Meeting Title: N8N Demo Dry Run Date: 2025-11-04 Meeting participants: Uttam Kumaran, Casie Aviles, Craig Foldes


WEBVTT

1 00:00:09.470 00:00:12.429 Uttam Kumaran: Hey, Casey, sorry, meeting ran over.

2 00:00:14.140 00:00:16.489 Casie Aviles: Oh, hey. Yeah, no worries.

3 00:02:32.310 00:02:33.900 Uttam Kumaran: I’m texting Craig, let’s see.

4 00:02:34.700 00:02:35.300 Casie Aviles: Okay.

5 00:02:39.910 00:02:43.509 Uttam Kumaran: I think you… you sent the NADN link in,

6 00:02:47.050 00:02:48.360 Uttam Kumaran: In Slack, right?

7 00:02:49.100 00:02:55.060 Casie Aviles: It’s a… it’s the… it’s Loom… it’s for Loom, it’s for the Loom video, but I do have, like, the…

8 00:02:55.440 00:02:57.570 Casie Aviles: And I don’t think if you also need that.

9 00:02:59.290 00:03:00.780 Uttam Kumaran: Yeah, that would be amazing.

10 00:03:02.960 00:03:08.839 Casie Aviles: Okay. Right now, I just did, like, a text entry for now,

11 00:03:09.760 00:03:14.179 Casie Aviles: Like, I wasn’t sure, like, if… It’s worth building up, like.

12 00:03:14.890 00:03:17.009 Casie Aviles: Something like what we did for Zoom.

13 00:03:29.660 00:03:32.050 Casie Aviles: Okay, I sent it via chat.

14 00:04:02.070 00:04:03.110 Uttam Kumaran: Hello, sir.

15 00:04:03.110 00:04:04.000 Craig Foldes: Dude.

16 00:04:04.750 00:04:06.039 Uttam Kumaran: I’m sorry!

17 00:04:06.040 00:04:09.920 Craig Foldes: 15 minutes without a heads up is not cool, I’m not trying to be a dick. What’s up? Run me through it.

18 00:04:09.920 00:04:14.079 Uttam Kumaran: Well, I’ve just… I was sharing my screen, I just couldn’t do anything on the.

19 00:04:14.080 00:04:16.290 Craig Foldes: I just didn’t sleep well last night, so I’m not in the best mood.

20 00:04:16.290 00:04:17.179 Uttam Kumaran: I’m sorry, I’m sorry.

21 00:04:17.180 00:04:19.389 Craig Foldes: tomorrow, and are we gonna do it or not?

22 00:04:19.390 00:04:26.910 Uttam Kumaran: Yes, we… we have it, I can walk us through it. So basically, we developed, like.

23 00:04:27.630 00:04:28.609 Craig Foldes: What’s up, Casey?

24 00:04:29.580 00:04:30.510 Casie Aviles: Hey, Craig.

25 00:04:30.510 00:04:31.410 Craig Foldes: How are you?

26 00:04:32.250 00:04:33.459 Casie Aviles: Yeah, doing good.

27 00:04:33.460 00:04:44.489 Craig Foldes: Alright. So, yeah, give me the rundown. So we got 30 minutes, there’s probably gonna be 70 fucking people, let’s… and it… let’s just, like, super tactical, like, very tactical, start to finish.

28 00:04:44.970 00:05:03.339 Uttam Kumaran: Yeah, so we basically… the demo is, like, we have a form where, you can just submit, like, a transcript, so I can show an example of me taking a sales call, putting it in here. It generates, examples of, like, the writing, and then, basically you could pick

29 00:05:04.910 00:05:09.600 Uttam Kumaran: You can label, what is it, Casey? It works like you label first, or do you label second?

30 00:05:10.270 00:05:14.249 Casie Aviles: I label second, so I create the draft first, and then it gets labeled.

