186 00:17:05.800 00:17:11.949 Miguel de Veyra: Okay, yeah, thanks for sharing. Let’s hop on a call. I think. Uttan, what could happen is, I hop on a call individually with them. Now.

187 00:17:11.950 00:17:17.979 Uttam Kumaran: Yeah, I actually told Ryan today to book a call next week. But I think each of these areas kind of become like.

188 00:17:18.520 00:17:31.480 Uttam Kumaran: how do we? OP, how do we use AI in different departments? Right? Like we have operations. We have engineering. We even have the AI department. We have like a bunch of stuff. So really, it’s just like.

189 00:17:32.090 00:17:37.629 Uttam Kumaran: partly, it’s like brain. Forge becomes a client for itself, right? So

190 00:17:37.950 00:17:44.359 Uttam Kumaran: that’s the thing that’s really lovely. I think these meetings help for everybody to kind of still share. But then, yeah, I think

191 00:17:44.440 00:17:50.100 Uttam Kumaran: I think you should just go chat with everybody, learn everybody’s use cases, because probably 10 min with you.

192 00:17:50.220 00:17:53.609 Uttam Kumaran: what take people on a completely different path?

193 00:17:54.057 00:18:10.140 Uttam Kumaran: And that’s really what I think a lot of people need. And also some people like for for me I use this stuff every day for like 2 years, but for some people this is new every day. So giving making the barrier to entry really low, giving access to all of our tools.

194 00:18:10.571 00:18:17.078 Ryan Brosas: We may not know what the problems are, but we definitely have the tools to solve. Probably every problem right in front of us.

195 00:18:17.460 00:18:23.829 Uttam Kumaran: And then, yeah, like, it’s just how do we? How do we get and make sure everybody’s using AI and their daily workflows?

196 00:18:23.980 00:18:32.794 Uttam Kumaran: That’s it cause even for me. It’s like I like. I wouldn’t have been able to do this business without the AI stuff like it’s way. It’s just like way too hard.

197 00:18:33.070 00:18:35.140 Miguel de Veyra: I’m not writing all those prompts.

198 00:18:35.340 00:18:38.780 Uttam Kumaran: Yeah. So I think it’s really great.

199 00:18:41.300 00:18:43.940 Nicolas Sucari: Yeah, I have some ideas, Miguel. So.

200 00:18:44.000 00:18:48.660 Nicolas Sucari: and I, I know how the AI can help me on my daily basis

201 00:18:49.064 00:19:09.620 Nicolas Sucari: to send updates to the clients knowing what was done like during the week. So maybe we can work towards something like that like reading all of the the latest conversation in slack for the past week, and creating like a status report, or something like that, or reading like a status of tasks in a notion board.

202 00:19:09.620 00:19:20.969 Nicolas Sucari: and seeing, like the the changes from one week to the other, one and then create like a slack message and send directly to the client something like that should would be like great.

203 00:19:21.550 00:19:28.370 Miguel de Veyra: Yeah, I was actually thinking of something similar earlier, because I was updating Eddie on stuff. So I think it’s like

204 00:19:28.550 00:19:40.820 Miguel de Veyra: I was thinking, maybe it would be like a good idea to just put the updates from either me or Casey into the slack channel, and then maybe just send it to a relevance agent, you know, hey, format this in a way that it’s, you know.

205 00:19:41.410 00:19:42.680 Miguel de Veyra: for the client.

206 00:19:42.890 00:19:49.140 Miguel de Veyra: and then we could review it first, st of course, before we send it to the client. Cause I know we

207 00:19:49.750 00:19:52.720 Miguel de Veyra: there’s some clients, especially on the data side, that are

208 00:19:53.390 00:19:59.529 Miguel de Veyra: high communication. Is that the term? Yeah. Yeah. So I was like, yeah, that’s probably one of the agents we want to build.

209 00:20:00.860 00:20:03.149 Nicolas Sucari: Yeah, I mean, I I like.

