Meeting Title: Sales-Hub-Automation-Discussion Date: 2024-09-10 Meeting participants: Joshuadeveyra, Uttam Kumaran
WEBVTT
1 00:01:47.070 ⇒ 00:01:47.730 Uttam Kumaran: Yeah.
2 00:01:48.830 ⇒ 00:01:49.610 joshuadeveyra: Hello! Hello!
3 00:01:50.070 ⇒ 00:01:51.161 Uttam Kumaran: Hey? You can hear me.
4 00:01:51.680 ⇒ 00:01:53.499 joshuadeveyra: Yeah. I can hear you.
5 00:01:53.500 ⇒ 00:01:54.150 Uttam Kumaran: Hey!
6 00:01:54.860 ⇒ 00:02:03.960 joshuadeveyra: Yeah, if my voice breaks out a bit sorry, because my cat, you know, he he nipped on the router wi-fi. So right now I’m on my mobile data.
7 00:02:04.470 ⇒ 00:02:06.686 Uttam Kumaran: Okay. You’re totally fine.
8 00:02:07.240 ⇒ 00:02:10.419 joshuadeveyra: I’ll probably have to call support tomorrow. Have them replace it.
9 00:02:10.669 ⇒ 00:02:13.019 Uttam Kumaran: That’s funny. How’s everything going.
10 00:02:14.301 ⇒ 00:02:22.030 joshuadeveyra: Yeah, pretty. Okay. I just got home from, you know. My mom celebrated her birthday. It was actually 9, 11
11 00:02:22.882 ⇒ 00:02:26.840 joshuadeveyra: but you wanted to celebrate earlier because she’s gonna be out of the country.
12 00:02:27.090 ⇒ 00:02:30.370 Uttam Kumaran: Okay, yeah. My dad’s birthday is today, actually 9, 10.
13 00:02:31.080 ⇒ 00:02:32.570 joshuadeveyra: Oh, okay, okay.
14 00:02:33.130 ⇒ 00:02:35.669 joshuadeveyra: you’re still working. You’re not celebrating with family.
15 00:02:35.670 ⇒ 00:02:39.000 Uttam Kumaran: No, I’m not. I’m not in the same city. I’m pretty. I.
16 00:02:39.000 ⇒ 00:02:39.679 joshuadeveyra: I want it.
17 00:02:39.680 ⇒ 00:02:43.239 Uttam Kumaran: They’re in. They’re in a they’re in California. So.
18 00:02:43.610 ⇒ 00:02:44.899 joshuadeveyra: Oh, yeah, pretty far.
19 00:02:45.310 ⇒ 00:02:46.220 Uttam Kumaran: Yeah.
20 00:02:47.720 ⇒ 00:02:50.430 Uttam Kumaran: okay. So tell me what you what you thought.
21 00:02:51.807 ⇒ 00:02:58.569 joshuadeveyra: Yeah, I have like a couple of questions, right? Is that gonna be for each sales lead? Do we create like
22 00:02:58.820 ⇒ 00:03:00.870 joshuadeveyra: it’s individual agent.
23 00:03:01.210 ⇒ 00:03:02.540 joshuadeveyra: or how, how.
24 00:03:02.880 ⇒ 00:03:16.859 Uttam Kumaran: I don’t know. So here’s like the the general overview. So when I get a lead in, you know, there’s a bunch of documents and processes that kick off right like. We have meetings with them. So there’s meeting notes. We have
25 00:03:16.930 ⇒ 00:03:22.949 Uttam Kumaran: proposals. So those are Pdfs, we have emails. So those are like email text back and forth.
26 00:03:23.320 ⇒ 00:03:40.419 Uttam Kumaran: We have, like internal communications. Right? This may be on slack. All of that. The the problem that’s happening is like all of that needs to be centralized in one place. And ideally, I can have an agent on top of that data so that I can say, Hey, they want to modify the proposal
27 00:03:40.580 ⇒ 00:03:43.150 Uttam Kumaran: like, here’s what it needs to look like or
28 00:03:43.190 ⇒ 00:03:53.069 Uttam Kumaran: given all the context like, what would be the best strategy here, right? So basically creating a chat bot for every lead. I don’t know whether it can be. It can be or probably be.
29 00:03:53.440 ⇒ 00:03:55.199 Uttam Kumaran: I mean, I don’t know. Maybe it has to be.
30 00:03:55.200 ⇒ 00:03:58.719 joshuadeveyra: It’s not. Yeah, it’s it has to be a separate agent for each.
31 00:03:58.720 ⇒ 00:04:02.789 Uttam Kumaran: Or like a or like a separate agent, and then the same sub agents.
32 00:04:04.520 ⇒ 00:04:05.480 Uttam Kumaran: right.