31 00:05:14.940 00:05:24.550 Uttam Kumaran: Yeah, so it creates an email draft, and then it gets labeled as, like, hey, AI wrote the draft for you, in your style. So this is something that I can…

32 00:05:24.770 00:05:44.110 Uttam Kumaran: I can either walk through creating all this live, if instead it’s, like, the biggest things for people to really learn about N8N is actually more about how to use, like, this LLM chain. Like, how do you actually call an LLM directly within here? Like, how do you write the prompt?

33 00:05:44.110 00:05:56.689 Uttam Kumaran: And then this is a great take-home, so everybody can take this with them and use within their workspace. Ideally, what we’re trying to show is the original pitch, which is you submit a form,

34 00:05:57.880 00:05:59.729 Uttam Kumaran: And then it just goes through this, like.

35 00:06:00.110 00:06:01.970 Uttam Kumaran: Writing a draft in your style.

36 00:06:02.530 00:06:07.780 Craig Foldes: Let’s… alright, so, I think… give me a second. So, for tomorrow.

37 00:06:07.880 00:06:10.110 Craig Foldes: Sorry, I didn’t mean to be a dick, I’m just not… I’m tired.

38 00:06:10.110 00:06:11.620 Uttam Kumaran: No, no, no, that’s fine, no, no, no.

39 00:06:11.820 00:06:30.080 Craig Foldes: Alright, let’s start with, like, obviously, like, 30 seconds on who you are, what you do, and then be like, hey, I’ll give a context, hey, we work together, I love these guys, and then I think you need to just talk about what N8N is. Nobody’s gonna know, kind of, what it is. So, like, literally just spend a minute or two, like, this is what N8N is, it operates, you know, workflows.

40 00:06:30.080 00:06:32.339 Craig Foldes: Craig asked that we do this thing.

41 00:06:32.340 00:06:33.580 Uttam Kumaran: Sure. So then…

42 00:06:33.790 00:06:38.010 Craig Foldes: I’m… I’m a little lost for a second, so what… what is this form?

43 00:06:38.620 00:06:43.309 Uttam Kumaran: Yeah, so, Casey, can I get the form link? Okay, yeah, here it is.

44 00:06:44.550 00:06:53.500 Craig Foldes: I would… I would treat this as more like showing off what NIT is, and it is, and what it’s capable of. So, thank you for doing the upfront work. So, what…

45 00:06:53.500 00:07:10.409 Uttam Kumaran: Okay, yeah, so basically we… there’s a simple form where, for example, I was… I will say, Craig tasked me with, hey, I’m very busy, a lot of my emails can pretty much be done by a VA, but I don’t have time for that, or I don’t want to get one. I would like AI to handle some portion of that.

46 00:07:10.500 00:07:25.220 Uttam Kumaran: we can… we built a simple thing that… that you plug in a couple of transcripts, it understands your voice, and then now you can able… you can label messages, it then sends a notification to NNN to write a draft.

47 00:07:25.460 00:07:30.479 Uttam Kumaran: The NADN agent looks at that email, understands your voice, writes a draft.

48 00:07:30.610 00:07:37.949 Uttam Kumaran: And that’s… that’s simple. This is something that you can set up on your machine for free, basically, in, like, an hour.

49 00:07:38.530 00:07:44.789 Craig Foldes: So, I would have, like, a visual, like, that end-and-end thing, like, I would create a slide, and I’m happy to do it with you, like.

50 00:07:44.790 00:07:45.160 Uttam Kumaran: Yeah.

51 00:07:45.160 00:07:51.249 Craig Foldes: like, we are here, and then constantly refer back to that slide in terms of… so can you…

52 00:07:51.470 00:07:52.039 Craig Foldes: Can you…

53 00:07:52.040 00:07:55.529 Uttam Kumaran: So, like, for example, like, we have… we have, like, an AI, like.

54 00:07:55.670 00:08:03.709 Uttam Kumaran: deck with some good, like, workflow visualizations. For example, I would show something like, here’s, like, where we are, right?

