Meeting Title: Uttam-Kumaran’s-Personal-Meeting-Room Date: 2024-11-15 Meeting participants: Unknown
WEBVTT
00:00:10.000 ⇒ 00:00:13.000 Hey, guys. Morning.
00:00:13.000 ⇒ 00:00:14.000 Evening.
00:00:14.000 ⇒ 00:00:18.000 Hello, everyone. Amigo. Hi, Utam.
00:00:18.000 ⇒ 00:00:23.000 Hello, Casey.
00:00:23.000 ⇒ 00:00:25.000 Hey, guys.
00:00:25.000 ⇒ 00:00:28.000 you’re looking fly.
00:00:28.000 ⇒ 00:00:29.000 Who?
00:00:29.000 ⇒ 00:00:30.000 Does it snow in Texas, Utah? You.
00:00:30.000 ⇒ 00:00:36.000 No, no, it’s not. It’s just cold. It’s cold in my house today. I opened the door.
00:00:36.000 ⇒ 00:00:37.000 Oh, I see, I see.
00:00:37.000 ⇒ 00:00:43.000 It’s cold outside too. So I just put a jacket on.
00:00:43.000 ⇒ 00:00:49.000 Thank you.
00:00:49.000 ⇒ 00:00:53.000 Will Marian be joining the call today or no?
00:00:53.000 ⇒ 00:00:57.000 I don’t think so because I didn’t give her a heads up.
00:00:57.000 ⇒ 00:00:58.000 Okay, okay.
00:00:58.000 ⇒ 00:01:01.000 But I told her she’ll start joining on Monday.
00:01:01.000 ⇒ 00:01:05.000 Okay.
00:01:05.000 ⇒ 00:01:09.000 Let’s see, it may just may just be this crew.
00:01:09.000 ⇒ 00:01:11.000 Yeah.
00:01:11.000 ⇒ 00:01:16.000 I’m surprised that Ryan isn’t joining. He never missed one.
00:01:16.000 ⇒ 00:01:17.000 Ryan.
00:01:17.000 ⇒ 00:01:18.000 Oh, well, he’s out of town. He’s on vacation.
00:01:18.000 ⇒ 00:01:23.000 Oh, that’s why I was like, he usually replies to me.
00:01:23.000 ⇒ 00:01:26.000 When they ask him something.
00:01:26.000 ⇒ 00:01:32.000 Let me ping Roy.
00:01:32.000 ⇒ 00:01:36.000 Fair enough.
00:01:36.000 ⇒ 00:01:38.000 Nico shall be back Monday, no?
00:01:38.000 ⇒ 00:01:39.000 Yeah.
00:01:39.000 ⇒ 00:02:02.000 Okay.
00:02:02.000 ⇒ 00:02:05.000 Let’s get started. I think it’s probably just us today.
00:02:05.000 ⇒ 00:02:09.000 We have a few people out.
00:02:09.000 ⇒ 00:02:17.000 So yeah, I guess I just wanted to talk about, I think probably on Monday, I’ll do a review of data since we’re like halfway through the month.
00:02:17.000 ⇒ 00:02:20.000 But we’re doing a pretty good job on the content side. In fact.
00:02:20.000 ⇒ 00:02:23.000 one of the posts that we did earlier this week
00:02:23.000 ⇒ 00:02:28.000 got a lot of promotion. And so I’m very, very excited
00:02:28.000 ⇒ 00:02:33.000 For that, because that’s those are some small wins that
00:02:33.000 ⇒ 00:02:37.000 you know we basically get in front of everybody and
00:02:37.000 ⇒ 00:02:44.000 you know, every day people tell me that they enjoy the site and that they’re learning a lot. And so we’re just going to get better there.
00:02:44.000 ⇒ 00:02:51.000 The other thing we’re working on on the design side is we’re kind of two big initiatives.
00:02:51.000 ⇒ 00:02:54.000 We’re closing out like our kind of like initial
00:02:54.000 ⇒ 00:03:00.000 redesign of everything, but we’re working on two big things. One is like from the design brainstorm.
00:03:00.000 ⇒ 00:03:03.000 We’re working on the priority, which is like how do we show
00:03:03.000 ⇒ 00:03:06.000 what we do and explain clearly like
00:03:06.000 ⇒ 00:03:13.000 the impact we make. So we’re going to be working on a few different items there. One is going to be like having like demos.
