Meeting Title: Brainforge AI Workflow Setup Sync Date: 2025-11-04 Meeting participants: Casie Aviles, Uttam Kumaran
WEBVTT
1 00:11:57.740 ⇒ 00:11:59.339 Uttam Kumaran: It is.
2 00:12:00.290 ⇒ 00:12:01.389 Casie Aviles: Heh.
3 00:12:03.110 ⇒ 00:12:04.140 Uttam Kumaran: Hey, sorry.
4 00:12:05.210 ⇒ 00:12:18.239 Uttam Kumaran: Okay… Let me… Okay, let me know where you ended up, and then I’m gonna read…
5 00:12:19.480 ⇒ 00:12:22.630 Casie Aviles: Okay, I can go share.
6 00:12:29.610 ⇒ 00:12:36.319 Casie Aviles: Yeah, so… what I did… For now, as I just updated the prompt,
7 00:12:38.110 ⇒ 00:12:46.159 Casie Aviles: I made sure that the AI knows that it’s you, that you’ll be the one sending out the emails, and then…
8 00:12:47.220 ⇒ 00:12:58.300 Casie Aviles: I added this… OpenAI nodes, but it’s using NHN free credits. I can’t seem to get our… OpenAI working…
9 00:12:59.700 ⇒ 00:13:00.400 Casie Aviles: Like…
10 00:13:00.400 ⇒ 00:13:04.099 Uttam Kumaran: Oh, okay, yum, I can check on it now.
11 00:13:07.660 ⇒ 00:13:11.760 Casie Aviles: Yeah, because it says here we have, like, zero Reddit balance.
12 00:13:14.510 ⇒ 00:13:16.060 Uttam Kumaran: Let me log in to chat.
13 00:13:58.340 ⇒ 00:14:03.170 Uttam Kumaran: Okay, so I’m in… our org.
14 00:14:03.300 ⇒ 00:14:05.870 Uttam Kumaran: Which project is it?
15 00:14:08.010 ⇒ 00:14:12.279 Casie Aviles: I’m… I created an API key under N8N.
16 00:14:15.210 ⇒ 00:14:16.099 Uttam Kumaran: Oh, good one.
17 00:14:22.250 ⇒ 00:14:24.890 Uttam Kumaran: Oh, I see, okay, okay.
18 00:14:42.480 ⇒ 00:14:44.799 Uttam Kumaran: Okay, I added, $20.
19 00:14:46.230 ⇒ 00:14:46.920 Casie Aviles: Alright.
20 00:15:13.840 ⇒ 00:15:14.880 Casie Aviles: Let’s fair.
21 00:15:15.790 ⇒ 00:15:24.110 Casie Aviles: Let’s see… I’m using this transcript from… The call source meaning.
22 00:15:31.440 ⇒ 00:15:35.679 Casie Aviles: Okay, let’s see… Okay, perfect, yeah.
23 00:15:40.290 ⇒ 00:15:45.180 Casie Aviles: So, this is the… Interest…
24 00:15:52.420 ⇒ 00:15:56.889 Casie Aviles: Okay, yeah, it did generate a draft, although it says call sourcing.
25 00:15:57.430 ⇒ 00:15:59.250 Casie Aviles: Should be ABC, right?
26 00:16:05.450 ⇒ 00:16:06.480 Uttam Kumaran: Yes.
27 00:16:34.900 ⇒ 00:16:40.779 Casie Aviles: I’m just gonna make a change to the prompt. I’m gonna run our change, and then…
28 00:16:51.370 ⇒ 00:16:53.009 Casie Aviles: Right, give me a second.
29 00:18:23.960 ⇒ 00:18:24.650 Casie Aviles: Okay.
30 00:18:25.080 ⇒ 00:18:32.160 Casie Aviles: Yeah, I think this is… I mean, yeah, this is better than… The last draft,
31 00:18:33.630 ⇒ 00:18:41.270 Casie Aviles: Yeah, I think that’s pretty much it. I just have to place this with your… Gmail now.
32 00:18:41.400 ⇒ 00:18:42.549 Casie Aviles: So I’m gonna…
33 00:18:42.850 ⇒ 00:18:47.350 Uttam Kumaran: Yeah, so, maybe one thing I wanna… I can run through with you, Casey, because I’m gonna…
34 00:18:47.720 ⇒ 00:18:53.290 Uttam Kumaran: try to run through it on my own. So maybe what I can do is I can share and then run through
35 00:18:54.530 ⇒ 00:18:57.969 Uttam Kumaran: kind of how I’m gonna present it, and maybe you can just let me know what you think.
36 00:18:58.700 ⇒ 00:18:59.790 Casie Aviles: Okay, yeah.
37 00:19:17.620 ⇒ 00:19:21.289 Casie Aviles: Alright, I’ll just, send the links here as well.
38 00:19:34.480 ⇒ 00:19:38.239 Casie Aviles: It should be connected to your Gmail account now.
39 00:19:44.780 ⇒ 00:19:45.820 Uttam Kumaran: So…
40 00:21:27.230 ⇒ 00:21:29.040 Uttam Kumaran: Okay, so…
41 00:21:32.560 ⇒ 00:21:40.760 Uttam Kumaran: I just want to save a couple of the core things here, so… I want to save, like,
42 00:21:45.310 ⇒ 00:21:47.180 Uttam Kumaran: This is gonna be the prompt.
43 00:21:47.860 ⇒ 00:21:51.000 Uttam Kumaran: And then, what else is there that’s, like.
44 00:21:54.650 ⇒ 00:21:57.590 Casie Aviles: What else did I just have, like, ready to copy-paste? Yeah.
45 00:21:57.880 ⇒ 00:22:04.549 Casie Aviles: Oh yeah, the prompt… oh, what else? I think that the other writing examples… I just generated.
46 00:22:04.550 ⇒ 00:22:07.199 Uttam Kumaran: This, but, but, okay, this is generated.
47 00:22:07.200 ⇒ 00:22:07.610 Casie Aviles: Yeah.
48 00:22:07.610 ⇒ 00:22:11.120 Uttam Kumaran: Prompt, okay.
49 00:22:17.000 ⇒ 00:22:17.710 Uttam Kumaran: A.
50 00:22:17.780 ⇒ 00:22:21.070 Casie Aviles: I think this structured output parser.
51 00:22:22.290 ⇒ 00:22:24.210 Uttam Kumaran: Oh, they don’t have a GPT-5 API?
52 00:22:26.360 ⇒ 00:22:28.620 Uttam Kumaran: But, should we just use 5 for the demo?
53 00:22:29.660 ⇒ 00:22:31.740 Casie Aviles: Yeah, I think we could also use 5.
54 00:22:33.320 ⇒ 00:22:35.060 Uttam Kumaran: LA, it’s just like me and grab.
55 00:22:36.200 ⇒ 00:22:41.259 Uttam Kumaran: liquid output parser, Oh, so this is an example.
56 00:22:41.400 ⇒ 00:22:42.100 Casie Aviles: Yes.
57 00:23:01.450 ⇒ 00:23:02.400 Uttam Kumaran: Cool.
58 00:23:08.450 ⇒ 00:23:13.899 Casie Aviles: It still has, like… That classification, I’m not sure if we still need that, though, but…
59 00:23:14.210 ⇒ 00:23:17.379 Uttam Kumaran: So, for the demo, do you think it would be easier to do it this way?
60 00:23:20.770 ⇒ 00:23:21.540 Casie Aviles: Can you go and.
61 00:23:21.540 ⇒ 00:23:24.869 Uttam Kumaran: Oh, this… oh, actually, nevermind. Yeah, this one is not a UI.
62 00:23:25.990 ⇒ 00:23:27.819 Casie Aviles: Yeah, I thought it would be Fields.
63 00:23:29.260 ⇒ 00:23:30.240 Uttam Kumaran: Yeah, okay.
64 00:23:36.010 ⇒ 00:23:38.029 Casie Aviles: Yeah, I think that one’s easier.
65 00:23:43.940 ⇒ 00:23:47.159 Uttam Kumaran: And in NAVN, you can’t connect both of these to one.
66 00:23:47.890 ⇒ 00:23:49.649 Uttam Kumaran: You have to have two model.
67 00:23:50.400 ⇒ 00:23:53.949 Casie Aviles: You could, but I… I try to avoid it.
68 00:23:54.240 ⇒ 00:23:58.550 Casie Aviles: Because sometimes, when it runs too quickly, it gets decorated.
69 00:23:58.550 ⇒ 00:24:01.020 Uttam Kumaran: Oh, okay, okay, okay.
70 00:24:18.540 ⇒ 00:24:19.480 Uttam Kumaran: Nice.
71 00:24:19.860 ⇒ 00:24:27.770 Uttam Kumaran: So, let’s say, I was to go build this from… Scratch.
72 00:24:27.990 ⇒ 00:24:28.780 Uttam Kumaran: Right.
73 00:24:29.150 ⇒ 00:24:31.020 Uttam Kumaran: And let’s say in the,
74 00:24:31.780 ⇒ 00:24:34.520 Uttam Kumaran: I want to actually end up doing it,
75 00:24:38.830 ⇒ 00:24:40.829 Uttam Kumaran: Oh, so this is our prompt, right?
76 00:24:41.670 ⇒ 00:24:43.859 Casie Aviles: Yeah, this is… For… yeah.
77 00:24:43.860 ⇒ 00:24:44.830 Uttam Kumaran: Our sales follow-up contact.
78 00:24:44.830 ⇒ 00:24:45.350 Casie Aviles: Yes.
79 00:24:48.580 ⇒ 00:24:49.260 Uttam Kumaran: Cool.
80 00:24:50.750 ⇒ 00:24:55.780 Uttam Kumaran: But basically, what I would do is, I’m like… I would tell them, like, this is what we’re trying to build.
81 00:24:56.390 ⇒ 00:25:01.059 Uttam Kumaran: And so, I’m just gonna go ahead and do it right underneath.
82 00:25:01.330 ⇒ 00:25:08.090 Uttam Kumaran: And so… let’s say we were to begin that, I’m gonna go to plus, I’m gonna say form.
83 00:25:08.580 ⇒ 00:25:14.120 Uttam Kumaran: So one thing I can just put in here is, like, The, like, components.
84 00:25:15.020 ⇒ 00:25:19.979 Uttam Kumaran: So we need… one is… And they then removed the storm component.
85 00:25:20.500 ⇒ 00:25:24.049 Uttam Kumaran: We need the… so I’ll click on any of the form.
86 00:25:24.780 ⇒ 00:25:28.340 Uttam Kumaran: And… I would just click… form ending.
87 00:25:28.960 ⇒ 00:25:31.929 Casie Aviles: Hmm, these are actions, it should be triggered.
88 00:25:31.930 ⇒ 00:25:33.520 Uttam Kumaran: Oh, okay, okay.