55 00:08:03.850 00:08:21.160 Uttam Kumaran: So this is something that we were gonna produce, it’s just like, this is you, you’re talking to NAN, there’s, like, knowledge, and then here’s, like, the flow. This is a little bit complex, this is from another client, but I would have, like, this, and then basically, side by side, I can either share the NAN, or I can just…

56 00:08:21.660 00:08:25.039 Uttam Kumaran: do-it-all slides. Where I don’t want to kind of get stuck is, like.

57 00:08:25.210 00:08:30.729 Uttam Kumaran: in how complicated NAN is. I actually just want to show the more complicated pieces and say.

58 00:08:30.870 00:08:38.320 Uttam Kumaran: the link is gonna get emailed to you, you can copy this directly to your free NNN account, plug in your Gmail, and try it today.

59 00:08:39.340 00:08:44.370 Craig Foldes: Can you just show it to me? Like, can you paste… can you walk me through the whole workflow, start to finish?

60 00:08:44.540 00:08:48.310 Uttam Kumaran: Yeah, so if I was to take… Like…

61 00:08:48.470 00:08:52.319 Uttam Kumaran: a transcript, and Casey, is this tied to your Gmail right now?

62 00:08:52.960 00:08:56.629 Casie Aviles: Oh, yeah, it’s only gonna show in my… in my drafts for now.

63 00:08:56.630 00:08:59.109 Uttam Kumaran: So, so basically what I would do is…

64 00:08:59.530 00:09:07.799 Uttam Kumaran: I’d reconnect in my email. What we would do here is we have this form, I would take, let’s just say I took,

65 00:09:08.360 00:09:16.589 Uttam Kumaran: Something from today, let’s say… this call… so I have, like… My granola here.

66 00:09:16.590 00:09:18.680 Craig Foldes: Oh, nice, dude, dope, congratulations.

67 00:09:18.990 00:09:21.710 Uttam Kumaran: Thank you. Yeah, Honey Singer’s cool. What are you doing?

68 00:09:22.290 00:09:23.410 Uttam Kumaran: I tried… huh?

69 00:09:23.410 00:09:24.700 Craig Foldes: What are you doing with them?

70 00:09:24.700 00:09:30.889 Uttam Kumaran: we’re doing omnichannel reporting for them, so they’re selling Amazon, Shopify.

71 00:09:30.890 00:09:31.400 Craig Foldes: Cool.

72 00:09:31.400 00:09:34.920 Uttam Kumaran: Starting to do wholesale, they just need help on, like, stitching it together.

73 00:09:34.920 00:09:37.190 Craig Foldes: Honey Stinger the waffle company?

74 00:09:37.190 00:09:38.099 Uttam Kumaran: Waffle Company, yeah.

75 00:09:38.100 00:09:39.780 Craig Foldes: How’d you get connected to them? Cool.

76 00:09:39.780 00:09:42.439 Uttam Kumaran: We went to Shop Talk in Chicago.

77 00:09:42.440 00:09:44.240 Craig Foldes: Nice. And just, like, met…

78 00:09:44.570 00:09:46.150 Uttam Kumaran: Met the head of e-comm.

79 00:09:46.450 00:09:52.560 Uttam Kumaran: On the floor, and then… We’re just like, oh shit, yeah, we can do that problem for you.

80 00:09:52.560 00:09:54.690 Craig Foldes: And is… are they in Steamboat?

81 00:09:55.550 00:09:56.819 Uttam Kumaran: What is that?

82 00:09:56.960 00:09:58.520 Craig Foldes: Steamboat Springs? Is that what.

83 00:09:58.520 00:10:01.540 Uttam Kumaran: Oh, yeah, in Portland somewhere, I think.

84 00:10:01.540 00:10:02.290 Craig Foldes: So cool, alright?

85 00:10:02.290 00:10:03.530 Uttam Kumaran: Somewhere on the West Coast.