00:03:13.000 ⇒ 00:03:18.000 of the AI chatbot and the AI like phone agent
00:03:18.000 ⇒ 00:03:23.000 Another couple of things that we’re working on, and I’ll just share um
00:03:23.000 ⇒ 00:03:26.000 I’ll just share this in Notion.
00:03:26.000 ⇒ 00:03:32.000 I’ll cut another couple of things that we’re working on.
00:03:32.000 ⇒ 00:03:37.000 So these two things, explaining clearly what we do and showing what we do
00:03:37.000 ⇒ 00:03:44.000 We’re going to be working on adding the AI phone agent to the site, adding chatbot to the site and the AI and data homepage.
00:03:44.000 ⇒ 00:03:51.000 We’ve also added due dates. So basically for the end of the year, this is the main stuff we’re working on.
00:03:51.000 ⇒ 00:03:56.000 These two are going to be a collaboration between, of course, like design team on the UX and then
00:03:56.000 ⇒ 00:04:00.000 AI team on the actual implementation
00:04:00.000 ⇒ 00:04:05.000 On the homepage, this is where Anne is going to be leading a little bit of research
00:04:05.000 ⇒ 00:04:14.000 To understand, to look at other companies who have this sort of like multi-pronged strategy on how we should
00:04:14.000 ⇒ 00:04:22.000 what parts of our homepage we need to improve in order to convey that we do these two things, what our pricing is, answer all the common questions upfront.
00:04:22.000 ⇒ 00:04:29.000 You know, I read something today that I wanted to share because, you know, I thought it was really impactful for me.
00:04:29.000 ⇒ 00:04:35.000 But one thing that I follow this guy who runs a design agency on Twitter and
00:04:35.000 ⇒ 00:04:40.000 One of the things that he talked about was this, which is like show the process.
00:04:40.000 ⇒ 00:04:46.000 And this is where I think we’re going to try and show as much of what we actually do
00:04:46.000 ⇒ 00:04:51.000 And the impact we make upfront as possible. I think our website is okay now. I think we’re definitely
00:04:51.000 ⇒ 00:04:57.000 on the design side, amazing, the UX side, but we can go even further in terms of actually sharing
00:04:57.000 ⇒ 00:05:01.000 demos of what we do, the impacts it’s made, testimonials.
00:05:01.000 ⇒ 00:05:04.000 And so that’ll be really the next forefront
00:05:04.000 ⇒ 00:05:06.000 of stuff.
00:05:06.000 ⇒ 00:05:11.000 The other things I want to share that we’ve been doing on the design side
00:05:11.000 ⇒ 00:05:16.000 I know, and I know Anne isn’t in this um
00:05:16.000 ⇒ 00:05:18.000 meeting so i’m just gonna
00:05:18.000 ⇒ 00:05:21.000 kind of steal some of her thunder, but I did want to share
00:05:21.000 ⇒ 00:05:25.000 is we’ve been working on sales slides
00:05:25.000 ⇒ 00:05:28.000 In particular.
00:05:28.000 ⇒ 00:05:33.000 We’re using Figma slides to design these. So these are all like amazing templates that we’ll use for sales conversations.
00:05:33.000 ⇒ 00:05:36.000 that and design.
00:05:36.000 ⇒ 00:05:38.000 So I’ll be leveraging this
00:05:38.000 ⇒ 00:05:41.000 anytime we go into a sales meeting and we want to present what we’ve done.
00:05:41.000 ⇒ 00:05:44.000 And of course, here’s everybody’s
00:05:44.000 ⇒ 00:05:48.000 Face here, but we have all this
00:05:48.000 ⇒ 00:05:52.000 stuff we’re working on in terms of like getting these like deck templates ready
00:05:52.000 ⇒ 00:05:57.000 Additionally, we’re working on a deck template for our case studies as well.
00:05:57.000 ⇒ 00:06:00.000 So when we go present what we’ve done
00:06:00.000 ⇒ 00:06:02.000 We’ll be using these slides as well.
00:06:02.000 ⇒ 00:06:07.000 Which is like, here’s what we’ve done. Here’s kind of like the services we offer.
