# Grok Bot for GTM with Krista Letz (Ent GTM, SpaceXAI) and Eric Nowoslawski (Founder, GrowthEngineX) Channel: ~ The Signal ~ Video: https://www.youtube.com/watch?v=DtiKujg-gfc Duration: 56 min Language: English Words: 10989 Transcript page: https://viewrankai.com/tools/youtube-transcript/DtiKujg-gfc --- [0:03] Alrighty, sweet. I see folks rolling in. Incredible. Um, all right. Cool. Well, thank you everybody for being here. I'm excited to have uh both Christa and Eric here to walk us through the magic of Grockbot. Grockbot or Grock bots? This is my my only question for you first. Grockbot, but you call them bots. So, I call them [0:27] my bots, but it's Grockbot. Amazing. Well, thank you for being here. I'm excited to uh to learn from you and uh I have been personally using uh Grockbot for a couple of weeks and I'll just say upfront like my experience with it has been um like I basically I think this is like the most magical AI experience I've had since sort of claude code whatever it was I think Opus 4.6 six in like December, January and sort of like chat GBT before that. It's it sort of feels to me like a new novel different um workflow that is sort of like the power of codeex uh or cursor or cloud code but like for [1:11] a non-technical person like me. Um and there's an amazing mobile experience. Uh it's very cool. So, um, I'm excited to get into go to market use cases and, uh, Christa, you're at SpaceX AI. Yes. On the enterprise go to market team. Um, the last thing I'd say is I'll put in the chat here in a second. Um, but you posted an amazing article about Grockbot use cases for GTM uh, a week or two ago that uh, is awesome. Everybody should read it. I'll put it in in the chat here. I think it was I think it's almost at 3 million impressions. So, uh that was amazing to see. Um so, thank you again [1:53] for being here. Thank you for giving us your time and uh yeah, let's jump right into it. Sounds good. I can kind of do a quick introduction to of Grockbot and how to work with it. I think one thing to keep in mind is Grockbot is a little bit different than you use most AI tools. So, I would approach it differently than uh what you're leveraging today. So what's kind of different and I can share [2:14] my screen here. Instead of like thinking of Grockbot as a taskbased tool, I like to think of them as teammates or colleagues. And I have a bot for like each person that I would work with normally. So you know, as Brendan said, I'm on the enterprise go to market team here at SpaceX. I work across uh a set of customers. I also do a lot of outbounding uh to reach new customers and have a lot of different products uh that we work with customers on across uh cursor across SpaceX now Grockbot and more. Uh so I think of it of like hey you know I have an engineer [2:56] that I typically ask any uh technical questions. Uh cursor is a very technical product so I definitely need a lot of help uh from this engineer if I'm live on a call. uh I you know sell products where we put together this like software factory. So I'm always kind of asking questions and uh instead of having to tap an engineer or a product manager, I can just use this bot and ask questions live and get customer f or facing answers. So if I wanted to say like hey I'm live on a call with a customer and they're like hey what are best practices on setting up cloud agents? I can ask my [3:33] engineer like hey can you put together a customerf facing answer on setting up cloud agents you ask like email format or usually I just kind of guess it's format so I'll just type this here so it'll uh actually look at our codebase I have this connected to uh GitHub but if you have a tool like glean or something tool or too you can also use that as well so if you use some type of like semantic search layer. Uh you can plug that in. Uh it's very fast as you can see. Uh so it gave me an answer and that felt like 2 seconds and I can either [4:12] send this or sometimes it like will recognize patterns and it'll create like even like a a quick Slack answer. Uh so this has helped me like tremendous tremendously because normally if I didn't know the answer I'd have to like you know go search through notion ask an engineer uh look through our docs and follow up after the call but now I can uh you know actually spend time with customers live and help them uh set up cloud agents while live on a call. Uh you can pin bots. So I have these kind of organized and this is my go to market [4:45] team. I have my mostused bots uh pinned. Uh, so this is my chief of staff. This is my dog. Her name is Olive. Uh, so she helps helps me out with my day-to-day. Uh, you can see that you can use these bots to kind of talk to each other. Uh, my chief of staff I primarily use more for admin related tasks. So, when I'm on my way to office, it'll prep me for my calls for the day. And as Brendan [5:09] mentioned, I use the mobile app a ton. Uh, I talk to my mobile app. I don't know if anyone is like this, but I have like a fear of leaving my laptop, especially at end of quarter. And I think most people at SpaceX are like this. Before cloud agents