86 00:10:03.850 00:10:08.429 Craig Foldes: So, long story short, you have this transcript, paste it in, then what happens?

87 00:10:08.430 00:10:10.490 Uttam Kumaran: We hit submit, and then…

88 00:10:12.650 00:10:17.879 Uttam Kumaran: It gets recorded. Can I see the execution happen, Casey? It’s this one, right?

89 00:10:17.880 00:10:18.590 Casie Aviles: Yes.

90 00:10:18.970 00:10:19.720 Casie Aviles: Yep.

91 00:10:20.890 00:10:23.950 Uttam Kumaran: So I would basically show… Bye.

92 00:10:24.460 00:10:43.519 Uttam Kumaran: we could either have this live, and I think it would just walk through, but basically, I would share that what it’s doing is it’s setting the writing examples, so it’s bringing the transcript in as a writing example. As you can see, this is the Honey Singer meeting. This writing example is then getting passed to AI,

93 00:10:43.620 00:10:54.720 Uttam Kumaran: we have a… we have a AI prompt in here that is saying, your goal is to identify, basically, it’s… it’s what we use for our sales follow-up emails.

94 00:10:54.720 00:11:03.769 Uttam Kumaran: So, you’re gonna take the writing example, and you’re trained as a sales follow-up assistant, your job is to write the follow-up.

95 00:11:04.040 00:11:15.400 Uttam Kumaran: I would show how you can bring your… any model you want. Here, we’re using OpenAI. It would then go into Gmail, it would then create a draft, and then, ultimately, it would…

96 00:11:16.480 00:11:19.830 Uttam Kumaran: it would… I could then pop my Gmail up, and…

97 00:11:20.070 00:11:22.200 Uttam Kumaran: show that the draft is written. And this…

98 00:11:22.200 00:11:25.730 Craig Foldes: show the draft now. I think that, like, structure, I’m just gonna drive this.

99 00:11:25.730 00:11:30.799 Uttam Kumaran: Yeah, Casey, do you wanna… do you wanna share the draft on… it’s on… it’s on his Gmail?

100 00:11:31.580 00:11:34.530 Casie Aviles: Oh, sure. I can share my screen right now.

101 00:11:38.270 00:11:38.940 Casie Aviles: Okay.

102 00:11:39.290 00:11:42.319 Uttam Kumaran: And so I can have granola, NADN, and…

103 00:11:42.320 00:11:43.030 Craig Foldes: Hold on one sec.

104 00:11:43.030 00:11:44.340 Uttam Kumaran: basically Gmail app.

105 00:11:44.670 00:11:50.280 Craig Foldes: So this is the draft, this access, setup, and next step type thing for, I mean, the setup earlier, I’m excited to get started with Main Forge.

106 00:11:51.020 00:11:54.530 Craig Foldes: wait, time out. I’m excited to get started with Brain Forge.

107 00:11:55.040 00:11:59.779 Craig Foldes: Africa, Klaviyo, Walmart seems to… so I’ll keep an eye. I’ll send you a shot. Wait.

108 00:11:59.910 00:12:07.509 Craig Foldes: Time out. But I’m excited to get started with Brainforge. Wouldn’t this be you sending the email out?

109 00:12:08.010 00:12:13.270 Uttam Kumaran: So yeah, I mean, but this is like a… so we can modify the prompt, but… Basically, it’s like.

110 00:12:13.640 00:12:25.990 Uttam Kumaran: We’re excited to get started. Like, this is just the draft that gets written. So, we can also change this so that it replies to an email draft. This situation is just creating a draft in your inbox.

111 00:12:26.390 00:12:29.869 Craig Foldes: No, I know, but wouldn’t this be an email that you’re sending back to Honey?

112 00:12:29.870 00:12:30.740 Casie Aviles: Yeah.

113 00:12:30.740 00:12:33.780 Craig Foldes: So why does it say, I’m excited to get started with Brainforge?