00:06:07.000 ⇒ 00:06:11.000 So I’m very excited that we’ve got this like basically in a really good spot
00:06:11.000 ⇒ 00:06:17.000 And of course, I love showing some of these slides, which is like our clients and partners and who we work with.
00:06:17.000 ⇒ 00:06:24.000 The other thing, I think I did that. The other thing is we’re working on a glossary as well.
00:06:24.000 ⇒ 00:06:28.000 It’s actually on the site right now, but this is mainly for
00:06:28.000 ⇒ 00:06:34.000 SEO.
00:06:34.000 ⇒ 00:06:40.000 And so we’ll actually be using this to kind of publish all of our terms and terminology and things like that.
00:06:40.000 ⇒ 00:06:43.000 And then these will all get picked up by
00:06:43.000 ⇒ 00:06:46.000 Google search and kind of start for us to get ranked.
00:06:46.000 ⇒ 00:06:50.000 Very excited for this. This was kind of joint effort between helene
00:06:50.000 ⇒ 00:06:53.000 Anne and Ryan to kind of put this
00:06:53.000 ⇒ 00:06:55.000 altogether.
00:06:55.000 ⇒ 00:07:06.000 And yeah, I guess that’s really like a lot of the stuff on the design side. On the sales side, we have a couple of really good motions. We have about like five or we have about five
00:07:06.000 ⇒ 00:07:12.000 potential leads that are in process right now, people who we’ve proposed, people who are doing demos for
00:07:12.000 ⇒ 00:07:15.000 So very excited, a lot of which is on the AI side.
00:07:15.000 ⇒ 00:07:19.000 And then I guess I’ll kind of with that, maybe if i
00:07:19.000 ⇒ 00:07:22.000 Miguel, if you want to kind of walk through like what’s on the
00:07:22.000 ⇒ 00:07:25.000 AI team’s agenda and like what you guys are working on
00:07:25.000 ⇒ 00:07:29.000 And anything you guys want to share, that would be
00:07:29.000 ⇒ 00:07:31.000 That would be great.
00:07:31.000 ⇒ 00:07:34.000 Yeah, so…
00:07:34.000 ⇒ 00:07:36.000 This week we got the HPI demo.
00:07:36.000 ⇒ 00:07:38.000 And it worked pretty well.
00:07:38.000 ⇒ 00:07:45.000 So that’s one of like the clients, other sales leads that Utem mentioned
00:07:45.000 ⇒ 00:07:53.000 Yeah, and then I believe there’s also like the Brazilian healthcare stuff. That’s why we want to dive deeper into voice AI.
00:07:53.000 ⇒ 00:07:55.000 Because…
00:07:55.000 ⇒ 00:07:59.000 Yeah, there’s like a lot of opportunities there. So that’s why
00:07:59.000 ⇒ 00:08:05.000 If you guys remember earlier, Utem showed like this voice AI agent on the website or sorry on the
00:08:05.000 ⇒ 00:08:11.000 on the to-do list because that’s something we want to demo to potential visitors of the website.
00:08:11.000 ⇒ 00:08:14.000 And yeah, right now we’re
00:08:14.000 ⇒ 00:08:23.000 Basically, Casey is working on some Zoom stuff, finalizing that. And then the next thing we’re going to be doing is like, we’re going to be creating a
00:08:23.000 ⇒ 00:08:27.000 basically a demo to better show
00:08:27.000 ⇒ 00:08:33.000 as you know Putin mentioned again earlier, to better show like what it does, what’s the capabilities you know
00:08:33.000 ⇒ 00:08:42.000 what it could bring for them. And yeah, to also just let them experience firsthand
00:08:42.000 ⇒ 00:08:43.000 So yeah, I think it’s
00:08:43.000 ⇒ 00:08:46.000 Yeah, so I’m super, super excited um i think
00:08:46.000 ⇒ 00:08:49.000 the agents that we’re going to be working on, we’re going to be
00:08:49.000 ⇒ 00:08:54.000 It’s going to be our ability to show what we’re capable of.
00:08:54.000 ⇒ 00:08:56.000 And so I’m super excited because
00:08:56.000 ⇒ 00:09:00.000 We have use cases and so it’s all going to be awesome to test those things out.