33 00:04:08.570 ⇒ 00:04:16.509 joshuadeveyra: Yeah, yeah, basic. Wait, let me think. Because if yeah, cause I was actually thinking of like that approach, then maybe we could have, like.
34 00:04:17.089 ⇒ 00:04:21.640 joshuadeveyra: basically, you know, a central agent, and then we’ll develop separate agents.
35 00:04:23.740 ⇒ 00:04:29.480 joshuadeveyra: But I think we should develop 1 1st and then see where it goes from there, and then we can just duplicate it right.
36 00:04:30.200 ⇒ 00:04:31.340 Uttam Kumaran: Okay. Cool.
37 00:04:33.240 ⇒ 00:04:36.341 joshuadeveyra: I did like. I did add some comments to this stuff
38 00:04:37.400 ⇒ 00:04:44.189 joshuadeveyra: data sources. If you could like, send me some examples of this, maybe, like a past lead, past lead, or maybe an existing lead.
39 00:04:44.320 ⇒ 00:04:46.139 joshuadeveyra: so I could obtain it.
40 00:04:46.920 ⇒ 00:04:47.320 Uttam Kumaran: Okay, cause.
41 00:04:47.320 ⇒ 00:04:49.660 joshuadeveyra: One thing I’m not
42 00:04:50.568 ⇒ 00:04:59.539 joshuadeveyra: cause. One thing I’m not really sure with is how good relevance is rug is. Last time I checked.
43 00:04:59.540 ⇒ 00:05:02.070 Uttam Kumaran: I want to find out a book.
44 00:05:02.360 ⇒ 00:05:12.620 Uttam Kumaran: It’s bad. Okay? I mean, then then we can. We’ll we have to do. We’ll we can try something else. Try it in relevance. If it’s not good, then fine. Maybe we switch to.
45 00:05:12.620 ⇒ 00:05:13.429 joshuadeveyra: Yeah, yeah.
46 00:05:13.430 ⇒ 00:05:16.229 Uttam Kumaran: Like. What else? What other options do you think we could switch to like?
47 00:05:17.630 ⇒ 00:05:19.720 Uttam Kumaran: You could try to switch to flow wise
48 00:05:21.990 ⇒ 00:05:22.690 Uttam Kumaran: because I heard the.
49 00:05:22.690 ⇒ 00:05:24.900 joshuadeveyra: I mean we could just do. Openai.
50 00:05:25.670 ⇒ 00:05:27.450 joshuadeveyra: Open air is pretty good.
51 00:05:28.000 ⇒ 00:05:35.132 Uttam Kumaran: Use an open, but then use Openai. And then what do you mean like? How do you actually get the documents in there? Like, let’s say it’s Pdf and stuff.
52 00:05:35.759 ⇒ 00:05:40.039 joshuadeveyra: We’ll create like an open, a assistant, basically another agent, right?
53 00:05:40.498 ⇒ 00:05:52.009 joshuadeveyra: And then it all it does is basically it just gets you information. So basically, what we’re gonna do is we’re gonna send like, for example, a query from the relevance agent all
54 00:05:52.420 ⇒ 00:05:59.929 joshuadeveyra: to the open AI assistant. That’s what they’re called. And then, because the rug in Openai is really good rug in
55 00:06:00.030 ⇒ 00:06:01.440 joshuadeveyra: relevance is not.
56 00:06:02.360 ⇒ 00:06:04.199 joshuadeveyra: But that’s probably something we can do.
57 00:06:04.590 ⇒ 00:06:05.150 Uttam Kumaran: Okay.
58 00:06:05.150 ⇒ 00:06:09.750 joshuadeveyra: To not, you know, complicate things. Because if we go through pinecone and all that stuff.
59 00:06:09.860 ⇒ 00:06:13.339 joshuadeveyra: it’s just Pdf, we just need to structure the data. We don’t need all the.
60 00:06:13.590 ⇒ 00:06:14.580 Uttam Kumaran: Okay. Okay.
61 00:06:17.510 ⇒ 00:06:23.089 joshuadeveyra: Although I will. I will need like. Do we have, like an open AI platform account.
62 00:06:23.090 ⇒ 00:06:23.810 Uttam Kumaran: Yes.
63 00:06:24.550 ⇒ 00:06:28.400 joshuadeveyra: Okay, yeah. Yeah. And then I probably have to write some.
64 00:06:29.221 ⇒ 00:06:32.639 joshuadeveyra: Code. I’ll just post it on Ng rock for now.
65 00:06:34.650 ⇒ 00:06:38.080 Uttam Kumaran: Okay, cool. So you’re gonna you’re gonna host it where.
66 00:06:39.570 ⇒ 00:06:40.280 joshuadeveyra: And
67 00:06:41.460 ⇒ 00:06:44.099 joshuadeveyra: it’s just local.