114 00:12:34.340 00:12:35.530 Casie Aviles: It’s just a prompt.

115 00:12:36.170 00:12:38.299 Casie Aviles: We’ll have to fix the prompt here.

116 00:12:39.000 00:12:41.749 Casie Aviles: We can definitely just, you know, look at the…

117 00:12:42.650 00:12:46.669 Casie Aviles: I’ll get the transcript, and then… Like, specify here.

118 00:12:47.140 00:12:51.180 Casie Aviles: That it should be… You know, we can add, like…

119 00:12:51.700 00:12:54.300 Casie Aviles: Make sure that Utam is the one who’s writing it.

120 00:12:55.120 00:12:58.650 Casie Aviles: And we can just basically improve the prompt here and fine-tune it.

121 00:12:58.930 00:13:04.529 Casie Aviles: So that it sends it that way, specified rules, for example.

122 00:13:05.820 00:13:15.650 Uttam Kumaran: Yeah, we can… so we basically would just change a prompt that says, you’re sending this from Brainforge. But I guess, what’s more… like, what is more compelling? Do we want it to reply to an existing thread, or is this…

123 00:13:15.950 00:13:21.629 Uttam Kumaran: All this is, is, like, shove something in, generate me… The follow-up email.

124 00:13:21.630 00:13:28.699 Craig Foldes: here’s… here’s how I think that we should do it. And Utami, like, I’d like for you to drive, but if you’re not, like… like…

125 00:13:29.240 00:13:35.909 Craig Foldes: Give me a second, just… and then I unfortunately have to jump in 3 minutes. So give me a second.

126 00:13:39.380 00:13:40.540 Craig Foldes: I would like…

127 00:13:40.730 00:13:50.769 Craig Foldes: Ideally, you or me or whoever, you show your screen, and you say, hey, I’m Utam, I run Brainforge, this is a call transcript that I had with a customer.

128 00:13:50.810 00:14:00.839 Craig Foldes: Sure. To close a deal, I’m gonna copy and paste that call transcript into this form, and then I’m gonna share my screen and show you the output, which is an email.

129 00:14:00.840 00:14:15.310 Craig Foldes: Right, and then from there, I would say, and here are the step-by-steps on how I built that. And then what I would like to do is, if we can, is literally start from scratch in N&, and show the entire workflow, start to finish, and, like, build it out.

130 00:14:15.310 00:14:29.599 Craig Foldes: Right? So, I showed you the output, here’s how I created it. By the way, in 15 minutes, you can too. Is that… like, I think the goal here is, like, tactical training on what N8N is and how it works. Are you able to drive that start to finish?

131 00:14:30.330 00:14:35.229 Uttam Kumaran: Yeah, we would need to slim down this workflow a bit, then.

132 00:14:38.130 00:14:45.149 Uttam Kumaran: The biggest things I want to get to people is how to use this, like, AI piece.

133 00:14:45.630 00:14:47.779 Uttam Kumaran: And then how to use the Gmail node.

134 00:14:47.950 00:14:54.000 Uttam Kumaran: So, I almost may just do… like…

135 00:14:54.190 00:14:55.350 Craig Foldes: this?

136 00:14:55.480 00:15:02.210 Uttam Kumaran: And I think… I could do that in, like, 20 minutes?

137 00:15:03.030 00:15:06.280 Craig Foldes: The challenge is, none of these people are technical, they’re the CEOs.

138 00:15:06.280 00:15:06.830 Uttam Kumaran: Yeah.

139 00:15:07.120 00:15:19.669 Craig Foldes: companies that do hundreds of millions a year, and none of them are gonna, gonna go through this. I think it’s just more like, holy shit, this is what’s possible. Now I know, by the way, let me reach out to Brainforge to build this for me or something.