89 00:25:36.250 ⇒ 00:25:36.970 Casie Aviles: Add an.
90 00:25:36.970 ⇒ 00:25:37.849 Uttam Kumaran: Oh, I don’t trigger it.
91 00:25:40.940 ⇒ 00:25:41.999 Casie Aviles: Yeah, that one.
92 00:25:44.550 ⇒ 00:25:48.310 Uttam Kumaran: Okay, so I click this…
93 00:25:49.010 ⇒ 00:25:56.770 Uttam Kumaran: The forum title is gonna be… My… Gmail… Follow up.
94 00:25:57.380 ⇒ 00:25:58.140 Uttam Kumaran: Writer.
95 00:25:59.080 ⇒ 00:25:59.650 Casie Aviles: Yeah.
96 00:25:59.760 ⇒ 00:26:07.040 Uttam Kumaran: And… so then the other thing is I’m gonna put in… office…
97 00:26:17.970 ⇒ 00:26:24.309 Uttam Kumaran: So all this is doing is setting up a simple form input, okay. And then…
98 00:26:24.310 ⇒ 00:26:26.460 Casie Aviles: With a farm of lemon.
99 00:26:28.190 ⇒ 00:26:30.739 Uttam Kumaran: Oh, add a form element, okay.
100 00:26:30.930 ⇒ 00:26:35.060 Uttam Kumaran: And then this would say, field name is transcript.
101 00:26:35.310 ⇒ 00:26:39.580 Uttam Kumaran: The element type is text, so say, enter transcript.
102 00:26:40.150 ⇒ 00:26:43.900 Uttam Kumaran: And we’ve got a required field. So this is very similar to a Google form.
103 00:26:44.580 ⇒ 00:26:44.970 Casie Aviles: Yes.
104 00:26:44.970 ⇒ 00:26:48.429 Uttam Kumaran: You can go in and leave that, right?
105 00:26:48.850 ⇒ 00:26:50.809 Uttam Kumaran: And so that gets created up here.
106 00:26:51.940 ⇒ 00:26:55.030 Uttam Kumaran: The next piece we would do is…
107 00:26:55.640 ⇒ 00:26:58.759 Uttam Kumaran: Let’s go ahead and do… this is just, like, text, right?
108 00:26:59.380 ⇒ 00:27:04.240 Casie Aviles: Yeah, that’s just to make it easier to… Kind of compartmentalize.
109 00:27:04.870 ⇒ 00:27:06.910 Casie Aviles: But we can do that all in one.
110 00:27:06.910 ⇒ 00:27:08.110 Uttam Kumaran: Oh, this is just input.
111 00:27:08.850 ⇒ 00:27:09.460 Casie Aviles: Yeah.
112 00:27:09.610 ⇒ 00:27:10.700 Uttam Kumaran: Oh…
113 00:27:12.750 ⇒ 00:27:16.369 Casie Aviles: I’m not sure if it’s much easier to understand it this way, but…
114 00:27:17.540 ⇒ 00:27:19.110 Casie Aviles: Technically, we could have it all in.
115 00:27:19.110 ⇒ 00:27:22.460 Uttam Kumaran: Like, we sh… we should just, like, have this go straight to this.
116 00:27:24.120 ⇒ 00:27:28.280 Casie Aviles: Okay, so we, we don’t need this, this, separating examples.
117 00:27:29.020 ⇒ 00:27:32.100 Uttam Kumaran: So let’s say if I was to just continue without the writing example.
118 00:27:32.760 ⇒ 00:27:38.079 Casie Aviles: The next thing we’re gonna do is… oh, so you’re saying the set writing example…
119 00:27:40.890 ⇒ 00:27:45.029 Casie Aviles: This is where they would ideally… Right? Like, they’re…
120 00:27:45.030 ⇒ 00:27:46.800 Uttam Kumaran: But all this is due… oh…
121 00:27:46.800 ⇒ 00:27:47.330 Casie Aviles: Yeah.
122 00:27:50.610 ⇒ 00:27:52.509 Uttam Kumaran: Oh, I see what he means.
123 00:27:54.980 ⇒ 00:27:59.170 Casie Aviles: But yeah, like, technically they could do it all in just the system prompt.
124 00:27:59.360 ⇒ 00:28:04.249 Casie Aviles: It’s really just… A variable that I’m…
125 00:28:04.250 ⇒ 00:28:06.370 Uttam Kumaran: Okay, okay, okay.
126 00:28:06.670 ⇒ 00:28:09.769 Uttam Kumaran: Well, oh, this is the writing example.
127 00:28:10.280 ⇒ 00:28:10.880 Casie Aviles: Yeah.
128 00:28:11.020 ⇒ 00:28:14.729 Uttam Kumaran: But the writing example right now comes from…
129 00:28:15.770 ⇒ 00:28:18.700 Uttam Kumaran: Oh, this is, like, an example email that I sent.
130 00:28:20.200 ⇒ 00:28:23.850 Casie Aviles: Oh, no, these are all AI-generated as well, for now.
131 00:28:25.470 ⇒ 00:28:26.730 Uttam Kumaran: I see, okay.
132 00:28:26.950 ⇒ 00:28:29.520 Uttam Kumaran: So what I’m gonna do here is…
133 00:28:29.720 ⇒ 00:28:36.230 Uttam Kumaran: I’m also gonna put in here… like… sample emails.
134 00:28:41.330 ⇒ 00:28:52.509 Casie Aviles: Here are… yeah, go ahead. There are these types, but earlier we had, like, the labels, that’s why I added this type thing, but… I mean, if we’re not gonna do labels…
135 00:28:52.740 ⇒ 00:28:55.779 Casie Aviles: Yeah, that’s important, yeah.
136 00:28:58.860 ⇒ 00:29:03.310 Uttam Kumaran: Okay, I’ll leave them, that’s fine. So then I’ll walk them through, in this next step.
137 00:29:03.600 ⇒ 00:29:11.230 Uttam Kumaran: We’re going to also give the AI some examples of emails, follow-up emails, we liked.
138 00:29:11.380 ⇒ 00:29:18.650 Uttam Kumaran: And so, here on the left, I have some examples of emails that I’ve sent before, so the AI can use as guidance.
139 00:29:18.770 ⇒ 00:29:22.190 Uttam Kumaran: And so we’re gonna go in here, type in input.
140 00:29:23.160 ⇒ 00:29:26.169 Uttam Kumaran: are… what does it remember? .
141 00:29:26.480 ⇒ 00:29:28.940 Casie Aviles: Edit, like, yeah, type and edit.
142 00:29:30.420 ⇒ 00:29:31.850 Casie Aviles: Edit here. Oh, okay.
143 00:29:32.860 ⇒ 00:29:37.359 Uttam Kumaran: We’re gonna go to Edit Fields. What this is gonna do is just basically create
144 00:29:37.850 ⇒ 00:29:41.909 Uttam Kumaran: Oh, you’re saying just shove this into this prompt.
145 00:29:42.880 ⇒ 00:29:44.229 Casie Aviles: Yeah, yeah, that’s what it’s doing.
146 00:29:44.250 ⇒ 00:29:48.329 Uttam Kumaran: Yeah, maybe that might be best, to be honest.
147 00:29:51.020 ⇒ 00:29:51.770 Uttam Kumaran: Yeah.
148 00:29:52.240 ⇒ 00:29:53.740 Uttam Kumaran: I might just do that, dude.
149 00:29:54.200 ⇒ 00:29:54.850 Casie Aviles: Okay.
150 00:29:55.050 ⇒ 00:29:56.680 Uttam Kumaran: I think they’re kinda confused.
151 00:29:59.460 ⇒ 00:30:07.699 Uttam Kumaran: Okay, so we go back, basically say, okay, we have this form that’s gonna take a transcript in. In the next step, we’re gonna set up our prompt.
152 00:30:07.980 ⇒ 00:30:11.720 Uttam Kumaran: I’ll walk people through, like, basically.
153 00:30:12.070 ⇒ 00:30:28.260 Uttam Kumaran: what a… how to structure a great prompt. You know, you have to first, at the top, set, like, the objective, tell the AI, like, what it is, and then give it sort of structured information. So, for example, we’ve given it examples of our emails.
154 00:30:28.440 ⇒ 00:30:33.129 Uttam Kumaran: We’ve then given it Our place for our meeting transcript.
155 00:30:33.350 ⇒ 00:30:38.579 Uttam Kumaran: This is gonna end up coming from our form submission. That’s why it looks like a variable.
156 00:30:38.940 ⇒ 00:30:42.619 Uttam Kumaran: We’re gonna give it some rules on how to handle transcripts.
157 00:30:42.890 ⇒ 00:30:45.880 Uttam Kumaran: How to handle names.
158 00:30:46.120 ⇒ 00:30:57.809 Uttam Kumaran: Things like that, so depending if you’re coming from Granola or Zoom, it’ll adapt. I’ve also listed… these are people on our team, and so you can also list, like, hey, if these people are mentioned, it’s context for you.
159 00:30:58.190 ⇒ 00:31:04.729 Uttam Kumaran: We’ve listed in here also information about what we do here at Brainforge.
160 00:31:04.840 ⇒ 00:31:09.690 Uttam Kumaran: And we’ve given it helpful instructions on how to identify the client.
161 00:31:10.230 ⇒ 00:31:16.119 Uttam Kumaran: the types of emails to identify, so I… I can provide them with this prompt, basically, and I’ll share, like.
162 00:31:16.750 ⇒ 00:31:20.599 Uttam Kumaran: This is something that… this is really the logic to iterate on.
163 00:31:20.870 ⇒ 00:31:27.920 Uttam Kumaran: And so what we’re gonna do is then go ahead and create our Our ability to actually,
164 00:31:28.520 ⇒ 00:31:34.120 Uttam Kumaran: create this, like, LLM prompt, so I’m gonna type in LLM… chain.
165 00:31:34.770 ⇒ 00:31:35.470 Casie Aviles: Yes.
166 00:31:35.770 ⇒ 00:31:40.820 Uttam Kumaran: And… in here… All I’m gonna do is… I don’t know.
167 00:31:40.820 ⇒ 00:31:41.939 Casie Aviles: I think we have to…
168 00:31:41.940 ⇒ 00:31:43.739 Uttam Kumaran: So we have to tie it together, right?
169 00:31:44.500 ⇒ 00:31:48.289 Casie Aviles: Yeah, so… So I’m gonna go in here, put my…
170 00:31:49.070 ⇒ 00:31:50.240 Uttam Kumaran: Prompton.
171 00:31:53.930 ⇒ 00:32:01.460 Uttam Kumaran: And then I’m gonna go ahead and exit out. As you can see, it got put up… Oh, what?
172 00:32:02.290 ⇒ 00:32:03.650 Casie Aviles: I got connected.