00:09:00.000 ⇒ 00:09:03.000 I mean, that’s what our team is amazing at.
00:09:03.000 ⇒ 00:09:12.000 So, and then kind of, I think Miguel’s team will work with Anne on how do we actually get this into the world on the website. And then on the content side.
00:09:12.000 ⇒ 00:09:16.000 Believe me, we’re going to be posting a ton about this stuff. So I’m kind of just like.
00:09:16.000 ⇒ 00:09:18.000 waiting for us to get a little bit further.
00:09:18.000 ⇒ 00:09:24.000 before just blasting it all out
00:09:24.000 ⇒ 00:09:29.000 I think for AI side, that’s pretty much it.
00:09:29.000 ⇒ 00:09:32.000 Yeah, I think that’s it, to be honest.
00:09:32.000 ⇒ 00:09:35.000 Cool.
00:09:35.000 ⇒ 00:09:39.000 Yeah, I guess I don’t know if we have much
00:09:39.000 ⇒ 00:09:45.000 else to share? I think the only, I think I’ll probably talk about a couple more larger company updates
00:09:45.000 ⇒ 00:09:49.000 When everybody’s in the call on Monday.
00:09:49.000 ⇒ 00:09:53.000 I guess maybe we could spend a
00:09:53.000 ⇒ 00:09:57.000 a minute talking about some AI stuff so i know
00:09:57.000 ⇒ 00:10:02.000 Luke, you were working on Cursor and that stuff was going
00:10:02.000 ⇒ 00:10:03.000 Yeah.
00:10:03.000 ⇒ 00:10:05.000 super well. So I’m really excited because I’ve been using cursor a lot too and
00:10:05.000 ⇒ 00:10:10.000 I think that’s basically what we’re going to try to have everybody’s on the engineering side use cursor for development.
00:10:10.000 ⇒ 00:10:16.000 Yeah, I pretty much ditched VS Code already for cursor because it’s a lot, you know.
00:10:16.000 ⇒ 00:10:17.000 Bye-bye.
00:10:17.000 ⇒ 00:10:19.000 It’s.
00:10:19.000 ⇒ 00:10:29.000 Like you don’t have to use Copilot anymore because it already does everything AI related. You can ask like questions. You can even let it generate code for you just have to
00:10:29.000 ⇒ 00:10:31.000 have the correct prompt or whatever.
00:10:31.000 ⇒ 00:10:33.000 Yeah, it does like auto
00:10:33.000 ⇒ 00:10:39.000 complete stuff in the code. So yeah, pretty much pretty cool
00:10:39.000 ⇒ 00:10:41.000 If you ask me, yeah.
00:10:41.000 ⇒ 00:10:45.000 Even in errors, I’ve tried like
00:10:45.000 ⇒ 00:10:50.000 doing dbt runs and running to errors. And I tried to let it
00:10:50.000 ⇒ 00:10:53.000 find the fix and it was able to do it.
00:10:53.000 ⇒ 00:10:54.000 Wow.
00:10:54.000 ⇒ 00:10:57.000 Although I haven’t tested a lot but
00:10:57.000 ⇒ 00:11:01.000 The first time I tested that, it showed like
00:11:01.000 ⇒ 00:11:08.000 the errors were like the cost it was just pretty simple, like missing comma and stuff like that so
00:11:08.000 ⇒ 00:11:13.000 But I wonder if it can solve like complex ones but yeah it’s pretty cool
00:11:13.000 ⇒ 00:11:16.000 Hell yeah.
00:11:16.000 ⇒ 00:11:20.000 I’ll start using it now. I haven’t started.
00:11:20.000 ⇒ 00:11:21.000 Yeah.
00:11:21.000 ⇒ 00:11:25.000 You should, dude. And then, yeah, if you run out of tokens or whatever, let me know. We could
00:11:25.000 ⇒ 00:11:33.000 And then, yeah, I guess, Miguel, you’re working on like sales agent stuff. How’s that going? Like, I know you sent some stuff just now.
00:11:33.000 ⇒ 00:11:38.000 Oh, yeah, yeah. Basically, I think I will probably communicate with you luke
00:11:38.000 ⇒ 00:11:39.000 Probably not today because it’s you know
00:11:39.000 ⇒ 00:11:41.000 Thank you.