140 00:15:19.670 00:15:26.960 Uttam Kumaran: This is pretty easy, like, what I will prep is probably, on my side, I’m gonna prep the prompt, and then, like, that way I’m not, like, writing

141 00:15:27.820 00:15:44.260 Uttam Kumaran: I guess it’s… if it’s okay with you, I’m not gonna write the prompt, like, from scratch or anything. Mainly, I’m gonna be like, okay, here’s how to create a node. And logically, next, we need to talk to Gmail. So here’s how to create the Gmail node, here’s how to connect it. In Gmail, as we mentioned, our goal is to create a draft, so I’m gonna select

142 00:15:44.260 00:15:51.019 Uttam Kumaran: the create draft option, and then tie the… take the input from the LLM, and put it in here, and then that’s…

143 00:15:51.050 00:15:52.850 Uttam Kumaran: I’ll kind of walk through like that.

144 00:15:53.690 00:16:03.069 Craig Foldes: cool, but I think that that’s perfect, but again, I just want to, like, have… you don’t have to create the prompts in real time, I just kind of want to have the,

145 00:16:03.230 00:16:11.310 Craig Foldes: like, I want to show the output, right? Like, start to finish, paste it into that, get a really strong output, and then be like, by the way, here’s how I built it all.

146 00:16:11.310 00:16:16.199 Uttam Kumaran: Okay, okay. Like, do you… do you… so what to… to kind of run… run it back.

147 00:16:16.580 00:16:30.849 Uttam Kumaran: I’ll start, I’ll intro Brainforge. What I’m gonna do is say, today we wanted to show you how, in just 20 minutes, you can build a pretty simple email follow-up writer that can

148 00:16:30.930 00:16:50.099 Uttam Kumaran: put emails right into your email drafts. I’ll mention that I use this, like, all the time, because we do use this system, a similar one, all the time, and I can now send emails that go… I don’t have to trust AI to send the emails out, but I also… it’s way faster than having to copy and paste from Chagbt. And then I’ll… Oh, yeah.

149 00:16:50.530 00:16:55.909 Craig Foldes: Do you have… unfortunately, I have to go. Do you have, utam, do you have another 30 minutes later today? Because I think.

150 00:16:55.910 00:16:56.770 Uttam Kumaran: Yeah, so…

151 00:16:56.770 00:17:01.550 Craig Foldes: We’ll cancel these, like, so if we’re not in a good place, like, we won’t run it, so I just wanted.

152 00:17:01.550 00:17:10.809 Uttam Kumaran: So I’ll also record a Loom and walking through… I can just record a Loom walkthrough of me doing it, send it to you, and then, yeah, we can also chat this afternoon.

153 00:17:10.819 00:17:14.409 Craig Foldes: Alright, that’s a pain in the ass, I’m sorry to make you kind of, like, have to record everything ahead of time.

154 00:17:14.410 00:17:22.739 Uttam Kumaran: Alright, no, I don’t care. I don’t mind doing it, yeah. I mean, it’s… I wanna… I’m gonna do a dry run at least once or twice on my own, so…

155 00:17:22.740 00:17:27.100 Craig Foldes: Yeah, I would suggest that… hold on, let me look at who’s attending.

156 00:17:37.130 00:17:39.460 Craig Foldes: Sorry, give me one second.

157 00:17:47.160 00:17:48.490 Craig Foldes: Yeah, they’ll be like.

158 00:17:48.760 00:17:53.449 Craig Foldes: 85 people have already confirmed yes, so it’s worth doing right, and if it’s not good.

159 00:17:53.450 00:18:01.499 Uttam Kumaran: No, no, no, I’m gonna go… I’m gonna do that myself anyways, so I can record it, send it to you, and then I’m free, basically, after…

160 00:18:01.620 00:18:04.109 Uttam Kumaran: 4.30 Central, like, the rest of the day.

161 00:18:04.200 00:18:06.089 Craig Foldes: So, that’s my plan.

162 00:18:06.090 00:18:06.990 Uttam Kumaran: This evening.