210 00:20:03.480 00:20:14.850 Nicolas Sucari: I I don’t mind on reviewing what the the agent can can create, but it’s like the previous step. What I’m trying to to build is like, how do we get all of

211 00:20:14.880 00:20:28.898 Nicolas Sucari: what what happens during the entire week into a status like? For now I I am kind of reviewing all of these like messages. I’m going to the notion boards. I’m seeing the pull requests. I’m seeing everything that was done, and then writing myself

212 00:20:29.489 00:20:50.439 Nicolas Sucari: like like a message, and sending it that to chat, gpt to format it, and then copy it and paste it to the client. But if there is something that can automatically go check all of these slack messages during the week, and the the tasks that were that were worked on and then create automatically the slack message. Then we just like check the message and see if that’s accurate or not.

213 00:20:51.030 00:20:59.659 Uttam Kumaran: Yeah, I think, Miguel. Also, it’s like it’s honestly kind of a mix of the stuff we’re doing on the content side. Because Ryan now has content brain that knows

214 00:20:59.730 00:21:04.090 Uttam Kumaran: how I wanted our verbiage to be like

215 00:21:04.410 00:21:28.570 Uttam Kumaran: when he generates stuff. Now, compared to 3 months ago, it actually sounds like pretty much like what I how I would write it, or how I would want it to write, and that took a journey right? So there’s probably prompts involved. Very similarly, I think for Nico, it’s like he has a structure by which he sends those updates. The inputs, though, need to come from slack email. And maybe even Github all get basically

216 00:21:28.700 00:21:34.952 Uttam Kumaran: process. And then client, it says, like, what do we do for this client this week? And then there’s an Update

217 00:21:35.320 00:21:38.329 Miguel de Veyra: Yeah, like that sort of thing

218 00:21:38.820 00:21:45.069 Miguel de Veyra: Nico, do you think it would be like? Let’s start, let’s start with something a bit simpler. Just so we could get our, you know.

219 00:21:45.130 00:21:53.029 Miguel de Veyra: put off the ground. But do you guys provide, do? Do we have like daily updates, for example, for a client like, hey, here’s what we did today.

220 00:21:53.100 00:22:01.539 Miguel de Veyra: Right? Cause I think what we could do if it’s like just daily, right may maybe someone working on data side. Here’s like the daily updates. Put it all. And then.

221 00:22:01.800 00:22:10.690 Miguel de Veyra: you know, we can just collect that at the end of the week. Send it all to an agent, process it. And I think that streamlines, basically everything you’re

222 00:22:11.210 00:22:12.109 Miguel de Veyra: working on.

223 00:22:12.110 00:22:34.710 Nicolas Sucari: Yeah, I mean, we don’t have like anything standardized like day in a daily basis. But we can. We can work on something. Yeah, totally. I mean, if if that’s the way, in order to start trying one of these agents and see how how that works. Yeah, let’s do it. I mean we can. We can. I think we can start using standardly again, maybe maybe just for me, so that I can get the habit of.

224 00:22:34.710 00:22:35.380 Uttam Kumaran: Oh, okay.

225 00:22:35.380 00:22:41.869 Nicolas Sucari: Everyday something, you know, and for each client. So I can write something for each client and see if that.

226 00:22:42.292 00:22:54.530 Uttam Kumaran: What’s even better is instead of stand up late, like, I’m sure these guys can create says, what what happened with these clients? You respond all that gets sent

227 00:22:54.740 00:23:08.729 Uttam Kumaran: to Csv. Or something. And then at the end of the week it’s all come. So that’s the biggest thing also is having it work with you like you can’t have it be a completely different workflow. Otherwise you’re never gonna adopt it right? So you’re almost building it for yourself.

228 00:23:08.930 00:23:23.570 Uttam Kumaran: So build something like that’s small. But also it’s like accomplishes the goal. And then, yeah, like, I mean again, I think we’re gonna try to gun for at least every 2 days. Updates, I think we’ll probably start there like next week.

229 00:23:23.800 00:23:29.120 Uttam Kumaran: So we I wanna try to find some way. For like, for example.

230 00:23:29.160 00:23:33.139 Uttam Kumaran: if you’re working on 4 clients, it should automatically say, Hey.

231 00:23:33.170 00:23:36.080 Uttam Kumaran: based on slack. Here’s probably what happened yesterday.