68 00:06:44.420 ⇒ 00:06:45.270 Uttam Kumaran: Okay. Okay.
69 00:06:47.630 ⇒ 00:06:50.090 joshuadeveyra: And it’s gonna you know.
70 00:06:50.180 ⇒ 00:06:51.290 joshuadeveyra: it’s just low.
71 00:06:51.960 ⇒ 00:06:58.240 Uttam Kumaran: Okay, cool. So why don’t I? I’m gonna send you open AI access. I’ll send you
72 00:06:59.820 ⇒ 00:07:06.329 Uttam Kumaran: like. I’ll send you just like a example of one lead. And like all the info, basically
73 00:07:06.460 ⇒ 00:07:07.240 Uttam Kumaran: again, like.
74 00:07:08.020 ⇒ 00:07:08.610 Uttam Kumaran: Email, okay.
75 00:07:08.835 ⇒ 00:07:09.060 joshuadeveyra: Yeah.
76 00:07:09.060 ⇒ 00:07:10.159 Uttam Kumaran: You know, will be
77 00:07:10.460 ⇒ 00:07:23.449 Uttam Kumaran: easier, but I mainly want to consider like the Pdfs. I mean. I don’t know. I’ll send you everything, and then you can let me know what the best route is. The other thing that would be helpful is like. If the agent can also have Internet access.
78 00:07:23.460 ⇒ 00:07:31.170 Uttam Kumaran: cause. All these companies are like they have websites and stuff. Cause. One of the things that I do is when a lead comes in. I have to go, research.
79 00:07:31.360 ⇒ 00:07:32.090 Uttam Kumaran: Right? So yeah.
80 00:07:32.090 ⇒ 00:07:32.590 joshuadeveyra: Yeah, yeah.
81 00:07:32.590 ⇒ 00:07:35.550 Uttam Kumaran: Spot. I wanted to own everywhere, from research
82 00:07:35.570 ⇒ 00:07:39.080 Uttam Kumaran: to proposal creation to negotiation
83 00:07:39.390 ⇒ 00:07:41.299 Uttam Kumaran: right? And so.
84 00:07:41.300 ⇒ 00:07:42.010 joshuadeveyra: Okay.
85 00:07:42.010 ⇒ 00:07:50.679 Uttam Kumaran: Let’s let’s like, you see what you can do. And I’ll also, I’m gonna just basically try to in that dock in that notion. I’m just gonna put a bunch of information now that I know kind of the question.
86 00:07:50.680 ⇒ 00:07:52.000 joshuadeveyra: Oh, yeah, yeah, that would be great.
87 00:07:52.020 ⇒ 00:07:53.180 Uttam Kumaran: Let’s go from there.
88 00:07:53.990 ⇒ 00:07:59.650 joshuadeveyra: Yeah. And then with government open AI account, we need that.
89 00:07:59.750 ⇒ 00:08:04.149 joshuadeveyra: So we need research. It’s probably just gonna be Google Linkedin.
90 00:08:05.110 ⇒ 00:08:08.579 joshuadeveyra: Yeah, is there any anywhere else? Glass door? Probably right.
91 00:08:08.580 ⇒ 00:08:09.840 Uttam Kumaran: Yeah. And then
92 00:08:10.530 ⇒ 00:08:15.949 Uttam Kumaran: I think you can. Also we have you. Have you? Did the Apollo right? So start taking the stuff from Apollo could be nice, too.
93 00:08:15.950 ⇒ 00:08:17.760 joshuadeveyra: Oh, yeah. Apollo, yeah. Yeah.
94 00:08:25.630 ⇒ 00:08:33.209 joshuadeveyra: Okay, yeah. And then proposal. If you could send me like an a sample proposal, maybe the one we sent to Vito something like that. Yeah.
95 00:08:34.860 ⇒ 00:08:37.579 joshuadeveyra: And the Pdf sample Pdfs.
96 00:08:37.580 ⇒ 00:08:40.459 Uttam Kumaran: If we take Vaticococo as the example. Yeah, that’s actually good.
97 00:08:40.460 ⇒ 00:08:42.120 joshuadeveyra: Oh, yeah, yeah, that’s a good idea.
98 00:08:45.640 ⇒ 00:08:46.880 joshuadeveyra: Okay, yeah.
99 00:08:48.240 ⇒ 00:08:51.510 Uttam Kumaran: Okay. Great anything else.
100 00:08:54.325 ⇒ 00:08:57.450 joshuadeveyra: So we need Asian to
101 00:08:57.590 ⇒ 00:09:00.359 joshuadeveyra: basically have the ability to Google search.
102 00:09:03.830 ⇒ 00:09:05.940 joshuadeveyra: I think so far, particularly.
103 00:09:07.053 ⇒ 00:09:08.760 joshuadeveyra: data sources.