173 00:32:04.850 ⇒ 00:32:10.360 Uttam Kumaran: Oh, because… Oh, this is on Form Submission 1. Okay.
174 00:32:10.820 ⇒ 00:32:11.670 Casie Aviles: Yeah.
175 00:32:14.590 ⇒ 00:32:18.760 Uttam Kumaran: Oh, so it’s gonna auto-connect. Okay, so on forms, emotional streamline…
176 00:32:20.990 ⇒ 00:32:23.940 Casie Aviles: And, yeah, we should delete the connection.
177 00:32:26.120 ⇒ 00:32:33.650 Uttam Kumaran: And I’m gonna go ahead and put this here. This is gonna connect to this one. When we go in here, what we’re gonna wanna see here is that
178 00:32:33.800 ⇒ 00:32:38.740 Uttam Kumaran: We have this, which is on Form Submission 1.
179 00:32:39.770 ⇒ 00:32:40.480 Uttam Kumaran: Right?
180 00:32:41.050 ⇒ 00:32:41.730 Casie Aviles: Yes.
181 00:32:42.860 ⇒ 00:32:48.620 Uttam Kumaran: Is that, Oh, okay, this isn’t right.
182 00:32:49.480 ⇒ 00:32:53.370 Casie Aviles: It’s… it’s grayed out, because it’s… it hasn’t run yet, but I…
183 00:32:53.370 ⇒ 00:32:53.830 Uttam Kumaran: I see.
184 00:32:53.830 ⇒ 00:32:54.849 Casie Aviles: It’s not red.
185 00:32:56.630 ⇒ 00:33:00.840 Uttam Kumaran: Great. And then, for this LLM node.
186 00:33:00.940 ⇒ 00:33:07.690 Uttam Kumaran: you’re gonna want to connect, your chat model, and so for this exercise, we’re gonna be using AI,
187 00:33:08.910 ⇒ 00:33:13.270 Uttam Kumaran: The next, we’re gonna be using the latest BibuPoo 5 today.
188 00:33:14.320 ⇒ 00:33:21.899 Uttam Kumaran: So this one should be fine, right? So I’m gonna go use… er… Alright, we continue.
189 00:33:26.700 ⇒ 00:33:27.390 Casie Aviles: Shouldn’t.
190 00:33:28.240 ⇒ 00:33:34.420 Uttam Kumaran: I searched in the GPD5… And then…
191 00:33:36.720 ⇒ 00:33:42.389 Uttam Kumaran: The other piece we’re gonna want it to do is, for the draft, we just… well…
192 00:33:45.550 ⇒ 00:33:54.499 Uttam Kumaran: I see, okay, so for the draft, we’re gonna want this to give us a structured output of the email body and an email subject, and so…
193 00:33:54.930 ⇒ 00:33:56.579 Casie Aviles: When I go to structure.
194 00:33:56.760 ⇒ 00:33:57.350 Uttam Kumaran: Output…
195 00:33:58.220 ⇒ 00:34:03.949 Casie Aviles: Oh, wait, sorry, yeah, yeah, that’s right, that’s right. I, I, I forgot, we just forgot to set the…
196 00:34:04.630 ⇒ 00:34:07.710 Casie Aviles: something in the basic LLM chain.
197 00:34:08.460 ⇒ 00:34:09.829 Uttam Kumaran: I’ll just say.
198 00:34:09.830 ⇒ 00:34:13.159 Casie Aviles: Yeah, that one. You have to toggle that.
199 00:34:15.159 ⇒ 00:34:16.339 Uttam Kumaran: I see, okay.
200 00:34:16.789 ⇒ 00:34:22.849 Uttam Kumaran: So, I’m gonna delete this, because I realized, okay, so I can go in here… And for this…
201 00:34:23.049 ⇒ 00:34:29.719 Uttam Kumaran: So, a couple pieces, I’m gonna show you the form… And, ad… We have the…
202 00:34:30.039 ⇒ 00:34:32.629 Uttam Kumaran: Great. Along the chain.
203 00:34:33.249 ⇒ 00:34:46.249 Uttam Kumaran: Prompt to LN… chain. We then have the… Configure… LLM chain for structured… Output.
204 00:34:47.869 ⇒ 00:34:49.969 Uttam Kumaran: So I’m gonna click Output Parser.
205 00:34:50.649 ⇒ 00:34:52.659 Uttam Kumaran: structure it out to the parser.
206 00:34:53.019 ⇒ 00:34:55.709 Uttam Kumaran: And in this, we’re gonna put…
207 00:34:55.899 ⇒ 00:34:57.859 Uttam Kumaran: Our structured output, which is here.
208 00:34:58.039 ⇒ 00:35:03.659 Uttam Kumaran: Which basically gives an example to the AI of how to… Generate the output.
209 00:35:04.149 ⇒ 00:35:07.209 Uttam Kumaran: And so we don’t actually need to type, right, anymore.
210 00:35:10.150 ⇒ 00:35:12.119 Casie Aviles: Yeah, we don’t really need it anymore.
211 00:35:14.250 ⇒ 00:35:23.060 Uttam Kumaran: I’m gonna give it this structured output. All this is doing is saying, from the AI, we request two outputs, the subject and the draft.
212 00:35:24.050 ⇒ 00:35:25.870 Casie Aviles: Forgot to add the subject here.
213 00:35:25.870 ⇒ 00:35:43.210 Uttam Kumaran: Yeah, for subject, I’m gonna be… I’m gonna do… Brain Forge… Bbc Company, Thank you, and my goodness.
214 00:35:43.590 ⇒ 00:35:51.940 Uttam Kumaran: And the subject… I’m gonna get out of this, and then finally, we’re… we’re gonna also add here our…
215 00:35:54.960 ⇒ 00:35:56.310 Uttam Kumaran: AI, right?
216 00:36:00.530 ⇒ 00:36:01.210 Casie Aviles: Is it…
217 00:36:01.380 ⇒ 00:36:02.869 Uttam Kumaran: Oh, this is… yeah, this isn’t easy.
218 00:36:02.870 ⇒ 00:36:04.060 Casie Aviles: Different nodes.
219 00:36:06.730 ⇒ 00:36:12.940 Casie Aviles: I don’t know what that unit is, I haven’t tried that one. Let me…
220 00:36:12.940 ⇒ 00:36:14.119 Uttam Kumaran: You think it’s gonna work?
221 00:36:15.040 ⇒ 00:36:17.010 Casie Aviles: I mean, we could try, we could try.
222 00:36:18.270 ⇒ 00:36:20.060 Uttam Kumaran: What is it that instructor?
223 00:36:21.630 ⇒ 00:36:22.670 Uttam Kumaran: There’s only one.
224 00:36:23.680 ⇒ 00:36:25.880 Casie Aviles: I think we have to select it from…
225 00:36:26.290 ⇒ 00:36:32.010 Casie Aviles: The, look, can you try disconnecting that for now, like, the output parser, and then…
226 00:36:33.130 ⇒ 00:36:36.920 Casie Aviles: Structured output parser… it’s the same logo.
227 00:36:37.900 ⇒ 00:36:38.580 Uttam Kumaran: Yeah.
228 00:36:40.180 ⇒ 00:36:40.850 Casie Aviles: What?
229 00:36:42.680 ⇒ 00:36:44.159 Casie Aviles: I don’t know, that’s weird.
230 00:36:46.020 ⇒ 00:36:46.780 Casie Aviles: That’s just…
231 00:36:46.780 ⇒ 00:36:47.649 Uttam Kumaran: Okay, let’s try it.
232 00:36:47.660 ⇒ 00:36:49.190 Casie Aviles: Yeah, that’s just right.
233 00:36:49.470 ⇒ 00:36:54.179 Uttam Kumaran: And then last, we’re gonna want to get our epicented email.
234 00:36:54.350 ⇒ 00:36:59.869 Uttam Kumaran: So I’m gonna click on Gmail, and then I’m gonna type in draft, and I’m gonna say create a draft.
235 00:37:00.790 ⇒ 00:37:03.919 Uttam Kumaran: For the draft, we want to…
236 00:37:04.080 ⇒ 00:37:10.700 Uttam Kumaran: have our resource say draft, we’re creating the subject here, we want this to be…
237 00:37:11.060 ⇒ 00:37:14.850 Uttam Kumaran: Oh, execute previous node to use input data.
238 00:37:15.160 ⇒ 00:37:16.210 Uttam Kumaran: So…
239 00:37:17.940 ⇒ 00:37:19.530 Casie Aviles: Okay, let’s see…
240 00:37:22.120 ⇒ 00:37:24.359 Uttam Kumaran: Well, I’d probably need to execute this when we’ll start, right?
241 00:37:25.140 ⇒ 00:37:27.590 Casie Aviles: Oh, yeah, we… it doesn’t have any data.
242 00:37:33.640 ⇒ 00:37:36.200 Casie Aviles: Why are there outstanding issues here?
243 00:37:42.220 ⇒ 00:37:44.460 Casie Aviles: I think it’s the Gmail node.
244 00:37:45.270 ⇒ 00:37:47.250 Casie Aviles: It’s the only one with the warning.
245 00:37:47.400 ⇒ 00:37:48.220 Casie Aviles: Simple.
246 00:37:48.220 ⇒ 00:37:48.870 Uttam Kumaran: Yeah.
247 00:37:49.990 ⇒ 00:37:50.940 Uttam Kumaran: So…
248 00:37:51.740 ⇒ 00:37:56.099 Uttam Kumaran: So what I’ll actually do at the start here is I’m gonna say, great, I’m gonna pull this up.
249 00:37:57.010 ⇒ 00:37:58.650 Uttam Kumaran: And I’m gonna enter it here.
250 00:37:59.310 ⇒ 00:38:08.799 Uttam Kumaran: I’m gonna say… I’m gonna take a sample transcript from granola, and… I’m gonna take in…
251 00:38:09.200 ⇒ 00:38:14.225 Uttam Kumaran: A call from earlier, let’s just say…
252 00:38:25.590 ⇒ 00:38:26.370 Uttam Kumaran: Yes.
253 00:38:26.740 ⇒ 00:38:33.890 Uttam Kumaran: It’s gonna take this call, I’m not aware.
254 00:38:36.080 ⇒ 00:38:39.049 Uttam Kumaran: I’m gonna place it here. I’m gonna submit it.
255 00:38:41.580 ⇒ 00:38:46.020 Uttam Kumaran: What we’re gonna see… is that… we should see R…
256 00:38:47.270 ⇒ 00:38:47.979 Casie Aviles: I think I have to…
257 00:38:47.980 ⇒ 00:38:48.980 Uttam Kumaran: I got you here.