and before Grockbot, we'd walk around with open laptops. And that's kind of why we've created Grockbot is these agents work overnight. So you can shut your laptop, continue working on something, or even kick off a project on your phone. And these work for long periods of time. So I would say be much more ambitious with Grockbot. Don't use it as you normally [5:50] would like a chat interface. Uh it's not just for asking questions or doing light research. Uh I like to say that like it's not just uh you know researching, it's actually going and doing the work for you. So it's very proactive. Say I'm doing research about a customer or something, watching a video. It'll be like, great, you know, video's ended. I put together an email on uh, you know, [6:15] these three things the CTO mentioned. Uh, this is how it relates to cursor. I drafted an email for you. It's in your Gmail drafts. So, you can be much more endtoend. It's more of like a doing partner uh rather than a thought partner. It can be a thought partner, too. So I do use it for like you know research but uh it can actually do things on your behalf uh and can run overnight on long long running tasks. Uh each bot has its own computer. Uh, so I think this is a differentiator too as at least for me with other AI tools, it felt like it took a long time to like [6:49] get started or I do feel like some sales tools are very behind uh with like plugins and MCPs, but since uh my bot has a computer, I can teach it tasks. So maybe I wanted to research uh what does like the stock market look today? So yeah, what does the stock market look like today? Any recent funding rounds? I'm recording [7:17] this task right here. And after I'm done, I can stop it and that will be turned into a routine and my bot will learn from what I demonstrated. I think our BDR team has used this. We use some tools where like the plugin isn't that great. So, uh, you can like click around and show your bot how you use these tools and it'll remember it the first time. So, I would onboard, uh, Grothbot like a teammate. Give it all the tools that you have access to. Uh, show uh, your bot how to do things and what your workflows look like. When I first onboarded at Cursor and now [7:59] SpaceX, uh, I think I would like, you know, look over people's shoulders. uh it would take a couple times for me to watch what people were doing for me to like really consume it on my own. Uh but what I like about Grockbot is it kind of learns the first time around. And uh like kind of like an interesting uh like thing that we did uh is we uh wanted to know like what uh bots or what like agents complained about. So, we had this like channel and it was like agents complaining and the primary thing that they complained about is if you go and tell them to do a job but you don't give [8:34] them access to all the tools that they need. It's kind of like hiring somebody uh in sales and not giving them access to like you know X or uh you know Salesforce. It's like giving somebody a job but no tools to go and do their job. So give your bot access to all the tools that you have currently. Uh, myb also writes like me, so all of the content [8:56] that it puts out. Uh, it takes my voice. So, when I set up Grockbot, I had it scan my Gmail. I had it scan my Slack messages. So, now everything that uh I like all the emails that it writes, it's in my voice. I think a problem I had with other AI tools as well is like it was like a bunch of M dashes. like I would have to spend more time like rewriting something that AI gave to me uh rather than uh you know just writing it myself the first time. So I couldn't really tell if I was being like more productive or not. Uh I just felt like I [9:33] was like in this like prompting loop instead of just uh writing the email myself. Um so this is my chief of staff. He can message or she can message other uh bots as well. So, for example, I had this um task. I was like, "What's trending on X today? Can you put a slide together?" So, it messaged my Xbot. You can actually see the conversation here, which is kind of funny. I don't like tend to look at this, but you can see when it messages your other bots. So, it pulled everything that's trending on X today. And then it handed that to my slides designer. and my slides agent has [10:10] access to uh like the like kind of design schema like our brand guidelines and it like remembers that over time and remembers my preferences. So it worked with this like slides designer to put together this deck for me. Um I think this is really helpful. I think all of us like have to stay up to date on what's happening with AI. I think the place to do that with is X. [laughter] if you're on like LinkedIn or even like watching the news like it's like topics that like happened two days ago or even weeks ago. Uh so I think it's funny to like always see the lag of like you know [10:44] you see things happen on X first and then you see it hit LinkedIn uh [laughter] much later. Uh so if you're in go go to market you're not on X uh like you don't need to post or anything but definitely like to keep up with the