163 00:18:07.180 00:18:16.920 Craig Foldes: Let’s… let’s do that. So again, this is more an N8N demo, and how to build an A to N using this email as an example. So I would kick off with, like, hey, like, wanted to show.

164 00:18:16.920 00:18:27.119 Uttam Kumaran: I’m not gonna do any slides then, like, necessarily, because this is already fairly visual, and it’s… it’s already tight on time, so I’m just gonna start with just showing, like.

165 00:18:27.340 00:18:36.979 Uttam Kumaran: Going from here, I’ll refresh the Gmail, it’ll be there, and then I’ll hop in NHN, and then I’ll try to leave a couple minutes for questions, or whatever you want to do.

166 00:18:37.200 00:18:43.269 Craig Foldes: I’ll do Q&A throughout, I’ll guide, and then I’ll collect questions.

167 00:18:43.420 00:18:47.249 Craig Foldes: Okay, I’ll send you a note, we’ll record a loom, we’ll connect later. Later, guys.

168 00:18:47.280 00:18:49.079 Uttam Kumaran: Casey, good to see you, I miss you.

169 00:18:51.030 00:18:51.850 Casie Aviles: Bye.

170 00:18:52.550 00:18:54.350 Uttam Kumaran: Casey, you wanna stay on for a sec?

171 00:18:55.250 00:18:56.159 Casie Aviles: Yeah, sure.

172 00:18:56.310 00:19:07.790 Uttam Kumaran: Yeah, so a couple of tweaks that would be great. One is… Let’s remove, let’s remove the…

173 00:19:07.970 00:19:09.880 Uttam Kumaran: labels piece?

174 00:19:10.690 00:19:15.439 Casie Aviles: Okay. And then… Alright, let me share.

175 00:19:16.470 00:19:20.619 Uttam Kumaran: I think my other question is going to be,

176 00:19:36.570 00:19:40.799 Uttam Kumaran: So yeah, let’s remove, like, that, like, everything kind of to the right of that.

177 00:19:40.940 00:19:42.940 Uttam Kumaran: And then…

178 00:19:43.190 00:19:51.730 Uttam Kumaran: Yeah, you can just leave it somewhere. And then, my other question is, can we, instead of, can we, instead of this, can we just use the normal ChatGPT?

179 00:19:52.580 00:19:54.460 Uttam Kumaran: like… node.

180 00:19:54.460 00:19:55.560 Casie Aviles: Oh, okay.

181 00:19:55.690 00:19:58.210 Uttam Kumaran: Like, the simplest OpenAI node? Yeah.

182 00:20:01.540 00:20:08.120 Uttam Kumaran: Like, and do we even need to do the whole thing, the LLM chain, if it’s just one output?

183 00:20:13.020 00:20:16.760 Casie Aviles: Wait, well, I mean, it’s only just this anyway,

184 00:20:17.340 00:20:23.149 Casie Aviles: And for here, yeah, I guess we can remove the output parser if…

185 00:20:24.050 00:20:27.289 Casie Aviles: If it’s too technical for the audience.

186 00:20:28.920 00:20:37.560 Casie Aviles: If we could just… Make it so it just outputs… But, but I guess, like…

187 00:20:38.240 00:20:43.020 Casie Aviles: We want to be able to generate the subject and the message as well.

188 00:20:43.360 00:20:44.890 Casie Aviles: That’s why I used to output.

189 00:20:44.890 00:20:45.360 Uttam Kumaran: Oh, okay.

190 00:20:45.360 00:20:46.140 Casie Aviles: anniversary.

191 00:20:48.380 00:20:49.060 Casie Aviles: Wow.

192 00:20:49.300 00:20:53.210 Casie Aviles: Let me change this with… I’m opening…

193 00:20:58.330 00:21:03.949 Casie Aviles: Not sure if it’s this one’s working, since we haven’t been using it, but let’s see.

194 00:21:05.190 00:21:10.310 Casie Aviles: I think… I wasn’t sure, like, what the granola had.