258 00:38:51.110 ⇒ 00:38:53.589 Casie Aviles: We have to check the executions. Can…
259 00:38:54.550 ⇒ 00:38:57.449 Casie Aviles: Yeah, in the middle. Top middle.
260 00:38:57.570 ⇒ 00:38:59.040 Casie Aviles: Executions.
261 00:38:59.550 ⇒ 00:39:01.110 Uttam Kumaran: Oh, yeah, gone.
262 00:39:01.800 ⇒ 00:39:03.989 Casie Aviles: It should be below… yeah, that one.
263 00:39:07.070 ⇒ 00:39:08.390 Casie Aviles: Yeah, that should be it.
264 00:39:08.640 ⇒ 00:39:09.410 Casie Aviles: And then…
265 00:39:12.530 ⇒ 00:39:14.420 Uttam Kumaran: So in this, I should see…
266 00:39:20.050 ⇒ 00:39:24.320 Casie Aviles: Let me go check the execution. Okay, it did flow.
267 00:39:27.380 ⇒ 00:39:31.990 Casie Aviles: Oh, okay, I see the issue. We forgot something in the LLM chain.
268 00:39:33.760 ⇒ 00:39:35.610 Casie Aviles: But I guess so.
269 00:39:36.660 ⇒ 00:39:37.680 Uttam Kumaran: I’m glad.
270 00:39:38.120 ⇒ 00:39:41.270 Casie Aviles: We’ll have to replace the… Prompt.
271 00:39:41.610 ⇒ 00:39:42.970 Casie Aviles: source for prompt.
272 00:39:44.560 ⇒ 00:39:45.470 Uttam Kumaran: I saw?
273 00:39:46.350 ⇒ 00:39:48.580 Casie Aviles: No, exit.
274 00:39:48.750 ⇒ 00:39:54.380 Casie Aviles: From that. Connected chat trigger node, can you see that one? We have to replace that with define below.
275 00:39:59.700 ⇒ 00:40:02.560 Casie Aviles: And it can be empty for now, since…
276 00:40:03.000 ⇒ 00:40:04.909 Casie Aviles: We’re not really chatting with it.
277 00:40:05.750 ⇒ 00:40:08.109 Casie Aviles: It’s all in the system prompt anyway.
278 00:40:08.930 ⇒ 00:40:10.100 Uttam Kumaran: But it says required.
279 00:40:10.940 ⇒ 00:40:12.249 Uttam Kumaran: Just put a space.
280 00:40:13.450 ⇒ 00:40:20.630 Casie Aviles: Yeah, that’s fine. Currently, the LLM chain that I have is empty, though, but it’s working.
281 00:40:23.520 ⇒ 00:40:24.200 Uttam Kumaran: Okay.
282 00:40:24.340 ⇒ 00:40:26.290 Uttam Kumaran: But let’s say if I was to try this again.
283 00:40:33.880 ⇒ 00:40:35.099 Casie Aviles: Did you save it?
284 00:40:37.000 ⇒ 00:40:38.480 Casie Aviles: Were you able to save it?
285 00:40:39.080 ⇒ 00:40:39.680 Uttam Kumaran: Yeah.
286 00:40:40.050 ⇒ 00:40:40.710 Casie Aviles: Okay.
287 00:40:47.110 ⇒ 00:40:48.110 Casie Aviles: Okay…
288 00:40:48.110 ⇒ 00:40:50.509 Uttam Kumaran: Oh, you think it’s a GPT-5, it’s taking a while?
289 00:40:53.400 ⇒ 00:40:55.239 Casie Aviles: Yeah, that might be the case.
290 00:40:57.770 ⇒ 00:40:59.279 Uttam Kumaran: I can switch it to minis.
291 00:41:04.480 ⇒ 00:41:06.129 Casie Aviles: It’s… it’s good now.
292 00:41:06.590 ⇒ 00:41:07.530 Casie Aviles: Holy…
293 00:41:11.680 ⇒ 00:41:16.510 Uttam Kumaran: Oh, we ran this one. Oh… Yeah, I know.
294 00:41:23.010 ⇒ 00:41:25.210 Casie Aviles: It’s a little confusing.
295 00:41:26.160 ⇒ 00:41:29.690 Casie Aviles: It’s… you have to use, like… the, the tests.
296 00:41:30.850 ⇒ 00:41:32.410 Casie Aviles: the test version.
297 00:41:33.380 ⇒ 00:41:40.189 Casie Aviles: But if it’s easier, then we could just… I’m just gonna deactivate the…
298 00:41:42.970 ⇒ 00:41:44.489 Casie Aviles: Let me go and deactivate that.
299 00:41:44.490 ⇒ 00:41:45.060 Uttam Kumaran: costs.
300 00:41:45.630 ⇒ 00:41:46.879 Casie Aviles: the old workflow.
301 00:41:57.830 ⇒ 00:41:58.819 Casie Aviles: And then…
302 00:41:59.260 ⇒ 00:42:04.410 Uttam Kumaran: So this is… so the problem is, is like, oh, I can set mock data, okay?
303 00:42:05.100 ⇒ 00:42:05.610 Casie Aviles: Yeah.
304 00:42:10.670 ⇒ 00:42:11.760 Uttam Kumaran: Oh…
305 00:42:14.920 ⇒ 00:42:18.399 Casie Aviles: I don’t really use that a lot,
306 00:42:20.560 ⇒ 00:42:28.189 Casie Aviles: what I do is I just activate the form, like, I mean, the workflow, and then just use the production URL.
307 00:42:29.380 ⇒ 00:42:30.260 Casie Aviles: For me, it’s fast.
308 00:42:30.260 ⇒ 00:42:35.310 Uttam Kumaran: For this one, like, how can I… how can I get… specify the exact… like, how can I get the…
309 00:42:35.750 ⇒ 00:42:38.430 Uttam Kumaran: Execute previous nodes to use input data.
310 00:42:40.650 ⇒ 00:42:47.540 Casie Aviles: We could add, like, the variable, so it would be something like…
311 00:42:48.930 ⇒ 00:42:51.650 Uttam Kumaran: But you have to type it in perfectly, it’s not gonna, like, do it.
312 00:42:52.600 ⇒ 00:42:55.829 Casie Aviles: Yeah, you have to type it in, like, this one.
313 00:42:56.060 ⇒ 00:42:58.140 Casie Aviles: I’m going to send ones over.
314 00:42:59.810 ⇒ 00:43:06.389 Casie Aviles: For example, we need to do this… it this way, and then for the body.
315 00:43:19.110 ⇒ 00:43:21.919 Casie Aviles: Yeah, it should be on… the Zoom chat.
316 00:43:43.670 ⇒ 00:43:45.420 Casie Aviles: Sorry, were you able to get it?
317 00:43:48.620 ⇒ 00:43:49.560 Uttam Kumaran: I’m sorry.
318 00:43:57.440 ⇒ 00:44:00.970 Casie Aviles: Yeah, I sent something to Zoom chat.
319 00:44:00.970 ⇒ 00:44:07.320 Uttam Kumaran: Oh, but we can’t… we can’t, we can’t do it dynamically, because there’s no output data from the previous name.
320 00:44:08.060 ⇒ 00:44:12.350 Casie Aviles: Yeah, that’s… That’s the reason why we can’t do that.
321 00:44:14.690 ⇒ 00:44:15.350 Uttam Kumaran: Here.
322 00:44:16.680 ⇒ 00:44:19.110 Uttam Kumaran: Trying to think about how to explain this.
323 00:44:24.870 ⇒ 00:44:28.050 Casie Aviles: Yeah, that’s one of the confusing things about NA10.
324 00:44:46.830 ⇒ 00:44:47.630 Uttam Kumaran: Okay.
325 00:45:01.370 ⇒ 00:45:02.480 Uttam Kumaran: I guess…
326 00:45:04.520 ⇒ 00:45:08.409 Uttam Kumaran: how do we get around that? Because I feel like they’re really not going to follow from, like.
327 00:45:10.760 ⇒ 00:45:12.560 Uttam Kumaran: Venture write this, you know?
328 00:45:13.050 ⇒ 00:45:14.150 Casie Aviles: Yeah,
329 00:45:15.700 ⇒ 00:45:26.350 Casie Aviles: Well, what I would do is, I would just set the… I would set the workflow to active, and then I would use the production URL in order to test it and get something out.
330 00:45:27.240 ⇒ 00:45:29.110 Casie Aviles: Like, get something flowing.
331 00:45:29.350 ⇒ 00:45:30.730 Casie Aviles: Okay, so…
332 00:45:32.660 ⇒ 00:45:34.520 Uttam Kumaran: Okay, so what I can do is, like.
333 00:45:35.690 ⇒ 00:45:40.089 Uttam Kumaran: at this point, I can… I can… what I can say is, like, I… let’s say we do.
334 00:45:40.090 ⇒ 00:45:40.680 Casie Aviles: Yeah, we can dance.
335 00:45:40.680 ⇒ 00:45:41.020 Uttam Kumaran: near.
336 00:45:41.020 ⇒ 00:45:41.380 Casie Aviles: Right.
337 00:45:41.580 ⇒ 00:45:45.649 Uttam Kumaran: And I want to just say, cool, we got to this point, and I want to just test.
338 00:45:46.150 ⇒ 00:45:46.640 Casie Aviles: Yes.
339 00:45:46.640 ⇒ 00:45:48.480 Uttam Kumaran: what we’re seeing from the AI.
340 00:45:48.660 ⇒ 00:45:52.820 Uttam Kumaran: And so… I would click on this.
341 00:45:52.820 ⇒ 00:45:54.109 Casie Aviles: And then do the production.
342 00:45:54.110 ⇒ 00:45:55.420 Uttam Kumaran: Science Production.
343 00:45:55.420 ⇒ 00:45:55.960 Casie Aviles: Yeah.
344 00:45:57.550 ⇒ 00:45:59.399 Uttam Kumaran: And then let’s go ahead and change this to go.
345 00:46:11.590 ⇒ 00:46:12.640 Uttam Kumaran: Committed.
346 00:46:13.630 ⇒ 00:46:18.240 Casie Aviles: And then you have to go to Execution, it’s not gonna show up in real time.
347 00:46:20.090 ⇒ 00:46:23.180 Uttam Kumaran: Yeah, so let’s go to Execution, so we can just see that.
348 00:46:23.870 ⇒ 00:46:26.130 Uttam Kumaran: Air…
349 00:46:31.930 ⇒ 00:46:33.700 Uttam Kumaran: Oh, because I didn’t save? Okay.
350 00:46:43.810 ⇒ 00:46:47.399 Uttam Kumaran: So let’s go back, let’s say I’m using the production URL…
351 00:46:52.230 ⇒ 00:46:54.960 Uttam Kumaran: And let’s go ahead and test our progress so far.
352 00:46:55.890 ⇒ 00:46:59.300 Uttam Kumaran: I’m gonna go ahead and put in my transcript.