232 00:23:36.270 00:23:37.650 Nicolas Sucari: Exactly. Yeah.

233 00:23:37.650 00:23:41.979 Uttam Kumaran: That’s right, and maybe at least it is 50%, or at least it like

234 00:23:42.090 00:23:45.840 Uttam Kumaran: gives you the reminder to do the rest right. Stuff like that.

235 00:23:45.840 00:23:46.490 Miguel de Veyra: Yeah.

236 00:23:46.490 00:23:49.989 Nicolas Sucari: Yeah, yeah, yeah, that would be awesome.

237 00:23:50.440 00:23:56.100 Nicolas Sucari: But yeah, we can start working on something like that, Miguel. We can get in touch later and and see how we can do

238 00:23:56.230 00:23:57.940 Nicolas Sucari: Kickoff kind of did things.

239 00:24:00.320 00:24:05.070 Miguel de Veyra: Okay. Yeah. Okay. So last, Maria, Hello.

240 00:24:05.790 00:24:06.650 Mariane Cequina: Hello. Yeah.

241 00:24:07.780 00:24:10.700 Miguel de Veyra: So I think, yeah, yeah, is there like

242 00:24:10.740 00:24:15.850 Miguel de Veyra: any specific thing in operations that because I know you’re working with a lot of documentation.

243 00:24:16.650 00:24:20.840 Miguel de Veyra: right? So is there like a specific, you know, scenario, that

244 00:24:21.000 00:24:25.820 Miguel de Veyra: or daily thing, you can do that. Basically, you could, you could use some help using an agent.

245 00:24:27.180 00:24:35.509 Mariane Cequina: In Asian or like in in. How can I integrate like the task in notion? I can, because Nico last said, that you know you can just

246 00:24:35.860 00:24:47.869 Mariane Cequina: notify. I can actually, I can automate the if there are like assigned tasks in a specific person in notion, it can be sent in the slack as notification. Is that something that you’d want me to do.

247 00:24:49.130 00:24:49.660 Nicolas Sucari: Yeah.

248 00:24:49.660 00:24:55.220 Nicolas Sucari: I think we can do that. But I don’t think we need an a agent to do that. Maybe maybe there.

249 00:24:55.700 00:24:58.680 Nicolas Sucari: between notion and slack, and we can directly do that.

250 00:24:58.790 00:25:02.889 Nicolas Sucari: But yeah, I mean, we can try it. And we can. Yeah, look into that.

251 00:25:03.455 00:25:05.634 Nicolas Sucari: The the only thing is

252 00:25:06.630 00:25:17.370 Nicolas Sucari: As for now, as we are using like a different client page for for each client, it’s a different databases for each kind of roadmap

253 00:25:17.380 00:25:36.060 Nicolas Sucari: and and backlog of tasks. Right? So if, for example, here Luke is working with 2 different clients, we should like integrate 2 different databases into that note into that slack message so that he gets hit by that kind of agent, and remind him about all of the tasks from the different projects, right.

254 00:25:36.070 00:25:39.879 Nicolas Sucari: because we don’t have, like all in the same in the same database, in notion.

255 00:25:40.470 00:25:41.160 Mariane Cequina: Okay.

256 00:25:42.010 00:25:44.500 Nicolas Sucari: Okay. But yeah, I mean, need something that we can try.

257 00:25:44.930 00:26:01.389 Mariane Cequina: Yeah, maybe I’ll just present, first, st because I’m trying to strategize a lot, especially in the leads, because I talked with Utam about, you know, as as you mentioned before, like, there are a lot of clients. And you you guys also wanted to have, like a different views from external, even the internal. So I’m trying

258 00:26:01.730 00:26:07.510 Mariane Cequina: this person, then create like a workflow and present it to you guys, I think that’s ideology.

259 00:26:08.120 00:26:19.220 Nicolas Sucari: I I think that’s better. I mean, that’s why I I was not like pushing into doing that automations on what we have right now, I was just waiting until you can get your hands on

260 00:26:19.838 00:26:36.709 Nicolas Sucari: working on on these new client pages and and task management workflow. Once that is ready. I think we can think of what is necessary in order to automate and to start like sending that those kind of messages as reminders or as tasks reminder for everyone.