00:11:41.000 ⇒ 00:11:43.000 Sorry, what Utah?
00:11:43.000 ⇒ 00:11:46.000 No, no, no. I guess I was also going to say, I don’t know if the slack
00:11:46.000 ⇒ 00:11:50.000 thing was ready, but maybe if you or Casey want to even demo
00:11:50.000 ⇒ 00:11:51.000 that that could be cool.
00:11:51.000 ⇒ 00:11:52.000 Oh, yeah.
00:11:52.000 ⇒ 00:11:56.000 Casey, I think because you were the main proponent there.
00:11:56.000 ⇒ 00:11:58.000 Do you want to demo it?
00:11:58.000 ⇒ 00:12:00.000 Oh, yeah, sure, sure.
00:12:00.000 ⇒ 00:12:07.000 So yeah, let me just share my screen.
00:12:07.000 ⇒ 00:12:10.000 So if we go over to the sales website
00:12:10.000 ⇒ 00:12:13.000 So basically how it works is
00:12:13.000 ⇒ 00:12:15.000 We’re just going to use this keyword
00:12:15.000 ⇒ 00:12:17.000 Stella.
00:12:17.000 ⇒ 00:12:21.000 And yeah, and you’re going to ask your query like
00:12:21.000 ⇒ 00:12:24.000 What could BrainForge do?
00:12:24.000 ⇒ 00:12:27.000 Or stolosaurus.
00:12:27.000 ⇒ 00:12:30.000 or something like that and
00:12:30.000 ⇒ 00:12:34.000 Yeah, there should be an automation that’s going to trigger
00:12:34.000 ⇒ 00:12:36.000 and reply to this
00:12:36.000 ⇒ 00:12:40.000 message us thread.
00:12:40.000 ⇒ 00:12:44.000 So I guess, yeah, one thing we could improve is like
00:12:44.000 ⇒ 00:12:47.000 how fast i guess the
00:12:47.000 ⇒ 00:12:49.000 agent would reply or
00:12:49.000 ⇒ 00:12:52.000 also in terms of like the knowledge base
00:12:52.000 ⇒ 00:12:54.000 that we fed it.
00:12:54.000 ⇒ 00:13:01.000 we could also improve it because you know we just dumped like the slack conversations there and
00:13:01.000 ⇒ 00:13:02.000 Okay, yeah.
00:13:02.000 ⇒ 00:13:08.000 I think it would be better if you could show the relevance thing
00:13:08.000 ⇒ 00:13:12.000 Yeah, also, yeah, I’m going to check.
00:13:12.000 ⇒ 00:13:15.000 Does it like just take a while to reply or something?
00:13:15.000 ⇒ 00:13:16.000 Currently.
00:13:16.000 ⇒ 00:13:18.000 Yeah, because the process
00:13:18.000 ⇒ 00:13:22.000 I don’t know. Actually, Casey, you should explain sorry
00:13:22.000 ⇒ 00:13:25.000 Oh yeah, sure. Yeah, it’s kind of slow at the moment but they’re
00:13:25.000 ⇒ 00:13:26.000 Oh, there you go. Yeah.
00:13:26.000 ⇒ 00:13:28.000 we saw the we saw the…
00:13:28.000 ⇒ 00:13:29.000 Yeah, the reply.
00:13:29.000 ⇒ 00:13:31.000 It’s about a minute.
00:13:31.000 ⇒ 00:13:38.000 you know because it strings together like different automations maybe we could improve like how fast it is
00:13:38.000 ⇒ 00:13:39.000 But, you know.
00:13:39.000 ⇒ 00:13:45.000 In case it was really not much we can do about the speed because the slow part there is the N810.
00:13:45.000 ⇒ 00:13:46.000 Okay.
00:13:46.000 ⇒ 00:13:49.000 Because we have so much data.
00:13:49.000 ⇒ 00:13:50.000 Yeah, yeah.
00:13:50.000 ⇒ 00:13:52.000 But that’s the thing. It’s like what we’ll start doing is we’ll measure
00:13:52.000 ⇒ 00:13:56.000 We’ll basically measure using helicone or something
00:13:56.000 ⇒ 00:13:57.000 Yeah.