353 00:47:00.370 ⇒ 00:47:01.740 Uttam Kumaran: Select Submit.
354 00:47:02.100 ⇒ 00:47:04.559 Uttam Kumaran: We’re gonna go ahead and go to Executions.
355 00:47:04.750 ⇒ 00:47:08.870 Uttam Kumaran: We’re gonna see… our latest.
356 00:47:09.810 ⇒ 00:47:10.840 Uttam Kumaran: Runnings.
357 00:47:14.080 ⇒ 00:47:15.420 Casie Aviles: We should be…
358 00:47:15.420 ⇒ 00:47:16.770 Uttam Kumaran: It’s gonna be fun.
359 00:47:20.090 ⇒ 00:47:25.729 Casie Aviles: Sometimes it takes a long… Like, it’s already done, but it still says it’s running.
360 00:47:29.690 ⇒ 00:47:30.979 Casie Aviles: So you’re good.
361 00:47:32.220 ⇒ 00:47:34.439 Casie Aviles: Yeah, on my end, it’s already done.
362 00:47:36.580 ⇒ 00:47:37.690 Casie Aviles: Oh, weird.
363 00:47:41.980 ⇒ 00:47:42.579 Uttam Kumaran: You know what?
364 00:47:49.820 ⇒ 00:47:50.610 Casie Aviles: Right.
365 00:47:51.260 ⇒ 00:47:52.619 Casie Aviles: Let me give you the link.
366 00:47:59.470 ⇒ 00:48:01.640 Casie Aviles: Yeah, I sent the link now.
367 00:48:17.220 ⇒ 00:48:21.690 Uttam Kumaran: I don’t think this is just so buggy, I don’t guess it’s awesome, this didn’t… still didn’t work.
368 00:48:25.910 ⇒ 00:48:26.430 Uttam Kumaran: Right?
369 00:48:26.430 ⇒ 00:48:26.980 Casie Aviles: Boom.
370 00:48:29.280 ⇒ 00:48:31.869 Casie Aviles: Anita may have run out of memory.
371 00:48:33.090 ⇒ 00:48:33.770 Casie Aviles: phenomenal.
372 00:48:39.450 ⇒ 00:48:42.479 Uttam Kumaran: Oh, okay, so maybe I just shouldn’t use the PPT4.
373 00:48:47.650 ⇒ 00:48:50.119 Uttam Kumaran: What is the memory? Like, what does that even mean?
374 00:48:55.460 ⇒ 00:48:59.370 Casie Aviles: Yeah, I haven’t been using the OpenAI node, so…
375 00:48:59.790 ⇒ 00:49:01.110 Uttam Kumaran: Sure, both down.
376 00:49:02.600 ⇒ 00:49:08.000 Uttam Kumaran: Okay, what I’m gonna do is just to… GPT5 Mini.
377 00:49:13.580 ⇒ 00:49:15.190 Uttam Kumaran: I’m gonna save this.
378 00:49:15.820 ⇒ 00:49:18.570 Uttam Kumaran: I’m gonna try to go again, reproduction app.
379 00:49:42.850 ⇒ 00:49:44.499 Casie Aviles: Okay, let me check…
380 00:49:49.520 ⇒ 00:49:50.920 Uttam Kumaran: in an application.
381 00:49:51.250 ⇒ 00:49:57.540 Casie Aviles: We can change the… the text entry, I think it’s just one line.
382 00:49:58.410 ⇒ 00:49:59.240 Casie Aviles: Oh, okay.
383 00:49:59.240 ⇒ 00:49:59.800 Uttam Kumaran: Okay.
384 00:50:00.830 ⇒ 00:50:02.020 Uttam Kumaran: Succeeded.
385 00:50:02.480 ⇒ 00:50:03.920 Casie Aviles: Yes, and so…
386 00:50:04.010 ⇒ 00:50:09.030 Uttam Kumaran: What we should see is that our nodes here should be green.
387 00:50:09.190 ⇒ 00:50:20.910 Uttam Kumaran: And, let’s say we go back to editor, And we’re gonna see here… Still not seeing… any data.
388 00:50:21.720 ⇒ 00:50:22.910 Casie Aviles: Let me double check.
389 00:50:24.300 ⇒ 00:50:26.500 Uttam Kumaran: I mean, you know what I should do? Like, I should just…
390 00:50:27.460 ⇒ 00:50:29.879 Uttam Kumaran: I should just put this in here, right?
391 00:50:30.280 ⇒ 00:50:32.250 Casie Aviles: You could, you could pin the…
392 00:50:32.510 ⇒ 00:50:37.050 Casie Aviles: the output. So you… can you go back to executions?
393 00:50:39.300 ⇒ 00:50:41.759 Casie Aviles: Yeah, that one, then just double-click.
394 00:50:42.300 ⇒ 00:50:44.660 Casie Aviles: And then I click the copy to editor.
395 00:50:45.260 ⇒ 00:50:47.190 Casie Aviles: Yeah, that one.
396 00:50:47.840 ⇒ 00:50:49.029 Uttam Kumaran: And what does this do?
397 00:50:49.400 ⇒ 00:50:54.200 Casie Aviles: It saves, like, the output that went through, so you can edit.
398 00:50:54.200 ⇒ 00:50:59.400 Uttam Kumaran: Is it… is it more helpful at this step to just paste in the mock data here?
399 00:51:00.820 ⇒ 00:51:02.919 Uttam Kumaran: Like, facing the transcript like this.
400 00:51:05.150 ⇒ 00:51:09.609 Casie Aviles: I, that probably doesn’t work that way, what?
401 00:51:09.740 ⇒ 00:51:11.799 Casie Aviles: Looks like you have to structure it.
402 00:51:12.030 ⇒ 00:51:15.190 Casie Aviles: But I haven’t really used this, because I just used the law.
403 00:51:15.580 ⇒ 00:51:17.430 Casie Aviles: Like, I just pinned the output.
404 00:51:22.910 ⇒ 00:51:25.679 Uttam Kumaran: Okay, so, like, let’s say we were to… well…
405 00:51:33.960 ⇒ 00:51:34.970 Casie Aviles: Hmm, let’s see…
406 00:51:36.790 ⇒ 00:51:38.460 Uttam Kumaran: Oh, so I would just do, like…
407 00:51:40.470 ⇒ 00:51:41.830 Casie Aviles: You without the pace.
408 00:51:42.330 ⇒ 00:51:43.559 Casie Aviles: This is a lot.
409 00:51:43.810 ⇒ 00:51:46.820 Uttam Kumaran: Oh, okay, so, like, okay, so let’s say I go to Executions…
410 00:51:47.120 ⇒ 00:51:51.130 Uttam Kumaran: We see that we had a successful institution, I want to copy this to our editor.
411 00:51:51.280 ⇒ 00:51:51.880 Casie Aviles: Yeah, and…
412 00:51:51.880 ⇒ 00:51:57.390 Uttam Kumaran: So what we should now see here is that you can see an example of our transcript that came in.
413 00:51:58.100 ⇒ 00:52:03.409 Uttam Kumaran: And then you’re gonna go into our LLM. We’re gonna see, okay, here’s the transcript.
414 00:52:03.680 ⇒ 00:52:12.640 Uttam Kumaran: Here is the structured output, and then now, finally, We want to add our… Email nodes…
415 00:52:12.950 ⇒ 00:52:18.180 Uttam Kumaran: We’re gonna see online… Let’s go ahead and add our Gmail.
416 00:52:18.580 ⇒ 00:52:22.869 Uttam Kumaran: Graph node… For creating a draft.
417 00:52:23.700 ⇒ 00:52:25.790 Casie Aviles: Yeah, and then you can drag and drop now.
418 00:52:26.710 ⇒ 00:52:32.270 Uttam Kumaran: And so what I’m gonna do is I’m gonna say, great, this is our subject, And…
419 00:52:34.050 ⇒ 00:52:36.959 Uttam Kumaran: And this is our message.
420 00:52:37.520 ⇒ 00:52:38.040 Casie Aviles: Yeah.
421 00:52:41.220 ⇒ 00:52:45.920 Uttam Kumaran: And… perfect. And then what we’re gonna do here is we’re gonna execute this step.
422 00:52:47.730 ⇒ 00:52:54.160 Uttam Kumaran: And what I’d like to do at this point is go back to my… Gmail…
423 00:52:58.110 ⇒ 00:53:00.279 Uttam Kumaran: I’ll already have it open to draft.
424 00:53:01.720 ⇒ 00:53:07.070 Uttam Kumaran: And there it is.
425 00:53:07.070 ⇒ 00:53:07.930 Casie Aviles: Yes.
426 00:53:12.090 ⇒ 00:53:17.710 Uttam Kumaran: And so, one thing that I want to do is… can we, it’s not getting this, like…
427 00:53:18.160 ⇒ 00:53:19.290 Uttam Kumaran: New line?
428 00:53:21.940 ⇒ 00:53:24.190 Uttam Kumaran: So, should I just put spaces in here?
429 00:53:26.590 ⇒ 00:53:28.019 Casie Aviles: Let’s see,
430 00:53:33.840 ⇒ 00:53:37.600 Casie Aviles: I don’t think spaces would work either.
431 00:53:40.390 ⇒ 00:53:46.169 Casie Aviles: Yeah, that’s kind of the tricky part, Pedar Nathan. I, I… I do use, like…
432 00:53:46.720 ⇒ 00:53:49.369 Casie Aviles: Or maybe you could do ENTER, but…
433 00:53:52.910 ⇒ 00:53:57.980 Uttam Kumaran: And then, well, I guess another thing is, like, we don’t have, like, subject examples.
434 00:53:59.060 ⇒ 00:53:59.840 Uttam Kumaran: Right?
435 00:54:00.010 ⇒ 00:54:01.740 Casie Aviles: Yeah, we don’t have one right now.
436 00:54:02.400 ⇒ 00:54:04.960 Uttam Kumaran: So, should I put the subject examples here?
437 00:54:08.070 ⇒ 00:54:11.459 Casie Aviles: You, yeah, you could put subjects here, and then…
438 00:54:12.530 ⇒ 00:54:15.590 Casie Aviles: Additionally, in the output parser, but…
439 00:54:17.760 ⇒ 00:54:19.549 Casie Aviles: Yeah, yeah, you can do that.
440 00:54:43.580 ⇒ 00:54:45.319 Uttam Kumaran: Okay, so…
441 00:54:48.080 ⇒ 00:54:53.510 Uttam Kumaran: And then I’ll go through the whole thing. So let’s… okay, so we ironed out a couple different things. Let’s try to do…
442 00:54:53.700 ⇒ 00:54:55.460 Uttam Kumaran: An entire dry run.