261 00:26:37.380 00:26:49.359 Mariane Cequina: Yeah, definitely cause cause. Honestly, I also wanted to understand more about the process as well. So that’s why I have to make a scratch first, st and then before we like, automate us and improve.

262 00:26:49.510 00:26:50.310 Mariane Cequina: Yeah.

263 00:26:50.310 00:26:52.640 Nicolas Sucari: Okay, perfect. Yeah.

264 00:26:52.940 00:26:53.590 Mariane Cequina: Okay.

265 00:26:54.350 00:27:01.750 Miguel de Veyra: Oh, I’m fine, Maureen. Sorry I’m not really specifically like super familiar. Do you also like, for example, Utah.

266 00:27:01.820 00:27:06.639 Miguel de Veyra: if there’s like a lead that comes in is Marian, the one working on documenting everything.

267 00:27:06.990 00:27:07.470 Mariane Cequina: No, no.

268 00:27:08.150 00:27:08.969 Miguel de Veyra: Oh, okay.

269 00:27:09.550 00:27:16.000 Uttam Kumaran: Well, we, I think we’re gonna we’re gonna basically right now, we’ve just basically finished the content departments

270 00:27:16.170 00:27:22.750 Uttam Kumaran: notion. We’re gonna we’re moving on one by one by one. So ideally, what will happen

271 00:27:22.780 00:27:48.900 Uttam Kumaran: like in my ideal world? Leads are gonna come from any which way, like I get an email, someone hits me up or something. They come to the site. All of it will get funneled, basically into slack and then also into notion, somehow. Right? So. But we’re figuring out the process right now. It’s mainly like, can we just figure out the manual thing that it gets tracked in notion, because right now, notion is gonna be basically, the company operates on notion slack and zoom

272 00:27:48.900 00:28:00.370 Uttam Kumaran: right? So we’re kind of restricting any other place where information can get dispersed. The other thing is all of that needs to end up back into an AI agent somehow. Right? So if you were to ask

273 00:28:00.400 00:28:10.990 Uttam Kumaran: the brain Forge AI agent. What leads came up this week that are new? It should answer that question right? And that’s the stuff that we are working on with Casey just like throwing all of notion into stuff

274 00:28:11.160 00:28:19.619 Uttam Kumaran: like that’s the stuff longer term we’ll figure out. But as long as it’s in notion categorized somewhere, that’s the ideal thing.

275 00:28:21.260 00:28:33.100 Uttam Kumaran: How it gets into notion, notion, added web hooks and stuff, too. So I think there’ll be some better ways of getting it into notion. But right now, it’s maybe one or 2 per week.

276 00:28:33.240 00:28:40.580 Uttam Kumaran: But again, like as we grow, these things get harder and harder. So yeah.

277 00:28:42.310 00:28:49.439 Nicolas Sucari: Yeah, the issue with notion is, if you’re not like on top of it and trying to update everything, it’s kind of

278 00:28:49.470 00:29:09.060 Nicolas Sucari: as soon as you finish creating a documentation database. And you don’t look at it anymore, like it becomes old. So that’s like our challenge is to be actively updating everything and keeping it, keeping it up to date. Until we figure out like, what’s the best way of doing this without doing it manually.

279 00:29:11.580 00:29:12.320 Nicolas Sucari: Yep.

280 00:29:16.250 00:29:29.889 Miguel de Veyra: Okay, yeah. Cause I was thinking earlier, if it helped me in documenting basically that stuff for the handover, if we’re doing it for a bunch of clients, I think you know long term it could really help out.

281 00:29:31.370 00:29:53.789 Nicolas Sucari: Yeah, I mean, what Marianne is doing is creating, like the the process on how we are gonna structure, all of the different areas in notion. But then, like creating the documentation for a client, that’s something we can work on, because that’s that that is gonna be needed for every client in at some point where we should like start

282 00:29:54.560 00:30:03.469 Nicolas Sucari: sending them the documentation in in a certain way. And that’s where you guys can step in and try to help us automate. How is that created right.