00:13:57.000 ⇒ 00:13:58.000 how long these requests take.
00:13:58.000 ⇒ 00:13:59.000 Yep.
00:13:59.000 ⇒ 00:14:04.000 And then, again, I think the number one thing is making sure it answers the question, right?
00:14:04.000 ⇒ 00:14:05.000 Yeah.
00:14:05.000 ⇒ 00:14:07.000 then it’s like, how do we
00:14:07.000 ⇒ 00:14:08.000 Speed it up.
00:14:08.000 ⇒ 00:14:12.000 use summarization techniques or metadata to speed up rag, basically.
00:14:12.000 ⇒ 00:14:13.000 Yep.
00:14:13.000 ⇒ 00:14:19.000 Yeah, and this is just an agent part. So yeah, I’ve actually tried to integrate like, yeah, what.
00:14:19.000 ⇒ 00:14:22.000 Utam mentioned the evals but
00:14:22.000 ⇒ 00:14:26.000 I’m still figuring out the best way to do it because uh i think
00:14:26.000 ⇒ 00:14:32.000 we can customize like the nodes because we would need to customize these nodes for
00:14:32.000 ⇒ 00:14:34.000 You know, to integrate the uh
00:14:34.000 ⇒ 00:14:35.000 helicon and also
00:14:35.000 ⇒ 00:14:40.000 I don’t think if you can do it there. It has to be through the webhook.
00:14:40.000 ⇒ 00:14:41.000 Hmm.
00:14:41.000 ⇒ 00:14:45.000 like if you go to like the sales hub I built or for the HPI
00:14:45.000 ⇒ 00:14:52.000 Because there’s no way, because the way the helicone I checked it, it has to be sent along with like an API request.
00:14:52.000 ⇒ 00:14:53.000 Yeah, yeah. The Heathers.
00:14:53.000 ⇒ 00:14:54.000 Yeah.
00:14:54.000 ⇒ 00:14:59.000 Yeah, chat doesn’t have headers, so it’s not there. It has to be through the API request.
00:14:59.000 ⇒ 00:15:00.000 So…
00:15:00.000 ⇒ 00:15:03.000 Yeah, it’ll be in the end this because this is just the ui
00:15:03.000 ⇒ 00:15:04.000 Yeah, it’s not going to be there.
00:15:04.000 ⇒ 00:15:07.000 Mm-hmm.
00:15:07.000 ⇒ 00:15:08.000 Yeah, yeah. So that’s something
00:15:08.000 ⇒ 00:15:10.000 I have an example where there’s like
00:15:10.000 ⇒ 00:15:18.000 Yeah, there is one. But yeah, we’ll speak to like, I forgot their name later the tracking guys.
00:15:18.000 ⇒ 00:15:19.000 But yeah.
00:15:19.000 ⇒ 00:15:21.000 Trace Lu.
00:15:21.000 ⇒ 00:15:22.000 Yeah, Snoopy. I hate that guy.
00:15:22.000 ⇒ 00:15:24.000 Yeah.
00:15:24.000 ⇒ 00:15:25.000 Okay.
00:15:25.000 ⇒ 00:15:26.000 I think the other thing is like we’re
00:15:26.000 ⇒ 00:15:32.000 for everybody on the call, we’re going to be generating the way I think about it is like, imagine we had
00:15:32.000 ⇒ 00:15:35.000 a sales expert on every single client
00:15:35.000 ⇒ 00:15:41.000 That’s what we’re going for, right? And so basically it’s like, I think right now we have one channel
00:15:41.000 ⇒ 00:15:48.000 where you can add something and it does happen through Zapier. I think also on the usability, I want it to go towards almost like
00:15:48.000 ⇒ 00:15:50.000 You have Slack bots
00:15:50.000 ⇒ 00:16:00.000 that you can just add at any channel, actually, probably, you know, and you can and it’ll it’ll reference it or we just start with one channel, but there’s like five, there’s like one Slack bot for
00:16:00.000 ⇒ 00:16:05.000 client that we worked on or what we do is we have a client Slack bot
00:16:05.000 ⇒ 00:16:08.000 I’ll kind of let like Miguel and the team think about the ui
00:16:08.000 ⇒ 00:16:09.000 Yeah.