443 00:54:55.990 ⇒ 00:55:00.849 Uttam Kumaran: So now, this turns into our production version, right?
444 00:55:01.780 ⇒ 00:55:02.860 Uttam Kumaran: Yes.
445 00:55:03.970 ⇒ 00:55:11.170 Uttam Kumaran: So… Can I go ahead and, delete this one?
446 00:55:13.140 ⇒ 00:55:14.300 Casie Aviles: Yeah, sure.
447 00:55:15.720 ⇒ 00:55:20.290 Uttam Kumaran: So, I’m gonna delete this one. Block, I’ll leave it, and then let’s just go ahead and we’re gonna create…
448 00:55:20.410 ⇒ 00:55:22.619 Uttam Kumaran: Let’s say this is actually the talking, right?
449 00:55:23.050 ⇒ 00:55:27.440 Uttam Kumaran: So… Let’s kind of go ahead and elevate that.
450 00:55:44.600 ⇒ 00:55:45.820 Uttam Kumaran: But, action.
451 00:55:50.670 ⇒ 00:55:51.999 Casie Aviles: It’s not working.
452 00:55:56.870 ⇒ 00:55:58.489 Uttam Kumaran: I don’t mind.
453 00:55:58.780 ⇒ 00:56:00.960 Uttam Kumaran: Am I, like, copy-pasting it really quick.
454 00:56:07.600 ⇒ 00:56:08.800 Uttam Kumaran: I don’t know what?
455 00:56:12.620 ⇒ 00:56:14.969 Uttam Kumaran: And I want to say, that’s what I’m sure.
456 00:56:24.960 ⇒ 00:56:28.929 Uttam Kumaran: Okay, so we have our Gmail follow-up writer up here.
457 00:56:29.380 ⇒ 00:56:41.440 Uttam Kumaran: And so… Let’s say we ran through this union. Store… That’s great.
458 00:56:41.570 ⇒ 00:56:43.939 Uttam Kumaran: Let me know if there’s going on.
459 00:56:58.790 ⇒ 00:57:02.330 Uttam Kumaran: Okay, and actually, this one, so let me just do, like, a…
460 00:57:11.260 ⇒ 00:57:12.860 Uttam Kumaran: You know, there’s also…
461 00:57:40.110 ⇒ 00:57:40.880 Uttam Kumaran: Things.
462 00:57:49.150 ⇒ 00:57:49.940 Uttam Kumaran: Okay.
463 00:57:50.060 ⇒ 00:57:52.390 Uttam Kumaran: So I want to have kind of walnuts here.
464 00:57:54.230 ⇒ 00:58:03.879 Uttam Kumaran: And then wanna have… So in this, I want to have my Gmail up open.
465 00:58:04.450 ⇒ 00:58:08.099 Uttam Kumaran: I want to have fear, and I wanna, control law.
466 00:58:08.960 ⇒ 00:58:10.009 Uttam Kumaran: What’s in here.
467 00:58:15.900 ⇒ 00:58:18.840 Uttam Kumaran: And I want to just have my graphs open.
468 00:58:20.930 ⇒ 00:58:26.490 Uttam Kumaran: Great. So…
469 00:58:30.690 ⇒ 00:58:32.769 Uttam Kumaran: Please show some of the staff here.
470 00:58:33.630 ⇒ 00:58:34.720 Uttam Kumaran: Fair.
471 00:58:36.400 ⇒ 00:58:41.440 Uttam Kumaran: So… Basically, start with… it won’t be…
472 00:58:45.950 ⇒ 00:58:48.100 Uttam Kumaran: So I’ll have this open…
473 00:58:49.660 ⇒ 00:58:53.829 Uttam Kumaran: kind of just say, okay, hey, we’re Brainforce, blah blah blah.
474 00:58:54.530 ⇒ 00:58:57.430 Uttam Kumaran: We help people with AI problems, blah blah blah.
475 00:58:57.890 ⇒ 00:59:02.050 Uttam Kumaran: Today, we’re gonna show you how you and your team can
476 00:59:02.380 ⇒ 00:59:09.280 Uttam Kumaran: develop a Gmail follow-up writer based on your…
477 00:59:11.110 ⇒ 00:59:18.319 Uttam Kumaran: writing style in, like, less than half hour. So I want to just demo it for you guys to start.
478 00:59:18.430 ⇒ 00:59:23.620 Uttam Kumaran: So today I just have one of my, granola calls from earlier this week.
479 00:59:23.880 ⇒ 00:59:26.269 Uttam Kumaran: I’m gonna go ahead and copy it.
480 00:59:26.620 ⇒ 00:59:29.190 Uttam Kumaran: I’m gonna have that paste it into the spoon arm.
481 00:59:29.310 ⇒ 00:59:31.560 Uttam Kumaran: And hit, submit.
482 00:59:31.970 ⇒ 00:59:36.370 Uttam Kumaran: What you’re gonna see on the right side is you’re gonna see a new execution pop-up.
483 00:59:36.640 ⇒ 00:59:44.690 Uttam Kumaran: And what this is doing… Is running through our, RNA em flow.
484 00:59:44.870 ⇒ 00:59:54.710 Uttam Kumaran: This flow is taking our transcript, taking our custom prompt that dissects our writing style, and creating a draft in my Gmail.
485 00:59:55.010 ⇒ 00:59:59.170 Uttam Kumaran: What you should see here… I refresh…
486 00:59:59.420 ⇒ 01:00:02.610 Uttam Kumaran: You see one recently completed execution.
487 01:00:02.740 ⇒ 01:00:07.820 Uttam Kumaran: And you’ll see that, it completed this.
488 01:00:08.180 ⇒ 01:00:18.130 Uttam Kumaran: flow, And if we go ahead and go to my drafts… We’re gonna see a…
489 01:00:18.760 ⇒ 01:00:23.269 Uttam Kumaran: an email. And this is the follow-up email that we’re expecting.
490 01:00:23.710 ⇒ 01:00:31.000 Uttam Kumaran: And so today, we’re gonna walk through how to build this. And so… We end up back in…
491 01:00:32.040 ⇒ 01:00:32.900 Uttam Kumaran: Halloween.
492 01:00:33.090 ⇒ 01:00:35.059 Uttam Kumaran: Okay, I got a little kid’s kid.
493 01:00:35.290 ⇒ 01:00:36.430 Uttam Kumaran: No, they’re not.
494 01:00:37.200 ⇒ 01:00:37.910 Uttam Kumaran: Captain.
495 01:00:38.940 ⇒ 01:00:40.200 Uttam Kumaran: I know, God.
496 01:00:42.790 ⇒ 01:00:43.630 Uttam Kumaran: No.
497 01:00:50.340 ⇒ 01:00:51.760 Uttam Kumaran: What am I letting you do?
498 01:00:52.970 ⇒ 01:00:54.010 Uttam Kumaran: Anyone’s get out.
499 01:00:54.460 ⇒ 01:00:55.239 Casie Aviles: Yeah, that one.
500 01:00:57.650 ⇒ 01:00:59.350 Uttam Kumaran: So, let’s go back to editor.
501 01:00:59.820 ⇒ 01:01:03.010 Uttam Kumaran: And let’s get started. And so…
502 01:01:03.460 ⇒ 01:01:07.329 Uttam Kumaran: Kind of, we’re gonna walk through it pretty linearly and test along the way.
503 01:01:07.450 ⇒ 01:01:12.689 Uttam Kumaran: So, I’m gonna start here by clicking the plus sign, and we want to add what’s called a trigger.
504 01:01:12.990 ⇒ 01:01:15.280 Uttam Kumaran: We want to look for that item sworn.
505 01:01:15.660 ⇒ 01:01:19.300 Uttam Kumaran: And we’re gonna say, on a new NHN forming event.
506 01:01:20.530 ⇒ 01:01:22.359 Uttam Kumaran: And what we’re gonna do here…
507 01:01:22.470 ⇒ 01:01:35.320 Uttam Kumaran: is we’re going to say our form title, which is my new Gmail… dash… Writer… And…
508 01:01:36.770 ⇒ 01:01:39.830 Uttam Kumaran: We don’t need to have, form descriptions.
509 01:01:40.290 ⇒ 01:01:44.550 Uttam Kumaran: And we can go ahead and just click out of this.
510 01:01:46.440 ⇒ 01:01:48.540 Uttam Kumaran: And let’s go ahead and just drag it down here.
511 01:01:48.950 ⇒ 01:01:52.280 Uttam Kumaran: Next, we’re gonna add what we call an LLM chain.
512 01:01:52.540 ⇒ 01:02:01.839 Uttam Kumaran: All this is doing is stringing a couple of helpful things together, including our language model, which is OpenAI, as well as our ability to extract structured outputs.
513 01:02:01.970 ⇒ 01:02:05.150 Uttam Kumaran: You’ll see how that’s important in a second.
514 01:02:05.590 ⇒ 01:02:09.410 Uttam Kumaran: And so, we go ahead and hit plus here.
515 01:02:09.720 ⇒ 01:02:11.920 Uttam Kumaran: We’ll go ahead and do LLM.
516 01:02:12.880 ⇒ 01:02:13.770 Uttam Kumaran: chain?
517 01:02:14.860 ⇒ 01:02:15.550 Casie Aviles: Yes.
518 01:02:16.970 ⇒ 01:02:20.030 Uttam Kumaran: And then here, we want to go ahead and say…
519 01:02:20.160 ⇒ 01:02:22.400 Uttam Kumaran: We want to define our prompt.
520 01:02:22.730 ⇒ 01:02:26.329 Uttam Kumaran: Below here, we’ve written this prompt.
521 01:02:26.590 ⇒ 01:02:31.080 Uttam Kumaran: I could talk a little bit about
522 01:02:31.240 ⇒ 01:02:47.270 Uttam Kumaran: what goes into a great prompt, but you want to make sure to follow the ChatGBT guidelines, which indicate to give, you know, your prompt a key understanding of its role, and give it helpful information. In this situation, we’re giving it examples of past videos that I’ve sent.
523 01:02:49.330 ⇒ 01:02:54.050 Uttam Kumaran: And I’m gonna put juncture here whenever we say plus.
524 01:02:55.430 ⇒ 01:02:56.170 Casie Aviles: Yeah.
525 01:02:56.860 ⇒ 01:02:57.380 Uttam Kumaran: Yeah.
526 01:02:57.380 ⇒ 01:02:58.250 Casie Aviles: They fix it.
527 01:02:58.930 ⇒ 01:03:00.309 Uttam Kumaran: That’s what kind of fix the weight.
528 01:03:21.580 ⇒ 01:03:24.440 Uttam Kumaran: And these I can edit and make a little bit lighter.
529 01:03:24.740 ⇒ 01:03:26.879 Uttam Kumaran: I can just repost in general.
530 01:03:27.000 ⇒ 01:03:31.530 Uttam Kumaran: And so, all we have here is examples of email types.