104 00:09:09.480 ⇒ 00:09:12.020 joshuadeveyra: The meeting transcripts. Yeah, this is another one.
105 00:09:12.150 ⇒ 00:09:15.420 joshuadeveyra: How how would we know? Like, it’s for a specific, you know.
106 00:09:18.810 ⇒ 00:09:22.529 Uttam Kumaran: So there’s a couple of ways. One is the title of the
107 00:09:23.080 ⇒ 00:09:28.450 Uttam Kumaran: of the meeting. The second one is the people, the the people who are on the meeting. The 3rd one.
108 00:09:28.450 ⇒ 00:09:29.030 joshuadeveyra: 5.
109 00:09:29.030 ⇒ 00:09:33.859 Uttam Kumaran: Like manual. Right? So, for example, if there’s a meeting with the client.
110 00:09:34.520 ⇒ 00:09:37.889 Uttam Kumaran: then that’s of course, like a meeting with them.
111 00:09:38.200 ⇒ 00:09:43.259 Uttam Kumaran: and maybe something else, or we manually move it again. We’re not having like hundreds of meetings
112 00:09:43.300 ⇒ 00:09:48.400 Uttam Kumaran: so ideally. We just have a process where I can throw the meeting notes into the
113 00:09:50.890 ⇒ 00:09:52.980 Uttam Kumaran: into the knowledge base basically.
114 00:09:54.930 ⇒ 00:09:55.910 joshuadeveyra: Yeah, yeah.
115 00:09:55.910 ⇒ 00:10:06.319 Uttam Kumaran: The best way could be possibly like, we have a Google drive folder for every lead and that way it’s easy to throw everything in there right like, what’s the interface for the file store?
116 00:10:09.050 ⇒ 00:10:11.639 joshuadeveyra: For open. AI. It’s honestly just
117 00:10:11.660 ⇒ 00:10:13.570 joshuadeveyra: you just upload something on it.
118 00:10:14.370 ⇒ 00:10:15.060 Uttam Kumaran: Okay.
119 00:10:15.430 ⇒ 00:10:17.470 Uttam Kumaran: is that, can you do that on the Ui.
120 00:10:18.440 ⇒ 00:10:26.570 joshuadeveyra: Yeah, yeah, you can do that on the Ui, they have like, a pretty good Ui, I’ll create a loom video, basically, just a quick introduction to open AI. I guess.
121 00:10:27.250 ⇒ 00:10:35.549 Uttam Kumaran: Okay, cool. Yeah. I know how the I I’ve used like the Api before. I never use the assistance. Api. But yeah, I mean again. Maybe we just let’s do whatever works.
122 00:10:35.550 ⇒ 00:10:36.730 joshuadeveyra: Yeah.
123 00:10:36.730 ⇒ 00:10:39.980 Uttam Kumaran: We can. We can see everything connected to Google drive and stuff. Later.
124 00:10:40.690 ⇒ 00:10:41.700 joshuadeveyra: Yeah, build it from there.
125 00:10:42.370 ⇒ 00:10:49.109 joshuadeveyra: Yeah, I mean, it saves us a lot of time. If the relevance one is good because there’s like an option.
126 00:10:50.420 ⇒ 00:10:51.220 joshuadeveyra: Thanks.
127 00:10:52.050 ⇒ 00:10:54.060 joshuadeveyra: where, if you input them.
128 00:10:54.410 ⇒ 00:10:56.196 joshuadeveyra: it’s good
129 00:10:57.620 ⇒ 00:10:58.820 joshuadeveyra: knowledge, base.
130 00:10:59.120 ⇒ 00:10:59.540 Uttam Kumaran: Okay.
131 00:11:00.140 ⇒ 00:11:04.549 joshuadeveyra: Where we can just add stuff to the knowledge base. But the issue there is.
132 00:11:04.920 ⇒ 00:11:10.320 joshuadeveyra: Yeah, I’ll go to this. I’ll test it out today the remainder of my shift. So yeah.
133 00:11:11.270 ⇒ 00:11:12.570 Uttam Kumaran: Okay, okay, cool.
134 00:11:13.840 ⇒ 00:11:18.319 Uttam Kumaran: Okay, perfect. Otherwise. You got the money. You got the confirmation for the money right?
135 00:11:18.800 ⇒ 00:11:20.030 joshuadeveyra: Yeah, yeah, just now.
136 00:11:20.030 ⇒ 00:11:21.180 Uttam Kumaran: Okay, perfect. Okay.
137 00:11:21.500 ⇒ 00:11:23.270 joshuadeveyra: Okay, thanks. Your time. Have a good day.
138 00:11:23.270 ⇒ 00:11:25.259 Uttam Kumaran: Thank you so much. Talk to you soon.
139 00:11:25.850 ⇒ 00:11:26.650 joshuadeveyra: Okay bye, bye.