00:16:09.000 ⇒ 00:16:12.000 Because we may have a routing agent versus like
00:16:12.000 ⇒ 00:16:14.000 each of the agents that handle stuff
00:16:14.000 ⇒ 00:16:16.000 But I do think that like this is
00:16:16.000 ⇒ 00:16:21.000 this is like super, super close.
00:16:21.000 ⇒ 00:16:26.000 Yeah, there’s probably like a way to route it
00:16:26.000 ⇒ 00:16:28.000 Yeah, I guess that’s it from
00:16:28.000 ⇒ 00:16:30.000 But yeah, let’s finalize the functionality first.
00:16:30.000 ⇒ 00:16:38.000 Yeah. And once we do one and we know it’s accurate, then it’s easy to expand because it’s just different knowledge bases mainly.
00:16:38.000 ⇒ 00:16:39.000 Yes.
00:16:39.000 ⇒ 00:16:41.000 It’s like, and then basically, I think what we try to work on is like, how do we
00:16:41.000 ⇒ 00:16:44.000 How do we chunk better how do we
00:16:44.000 ⇒ 00:16:47.000 ingest the knowledge bases book more
00:16:47.000 ⇒ 00:16:52.000 But ideally, also the knowledge base engine is what we’re going to be using across
00:16:52.000 ⇒ 00:16:56.000 everything right from operations to sales, also on the engineering side
00:16:56.000 ⇒ 00:16:59.000 we’re going to start to have the same knowledge base so really
00:16:59.000 ⇒ 00:17:06.000 The thing here that’s the most important to get right is the knowledge base rag
00:17:06.000 ⇒ 00:17:07.000 Yeah, yeah.
00:17:07.000 ⇒ 00:17:11.000 And because again, let’s say we let’s say we want to say like
00:17:11.000 ⇒ 00:17:16.000 you want to ask an engineering question to it. Is it going to be able to answer versus a sales question versus like.
00:17:16.000 ⇒ 00:17:18.000 something else you know so
00:17:18.000 ⇒ 00:17:28.000 That would be very, very cool to do.
00:17:28.000 ⇒ 00:17:31.000 Okay.
00:17:31.000 ⇒ 00:17:34.000 Casey, sorry, I have a question.
00:17:34.000 ⇒ 00:17:43.000 The one you demoed earlier, is that all already using that reset that 3072 or the 1536?
00:17:43.000 ⇒ 00:17:45.000 You mean for the vector embeddings?
00:17:45.000 ⇒ 00:17:47.000 Yeah, yeah, yeah.
00:17:47.000 ⇒ 00:17:52.000 Yeah, let me check but it should be used should be using the large one yeah it’s using the large one
00:17:52.000 ⇒ 00:17:56.000 Okay. Okay, yeah.
00:17:56.000 ⇒ 00:18:01.000 And yeah, we’ll play around once we finish with the demos. We’ll play around how to
00:18:01.000 ⇒ 00:18:05.000 chunk it. Also, Utah, I think what we could do is for example
00:18:05.000 ⇒ 00:18:15.000 we could have like three separate knowledge bases i would say. So for example, the AI agent will determine is this like an engineering question
00:18:15.000 ⇒ 00:18:17.000 So it only searches the engineering knowledge base
00:18:17.000 ⇒ 00:18:19.000 stuff like that because
00:18:19.000 ⇒ 00:18:23.000 Once it’s in the knowledge base, there’s really nothing much we can do.
00:18:23.000 ⇒ 00:18:24.000 That’s why I want to separate it.
00:18:24.000 ⇒ 00:18:27.000 Well, here’s the thing is like, I think the engineering knowledge base may not need
00:18:27.000 ⇒ 00:18:33.000 to have an understanding of the sales, but the sales stuff will need to know the engineering.
00:18:33.000 ⇒ 00:18:34.000 Yeah, yeah.
00:18:34.000 ⇒ 00:18:41.000 Because for some clients, like for pool parts, we haven’t written down all the stuff we’ve done. All of that is in the GitHub.
00:18:41.000 ⇒ 00:18:43.000 Okay.
00:18:43.000 ⇒ 00:18:44.000 So…
00:18:44.000 ⇒ 00:18:47.000 Yeah.