531 01:03:31.880 ⇒ 01:03:35.400 Uttam Kumaran: And so…
532 01:03:35.800 ⇒ 01:03:51.680 Uttam Kumaran: the whole prompt, the only thing to kind of keep in mind is this area, where we’re actually going to be taking the output from our form and adding it to our prompt dynamically. This is using NNN variables, and so what we’re going to do is we’re going to go ahead and select
533 01:03:52.150 ⇒ 01:03:57.140 Uttam Kumaran: this, and so in this situation, this is gonna be on form submit 2, right? So we’re gonna go ahead and copy this.
534 01:03:58.630 ⇒ 01:04:01.959 Uttam Kumaran: Go ahead and paste it into the prompt here.
535 01:04:03.110 ⇒ 01:04:05.860 Uttam Kumaran: And… .
536 01:04:06.210 ⇒ 01:04:09.770 Casie Aviles: Oh, we do have the system prompt as well. This one’s…
537 01:04:09.970 ⇒ 01:04:12.809 Casie Aviles: I mean, they’re both going into the AI, but…
538 01:04:13.450 ⇒ 01:04:14.410 Uttam Kumaran: Does it matter?
539 01:04:14.930 ⇒ 01:04:17.720 Casie Aviles: Not really, but this is the user message.
540 01:04:17.840 ⇒ 01:04:20.819 Casie Aviles: And you can define, like, a system run.
541 01:04:21.370 ⇒ 01:04:22.419 Casie Aviles: Here at the Boston.
542 01:04:22.420 ⇒ 01:04:28.150 Uttam Kumaran: I guess I wouldn’t… I guess let’s… should we just see if it works with just doing this?
543 01:04:28.620 ⇒ 01:04:29.290 Casie Aviles: Yeah, sure, sure.
544 01:04:29.800 ⇒ 01:04:30.500 Uttam Kumaran: Okay.
545 01:04:31.100 ⇒ 01:04:34.530 Uttam Kumaran: And then we’re gonna go ahead and add a model. Today we’re gonna use AI.
546 01:04:35.010 ⇒ 01:04:39.959 Uttam Kumaran: I want to use the latest ChatGPT… 5 minutes.
547 01:04:41.100 ⇒ 01:04:44.280 Uttam Kumaran: LuPrate model, Gonna click on that.
548 01:04:44.590 ⇒ 01:04:45.540 Uttam Kumaran: Great.
549 01:04:46.180 ⇒ 01:04:56.759 Uttam Kumaran: And at this point, we want to do one last thing, which is make sure to require structured output. This is going to allow the AI to actually
550 01:04:57.320 ⇒ 01:05:05.740 Uttam Kumaran: output both a subject and a body, instead of just one blurb of text. So we’re gonna click here and say require specific output formats.
551 01:05:06.040 ⇒ 01:05:08.430 Uttam Kumaran: We’re gonna click on add out to the parser.
552 01:05:08.770 ⇒ 01:05:11.180 Uttam Kumaran: I’m here to click on Structured Output Farser.
553 01:05:11.450 ⇒ 01:05:17.400 Uttam Kumaran: And all we’re gonna do here is provide it with a very clear understanding of how to provide an output.
554 01:05:17.650 ⇒ 01:05:24.519 Uttam Kumaran: And so, I have pre-wrote this, but basically, we want to have one variable that’s sub… one variable that’s draft.
555 01:05:25.310 ⇒ 01:05:27.630 Uttam Kumaran: I’m gonna go ahead and paste this in here.
556 01:05:28.330 ⇒ 01:05:29.430 Uttam Kumaran: And then…
557 01:05:29.600 ⇒ 01:05:41.819 Uttam Kumaran: Great. So at this point, just to make sure everyone’s following, we’ve created a form, and we’ve created a simple language model chain. At this point, let’s go ahead and test what we’re looking at.
558 01:05:41.930 ⇒ 01:05:43.750 Uttam Kumaran: So we’re gonna hit save.
559 01:05:44.780 ⇒ 01:05:49.999 Uttam Kumaran: We are going to go ahead and open our form, and get the production URL.
560 01:05:51.130 ⇒ 01:05:55.109 Uttam Kumaran: I’m gonna go to here, and I’m going to paste it in.
561 01:05:56.100 ⇒ 01:06:00.789 Casie Aviles: And then… I didn’t add. Oh, forget that.
562 01:06:01.400 ⇒ 01:06:03.990 Uttam Kumaran: Input, transcript.
563 01:06:05.010 ⇒ 01:06:08.030 Uttam Kumaran: Oh, it’s not gonna be right, right? What is it called?
564 01:06:08.780 ⇒ 01:06:11.309 Uttam Kumaran: it’s just called transcript, right?
565 01:06:11.310 ⇒ 01:06:11.930 Casie Aviles: Yeah.
566 01:06:12.550 ⇒ 01:06:20.580 Casie Aviles: And then we could change, like, the text type, the element type as well. We can do text area.
567 01:06:21.480 ⇒ 01:06:22.620 Casie Aviles: I don’t want to go.
568 01:06:23.580 ⇒ 01:06:25.460 Casie Aviles: So it’s larger, yeah.
569 01:06:28.400 ⇒ 01:06:30.720 Uttam Kumaran: Great, and so now, let’s go ahead and test.
570 01:06:30.920 ⇒ 01:06:34.720 Uttam Kumaran: So that’s the cost of income. I’m sure that I have the steps here. So form…
571 01:06:35.050 ⇒ 01:06:42.920 Uttam Kumaran: I wanna say, text area… Add input called transcript.
572 01:06:43.680 ⇒ 01:06:50.669 Uttam Kumaran: Create LLM chain… And so at this point, let’s go ahead and test again.
573 01:06:50.890 ⇒ 01:06:56.100 Uttam Kumaran: So, gonna copy this in… Paste in our transcript, go ahead and hit submit.
574 01:06:56.750 ⇒ 01:06:58.710 Uttam Kumaran: We’re gonna go to our executions.
575 01:06:59.320 ⇒ 01:07:01.389 Uttam Kumaran: We’re gonna see this execution running.
576 01:07:03.590 ⇒ 01:07:11.329 Uttam Kumaran: And let’s go ahead and refresh, and we can just see that it should have completed at some point. Takes 20 seconds.
577 01:07:14.390 ⇒ 01:07:18.200 Uttam Kumaran: So much for, so far, so that much for a movie model.
578 01:07:20.580 ⇒ 01:07:21.590 Casie Aviles: Yeah, it’s done now.
579 01:07:22.320 ⇒ 01:07:22.910 Uttam Kumaran: That’s it.
580 01:07:23.750 ⇒ 01:07:26.930 Uttam Kumaran: And so, what we should see here, if we zoom in…
581 01:07:28.520 ⇒ 01:07:31.720 Uttam Kumaran: We’re gonna see green around our recent runs.
582 01:07:32.070 ⇒ 01:07:35.380 Uttam Kumaran: Let’s go ahead and copy this to our editor.
583 01:07:36.860 ⇒ 01:07:40.339 Uttam Kumaran: And what we’re gonna go ahead and see here is…
584 01:07:40.550 ⇒ 01:07:46.670 Uttam Kumaran: We click on our form submission, and we’re gonna see our most recent submission information?
585 01:07:46.940 ⇒ 01:07:51.870 Uttam Kumaran: And let’s go ahead and look at our LLM chain. We’re gonna see that this was the input.
586 01:07:52.040 ⇒ 01:07:55.449 Uttam Kumaran: From the form, our transcript, and this is the output.
587 01:07:55.680 ⇒ 01:08:04.449 Uttam Kumaran: the email subject, and the draft, and so the last piece we want to work on here is sending this to Gmail, and so we’re going to type in Gmail.
588 01:08:04.890 ⇒ 01:08:08.150 Uttam Kumaran: We’re gonna go ahead and search for create a draft node.
589 01:08:08.540 ⇒ 01:08:12.319 Uttam Kumaran: And all we’re gonna wanna do here is…
590 01:08:12.480 ⇒ 01:08:20.299 Uttam Kumaran: just map the outputs from our LLM chain to the draft. And so we’re gonna go ahead and drag subject here.
591 01:08:20.450 ⇒ 01:08:23.180 Uttam Kumaran: And we’re gonna go ahead and have an email drop with a message.
592 01:08:23.660 ⇒ 01:08:28.630 Uttam Kumaran: And what we can do is we can go ahead and execute SEP to just test this piece.
593 01:08:28.890 ⇒ 01:08:30.890 Uttam Kumaran: Let’s go ahead and go back to him.
594 01:08:31.479 ⇒ 01:08:33.729 Uttam Kumaran: Here, and we should see our draft here.
595 01:08:34.430 ⇒ 01:08:38.259 Uttam Kumaran: If we have 7, which is soon just happened right here.
596 01:08:38.750 ⇒ 01:08:40.709 Uttam Kumaran: Okay, so this is 7.34.
597 01:08:41.029 ⇒ 01:08:45.010 Uttam Kumaran: Great, and so at this point, let’s go ahead and test end-to-end.
598 01:08:45.210 ⇒ 01:08:47.620 Uttam Kumaran: So, what we’re gonna do, we’re gonna hit save.
599 01:08:48.970 ⇒ 01:08:54.760 Uttam Kumaran: We’re gonna reopen our… Production form.
600 01:09:00.399 ⇒ 01:09:05.330 Uttam Kumaran: And let’s go ahead and test. So, what we should be seeing, if we have our Gmail open.
601 01:09:05.920 ⇒ 01:09:13.250 Uttam Kumaran: We should be seeing… That, once we put the transcript here, press submit.
602 01:09:14.529 ⇒ 01:09:15.889 Uttam Kumaran: We’ll go to NA then.
603 01:09:16.430 ⇒ 01:09:20.339 Uttam Kumaran: We should see in our executions that this form is executing.
604 01:09:20.630 ⇒ 01:09:24.429 Uttam Kumaran: And I’ll stall for a little bit, because it probably doesn’t take any seconds.
605 01:09:24.649 ⇒ 01:09:26.420 Uttam Kumaran: And we can go back to Draft.
606 01:09:26.569 ⇒ 01:09:31.890 Uttam Kumaran: At any moment now… We should see… this.
607 01:09:32.229 ⇒ 01:09:33.010 Uttam Kumaran: We’re here.
608 01:09:35.170 ⇒ 01:09:35.850 Uttam Kumaran: Great.
609 01:09:38.960 ⇒ 01:09:40.140 Casie Aviles: Yes, yeah.
610 01:09:41.470 ⇒ 01:09:42.630 Uttam Kumaran: Cool,
611 01:09:42.800 ⇒ 01:09:46.919 Uttam Kumaran: So I guess one thing I could… I can start adding new lines to this prompt, probably, too, right?