195 00:21:10.640 00:21:15.089 Casie Aviles: So, I was using one of your sales calls before.

196 00:21:15.320 00:21:17.359 Casie Aviles: Which had, like, your name and…

197 00:21:17.360 00:21:18.139 Uttam Kumaran: Yeah, yeah, yeah.

198 00:21:18.140 00:21:19.050 Casie Aviles: other person.

199 00:21:19.530 00:21:24.319 Uttam Kumaran: That’s fine, that’s fine. I could… I’ll use that, or I’ll test it a little bit more.

200 00:21:25.140 00:21:26.050 Casie Aviles: Oh, wait.

201 00:21:32.280 00:21:35.960 Casie Aviles: Okay, I think we are… we need to set something. Oh, the quota.

202 00:21:36.350 00:21:37.530 Casie Aviles: I don’t have…

203 00:21:46.240 00:21:53.190 Casie Aviles: I don’t think… are we still… are we, like, paying for our OpenAI platform still?

204 00:21:53.440 00:21:54.090 Uttam Kumaran: Yeah.

205 00:21:59.960 00:22:05.880 Uttam Kumaran: Maybe let me, yeah, let me send you a voice note. I just have to run, because I have to drive to another…

206 00:22:06.070 00:22:11.010 Casie Aviles: Okay. But yeah, basically, I also don’t know, like, how easy it would be for us to, like.

207 00:22:17.550 00:22:21.940 Uttam Kumaran: Yeah, I think this is the easiest use case versus responding to an email, right?

208 00:22:23.200 00:22:26.389 Uttam Kumaran: Otherwise, we have to somehow be, like, select the email.

209 00:22:27.920 00:22:31.399 Uttam Kumaran: Can the form potentially let us select an email, or no?

210 00:22:34.610 00:22:36.010 Uttam Kumaran: Not really.

211 00:22:37.050 00:22:41.769 Casie Aviles: Yeah, this is… this just takes in, like, the transcript, so I think that’s gonna be more…

212 00:22:42.150 00:22:53.060 Uttam Kumaran: Okay, yeah, I think this is enough. I think, yeah, let’s just use OpenAI. I can show the chain, like, the point of the chain, and the parser, and then we’ll create a draft.

213 00:22:54.870 00:22:55.510 Casie Aviles: Okay.

214 00:22:58.000 00:23:10.469 Uttam Kumaran: Yeah, just Slack me. I’m gonna be kind of in and out on my laptop over the next 2 hours, but I’ll Slack with you. And then, yeah, if you could just test, like, for example, if we could test with…

215 00:23:23.910 00:23:27.579 Uttam Kumaran: I’ll send you a granola, that we could test with.

216 00:23:27.760 00:23:31.900 Uttam Kumaran: It’s not gonna have… like… everything.

217 00:23:31.900 00:23:32.520 Casie Aviles: So I think…

218 00:23:32.520 00:23:37.099 Uttam Kumaran: the best thing to… for us to figure out is, like, what’s the output format for Granola?

219 00:23:37.890 00:23:39.300 Casie Aviles: I see, yeah, okay.

220 00:23:42.630 00:23:45.270 Uttam Kumaran: But I’ll send you this snippet,

221 00:23:48.120 00:23:54.490 Uttam Kumaran: It says me and them, though, so it’s… you should be able to infer. And yeah, as long as this is working…

222 00:23:55.020 00:23:56.069 Uttam Kumaran: Should be good.

223 00:23:57.890 00:23:58.530 Casie Aviles: Okay.

224 00:23:59.290 00:24:02.099 Uttam Kumaran: Okay, cool. Thanks, Sue, this is really helpful.

225 00:24:02.220 00:24:04.660 Uttam Kumaran: Yeah, I’ll talk to you later today, probably.

226 00:24:05.140 00:24:06.039 Casie Aviles: Okay, thank you, Vital.

227 00:24:06.310 00:24:07.160 Uttam Kumaran: Thank you.