283 00:30:04.750 00:30:05.840 Miguel de Veyra: Yep, definitely.

284 00:30:07.940 00:30:13.009 Miguel de Veyra: Okay. Yeah, I’m I mean, we’re a bit over like 3 min, but I think that’s

285 00:30:13.740 00:30:18.480 Miguel de Veyra: pretty much it forever. For everyone, unless.

286 00:30:18.480 00:30:20.869 Nicolas Sucari: Guys. It’s a great, great 1st meeting of.

287 00:30:20.870 00:30:21.709 Miguel de Veyra: Yeah, I know.

288 00:30:23.160 00:30:27.729 Uttam Kumaran: So what’s next? What’s next, Miguel like? What’s the plan? After this.

289 00:30:27.730 00:30:32.400 Miguel de Veyra: I’ll hop on a call, probably next week or tomorrow

290 00:30:32.450 00:30:38.840 Miguel de Veyra: I’ll start hopping because I took down notes, so I’ll start hopping, hopping into calls with everyone.

291 00:30:39.030 00:30:43.100 Miguel de Veyra: And then basically just plan something out with them individually.

292 00:30:43.100 00:30:44.010 Nicolas Sucari: Excellent.

293 00:30:44.010 00:30:44.630 Mariane Cequina: Okay.

294 00:30:44.630 00:30:47.639 Miguel de Veyra: And then maybe I’ll create like a new channel in slack

295 00:30:48.100 00:30:51.890 Miguel de Veyra: for that specific case, because I don’t wanna fill up the Aid chat right.

296 00:30:51.890 00:30:54.500 Uttam Kumaran: Yeah, we have. There’s there’s another AI channel.

297 00:30:57.340 00:30:59.370 Uttam Kumaran: There’s a lot of channels, though. I’m gonna.

298 00:30:59.370 00:31:05.219 Nicolas Sucari: Yeah, yeah, there is one that’s automations. AI, I think that’s the one that you don’t wanna like.

299 00:31:05.890 00:31:09.920 Uttam Kumaran: There is automations. AI, that’s the one. Yeah, let’s just use that one.

300 00:31:10.930 00:31:13.950 Miguel de Veyra: Oh, should we do it in automations? AI. Then.

301 00:31:15.190 00:31:15.750 Nicolas Sucari: Yeah.

302 00:31:16.160 00:31:16.853 Miguel de Veyra: Okay. Okay.

303 00:31:17.200 00:31:23.980 Uttam Kumaran: Because I actually want everybody to see it. Like, frankly, I’m also thinking, maybe we just at some point, May, I’m thinking we even

304 00:31:24.560 00:31:28.409 Uttam Kumaran: one of the ideas I have is like to just push everything to Brainforge team.

305 00:31:28.450 00:31:29.990 Mariane Cequina: But I do think that.

306 00:31:30.210 00:31:38.830 Uttam Kumaran: Right now, I think we have a good enough of channels. What I’ll probably do is I’m gonna look at the data actually, and just figure out which channels aren’t being used. And then.

307 00:31:38.870 00:31:40.910 Uttam Kumaran: yeah, probably deleting them. But.

308 00:31:41.180 00:31:41.710 Miguel de Veyra: You’re not.

309 00:31:41.710 00:31:42.330 Nicolas Sucari: Yeah.

310 00:31:42.890 00:31:43.736 Miguel de Veyra: Okay, yeah.

311 00:31:44.380 00:31:45.820 Miguel de Veyra: I think we had a good session.

312 00:31:46.650 00:31:47.180 Uttam Kumaran: Okay.

313 00:31:47.650 00:31:48.670 Nicolas Sucari: Thanks. Miguel.

314 00:31:48.670 00:31:52.420 Miguel de Veyra: Okay, thank you. Okay, thank you. Guys, thanks.

315 00:31:52.420 00:31:54.159 Mariane Cequina: Thank you. Thank you.

316 00:31:54.160 00:31:54.730 Mariane Cequina: Day.

317 00:31:54.730 00:31:55.330 Luke Daque: Bye, bye.