612 01:09:48.109 ⇒ 01:09:48.769 Casie Aviles: Yeah.
613 01:09:50.189 ⇒ 01:09:53.249 Casie Aviles: And then the subject there, you could do that.
614 01:09:56.470 ⇒ 01:09:57.779 Uttam Kumaran: And do you.
615 01:09:59.190 ⇒ 01:09:59.669 Casie Aviles: I didn’t pick.
616 01:09:59.670 ⇒ 01:10:01.000 Uttam Kumaran: of this subject, right?
617 01:10:02.520 ⇒ 01:10:03.060 Casie Aviles: Yep.
618 01:10:11.590 ⇒ 01:10:13.129 Uttam Kumaran: Why do you think that is?
619 01:10:20.220 ⇒ 01:10:22.840 Casie Aviles: Maybe there’s… I think there might be some.
620 01:10:22.840 ⇒ 01:10:23.500 Uttam Kumaran: Oh, that’s…
621 01:10:23.500 ⇒ 01:10:24.579 Casie Aviles: Sending the prompt.
622 01:10:25.230 ⇒ 01:10:26.190 Uttam Kumaran: Yeah…
623 01:10:33.980 ⇒ 01:10:39.450 Uttam Kumaran: So I should tell the output format to reference The user samples, right?
624 01:10:40.030 ⇒ 01:10:41.819 Casie Aviles: Yeah, we can do that as well.
625 01:10:51.580 ⇒ 01:10:52.570 Uttam Kumaran: Yeah.
626 01:10:53.420 ⇒ 01:10:55.479 Uttam Kumaran: What’s up, dude? Hold on one second.
627 01:10:55.820 ⇒ 01:10:56.400 Casie Aviles: Sure.
628 01:11:42.380 ⇒ 01:11:45.259 Uttam Kumaran: Sorry, I’m just at a coffee shop, a friend of mine.
629 01:11:45.450 ⇒ 01:11:48.150 Uttam Kumaran: was just ordering food, he’s like, yo, what’s up?
630 01:11:48.150 ⇒ 01:11:49.080 Casie Aviles: Oh, nice.
631 01:11:49.880 ⇒ 01:11:52.119 Casie Aviles: Oh, you’re at a coffee shop right now?
632 01:11:52.420 ⇒ 01:11:58.950 Uttam Kumaran: Yeah, yeah, well, the coffee shop closed, but there’s, like, a… an adjacent, like, sort of, like, food court area.
633 01:11:59.150 ⇒ 01:12:02.400 Uttam Kumaran: Oh. And I’m just… just sitting here, I didn’t…
634 01:12:02.520 ⇒ 01:12:04.240 Uttam Kumaran: Maybe you can see my video.
635 01:12:05.870 ⇒ 01:12:07.750 Casie Aviles: Oh, nice.
636 01:12:09.690 ⇒ 01:12:12.279 Uttam Kumaran: But hold on, I can’t even see myself. Yes.
637 01:12:12.480 ⇒ 01:12:13.860 Uttam Kumaran: So there’s, like, listen.
638 01:12:14.460 ⇒ 01:12:15.450 Uttam Kumaran: So, yeah, we’re…
639 01:12:15.750 ⇒ 01:12:16.779 Casie Aviles: Messy.
640 01:12:16.940 ⇒ 01:12:21.720 Uttam Kumaran: There’s some people eating, and I’m here just, like, talking to you.
641 01:12:22.450 ⇒ 01:12:27.440 Uttam Kumaran: Okay, cool. So, how about I’m gonna go ahead and… and…
642 01:12:27.560 ⇒ 01:12:29.760 Uttam Kumaran: do a recording, and yeah, I feel pretty good.
643 01:12:30.040 ⇒ 01:12:36.200 Uttam Kumaran: Let me record this, I’m gonna… I’ll send it to the channel, and I’ll send it to…
644 01:12:36.380 ⇒ 01:12:38.530 Uttam Kumaran: Greg, and then, yeah, perfect.
645 01:12:39.180 ⇒ 01:12:42.820 Casie Aviles: Okay, cool. Oh, you’re… it’s not going to be, like, live? Or…
646 01:12:42.820 ⇒ 01:12:45.980 Uttam Kumaran: No, it is gonna be live, he just wants to see, like, an example.
647 01:12:45.980 ⇒ 01:12:46.810 Casie Aviles: Oh, fuck.
648 01:12:47.800 ⇒ 01:12:48.210 Casie Aviles: I see.
649 01:12:48.730 ⇒ 01:12:56.559 Uttam Kumaran: So, I feel like roughly I can get this whole thing done in, like, 15 minutes or so, and I’ll go slow, and then I’ll leave the rest for time for questions.
650 01:12:57.240 ⇒ 01:12:57.950 Casie Aviles: Okay.
651 01:12:59.060 ⇒ 01:13:00.250 Casie Aviles: Yeah, this looks good now.
652 01:13:01.630 ⇒ 01:13:02.660 Uttam Kumaran: Perfect, dude.
653 01:13:03.150 ⇒ 01:13:08.810 Uttam Kumaran: The other… I also wanted to share, we just started working with, this client, I need to send,
654 01:13:09.000 ⇒ 01:13:11.609 Uttam Kumaran: I don’t know if you’ve heard of them, Hedra? Or Hedra?
655 01:13:12.550 ⇒ 01:13:19.870 Casie Aviles: No, not yet. But I was checking them out earlier, it looks like they do, like, generated, like, videos, I think.
656 01:13:20.570 ⇒ 01:13:21.210 Uttam Kumaran: Yeah.
657 01:13:21.210 ⇒ 01:13:22.790 Casie Aviles: Just started.
658 01:13:22.790 ⇒ 01:13:26.209 Uttam Kumaran: Yeah, we just started working with them today, and so…
659 01:13:26.900 ⇒ 01:13:33.169 Uttam Kumaran: I need to finish this loom and go finish some more work for them, but, it’s like data work, so… yeah, it should be good.
660 01:13:33.900 ⇒ 01:13:34.790 Casie Aviles: Nice.
661 01:13:36.310 ⇒ 01:13:44.940 Uttam Kumaran: How is it, how is it today? I haven’t… didn’t check out any progress on the AI side. How’s it been working with Gabe, and, like, what do you think so far on this, like, new sprint process?
662 01:13:46.350 ⇒ 01:13:53.620 Casie Aviles: Oh, it’s… no, I think that’s… it’s good, because, you know, before we were kind of not very organized, and…
663 01:13:53.960 ⇒ 01:14:00.489 Casie Aviles: Yeah, there’s a lot of, like, phases in the whole… Development process that,
664 01:14:00.770 ⇒ 01:14:03.950 Casie Aviles: I wasn’t familiar with, but it’s nice that, you know.
665 01:14:04.300 ⇒ 01:14:09.229 Casie Aviles: It’s nice to have, like, a bit more structure now, and… Being able to differentiate.
666 01:14:09.230 ⇒ 01:14:09.590 Uttam Kumaran: Yeah.
667 01:14:09.590 ⇒ 01:14:12.250 Casie Aviles: Like, our output, because sometimes, like.
668 01:14:12.930 ⇒ 01:14:20.610 Casie Aviles: we didn’t really have, like, clear definitions on our output, like, we have a POC now, we have MVPs, V1s.
669 01:14:21.170 ⇒ 01:14:27.399 Casie Aviles: So it’s nice that we’re able to kind of clearly… Differentiate those, and…
670 01:14:27.550 ⇒ 01:14:32.119 Casie Aviles: Yeah, I think it’s gonna… it’s a lot better now, and I’m learning as well as we go.
671 01:14:33.650 ⇒ 01:14:39.600 Uttam Kumaran: Cool. Yeah, I mean, I’m hoping that you also get to see a little bit of the true product development process.
672 01:14:40.070 ⇒ 01:14:55.820 Uttam Kumaran: But also, like, it’s very, like, this… the way we’re doing things is not common. Nobody does week-long sprints, so, like, most places, they do, like, month- or two-month-long end-to-end. So I’m definitely putting y’all on the fast track, but…
673 01:14:56.010 ⇒ 01:15:00.970 Uttam Kumaran: that’s… that’s my challenge for y’all, is like, how can we go end-to-end in one week? Look, what…
674 01:15:01.270 ⇒ 01:15:08.769 Uttam Kumaran: It’s sort of like when Elon Musk says, like, we need to ship this old car in, like, 6 months. Okay, like, what needs to be true?
675 01:15:08.910 ⇒ 01:15:12.080 Uttam Kumaran: Right? Like, what can you truly push off to the side?
676 01:15:12.200 ⇒ 01:15:14.850 Uttam Kumaran: So that’s sort of my reason why I’m…
677 01:15:15.390 ⇒ 01:15:19.349 Uttam Kumaran: I want us to think about it in a week long, because…
678 01:15:19.890 ⇒ 01:15:29.810 Uttam Kumaran: Typically tasks fill up the time that they’re given, and I really think that we can start to ship some of these high-quality features in a week, and…
679 01:15:30.270 ⇒ 01:15:34.030 Uttam Kumaran: I think it’s up to our team to figure out, like, to kind of, like.
680 01:15:34.440 ⇒ 01:15:36.619 Uttam Kumaran: Shave off the useless stuff.
681 01:15:36.830 ⇒ 01:15:43.369 Uttam Kumaran: understand what we need to nail in discovery, and then kind of drive it out, and then use AI as much to help, right? So…
682 01:15:44.420 ⇒ 01:16:01.209 Uttam Kumaran: Yeah, I’m glad, and I think, like, the outputs of things that we wanted to do haven’t changed. I think more of, like, what I… before, I think we had, like, rough goals in mind, and we were developing features. Instead, I was like, okay, ship something cohesive just around this area.
683 01:16:02.820 ⇒ 01:16:05.129 Casie Aviles: And so, yeah, I mean, I’m hopeful that Gabe can…
684 01:16:05.130 ⇒ 01:16:11.590 Uttam Kumaran: can really be helpful, you know, in organizing, so I would definitely try to lean on him and spend time with him if you haven’t yet.
685 01:16:12.480 ⇒ 01:16:13.270 Casie Aviles: Sure.
686 01:16:14.910 ⇒ 01:16:19.240 Uttam Kumaran: Okay, dude, I know it’s late. Thank you so much. Sorry for, to keep you up.
687 01:16:19.690 ⇒ 01:16:21.140 Casie Aviles: Yeah, no worries.
688 01:16:21.520 ⇒ 01:16:22.649 Casie Aviles: Thank you, Tom.
689 01:16:23.230 ⇒ 01:16:25.160 Uttam Kumaran: Okay, thank you. I’ll talk to you soon.
690 01:16:25.970 ⇒ 01:16:26.650 Casie Aviles: Bye.