00:18:47.000 ⇒ 00:18:51.000 We can probably just prompt the agent if it’s a sales question, you’ll probably need to use
00:18:51.000 ⇒ 00:18:55.000 But if it’s an engineering question, then this one
00:18:55.000 ⇒ 00:18:57.000 Yeah.
00:18:57.000 ⇒ 00:19:00.000 I think, Casey, we had like a similar thing
00:19:00.000 ⇒ 00:19:06.000 We did this before for i forgot the company name, Lushcovoliche, right?
00:19:06.000 ⇒ 00:19:08.000 oh yeah it’s a e-commerce company.
00:19:08.000 ⇒ 00:19:10.000 Yeah, yeah.
00:19:10.000 ⇒ 00:19:15.000 Oh, yeah. We forgot. Casey, we should probably tell Utem about that now.
00:19:15.000 ⇒ 00:19:21.000 We built like a chat bot for like an e-commerce company before
00:19:21.000 ⇒ 00:19:22.000 Hmm.
00:19:22.000 ⇒ 00:19:24.000 They were a polish company
00:19:24.000 ⇒ 00:19:26.000 basically bed sheets, towels
00:19:26.000 ⇒ 00:19:28.000 everything like that curtains
00:19:28.000 ⇒ 00:19:31.000 IKEA stuff, basically. And then…
00:19:31.000 ⇒ 00:19:36.000 that chatbot is basically, you know, it has access to over 20,000 products i think
00:19:36.000 ⇒ 00:19:39.000 And then it also has like this ability to like check
00:19:39.000 ⇒ 00:19:46.000 So for example, hey, my order number is 12345. It’ll check it. It’ll send like a request and then it’ll come back.
00:19:46.000 ⇒ 00:19:47.000 Hmm.
00:19:47.000 ⇒ 00:19:51.000 I think that’s something we can offer. I don’t know. I forgot about that. That was one of our bigger projects.
00:19:51.000 ⇒ 00:19:53.000 Me and Casey.
00:19:53.000 ⇒ 00:19:55.000 Oh, okay.
00:19:55.000 ⇒ 00:19:59.000 Interesting.
00:19:59.000 ⇒ 00:20:06.000 Yeah, so that’s another thing too is like, let’s say we’re going to a client, existing client, we’re like, how can we use AI?
00:20:06.000 ⇒ 00:20:09.000 for your business, that’s something that would be, that’s a great question that I would ask the agent.
00:20:09.000 ⇒ 00:20:10.000 You know?
00:20:10.000 ⇒ 00:20:13.000 Yeah.
00:20:13.000 ⇒ 00:20:14.000 Cool.
00:20:14.000 ⇒ 00:20:15.000 you know order status, I remember, yeah.
00:20:15.000 ⇒ 00:20:19.000 Yeah, yeah.
00:20:19.000 ⇒ 00:20:24.000 Okay, great. I think probably short meeting today. Anything else?
00:20:24.000 ⇒ 00:20:32.000 we want to cover.
00:20:32.000 ⇒ 00:20:33.000 Not from my end.
00:20:33.000 ⇒ 00:20:34.000 Cool.
00:20:34.000 ⇒ 00:20:35.000 I think I’m good as well.
00:20:35.000 ⇒ 00:20:36.000 Okay.
00:20:36.000 ⇒ 00:20:41.000 Okay, I’ll give everyone some time back and then, yeah, let’s talk in Slack. I think I’m going to be
00:20:41.000 ⇒ 00:20:45.000 working on some AI stuff tomorrow.
00:20:45.000 ⇒ 00:20:46.000 We don’t see either.
00:20:46.000 ⇒ 00:20:49.000 But again, don’t feel pressured to do anything over the weekend. I think me and Miguel will be
00:20:49.000 ⇒ 00:20:50.000 Yeah.
00:20:50.000 ⇒ 00:20:51.000 trying to do some voice stuff tomorrow.
00:20:51.000 ⇒ 00:20:56.000 But yeah, otherwise I’ll look forward to talking to everybody on Monday.
00:20:56.000 ⇒ 00:20:58.000 Cool. See you guys. Thanks.
00:20:58.000 ⇒ 00:20:59.000 See you guys. Bye-bye. Have a good one.
00:20:59.000 ⇒ 00:21:01.000 Okay. Thank you.