trends. I'm always asked about things outside of SpaceX AI. Um maybe it's like Nvidia buying hug hugging face or something. Uh it's good to stay up todate and be more of a um you know consultant uh when you're working with [11:11] your customers. Uh bot again it has its computers has a computer so it can watch uh content. I had it watch uh Amplitude's last product launch and I could use that to be like, "Okay, go and watch uh and you watch Decagon's latest launch video on YouTube." Um give me the rundown. Uh so I think I uh the first time I started using graphbot it was still an internal product and why it was so successful for me at prospecting was it wasn't just like finding like general company news or like okay like this company has x over xyear growth. Uh like yes it finds those things but I think [12:01] what uh works best is like those personal hooks. Like if I was like, I don't know, a sales rep reaching out to Brendan, I would have my bot like scan his ex feed, watch any YouTube videos associated with him, and then personalize outreach based on his personal content um rather than, you know, general company uh content. So you can do this at a large scale. I'm just kind of showing uh some like examples here, but when I have like a large book of business to reach out to, I will uh give my bot like a list of companies. Uh and then usually I'm like pretty like [12:41] generous with it. Like I have this like PG skill that's basically it builds like a uh it pulls your accounts from Salesforce. It then creates a Google sheet with intent data. So, it'll pull the like top accounts to reach out to. I do five at a time, but you can do much more. Uh, it'll choose five contacts from that account. Uh, I like, you know, tell it to pull like leadership teams or people with like warmer content. It'll [13:07] create this spreadsheet. Um, I don't I think I'm sharing the wrong screen, but it it creates this uh spreadsheet. I'll ask you pull this up in your computer. uh filled with all the intent data. So it'll look at job postings. It'll look at uh you know maybe you want to look at engineering count growth or um pull like context contacts that are the warmest uh [13:36] and then it'll go watch YouTube videos. It'll go wa or listen to podcasts. Like those are all the things that I wish I had the time to do. Uh but I I simply don't. Um but your bot will kind of do that for you. Um, and I think just giving your bot like a database just [13:55] helps it stay more organized. Uh, something I do, I'll let that load for a second, but I also have like a few strategic customers and I have customer experts per customers and I run a routine uh, like a weekly content routine. Um, and it'll essentially every week, every Monday morning, it'll go watch YouTube videos, look at blog posts, and just give me like a rundown on the account. Um, and any leadership um, and that will just help me stay up to date. I also have I I have like Slack channels with all my customers, so it'll like listen to the Slack channels. Any feature requests that come up in those [14:37] channels, it'll let me know. And then when our change log drops, it'll be like, "Hey, uh, you know, Candle Wick Labs pinged you a month or two ago. They uh were interested in this feature. It just launched today. Like here's an email to reach out to them." So, it's just like so like dialed in on the customer. Uh, instead of like me having to like remember that this customer asked me about this feature two months ago, it will bring it up for me. Uh it can even track like linear tickets if you want to like if there's like any bugs that they uh you know flagged and [15:10] that the team was working on. So it can tell me when those were resolved. Uh it'll monitor usage. So for this one I have it monitoring like uh CLI users um uh any like next steps. I also have like a database like notion document. You can also share these bots. So I can share these after the call. Um, let me pull up my notion really quick. I'll [15:35] stop sharing. While you pull that up, I just wanted to interject. Um, there's also, yeah, people are asking if you can watch videos. Yes, you can. I I, for in, for example, I was watching a video the other night. Uh, and people were talking about like movies throughout it. It was like an hour and a half long thing. And [15:49] I just had bot watch the the interview. I just dropped the link in. And then I had it also give me stackranked on their Rotten Tomato scores. So then I had this list because I kept thinking like, "Oh, I should go watch that video, that movie." Um, and so then there was like 17 awesome movie recommendations. So it definitely can and like you can plug in [16:08] other things to do too. Yeah. I think like how I first thought of it is like I created this EG plan at first and it gave me the links and I was like, "Okay, that's great, but I'm not going to like go watch a 45minut interview." So I was like, "Okay, just go watch those podcasts or videos." So that's why I like say you need to think bigger. like even if the agent gives you like a link, just be like, "Okay, like go watch that." And I think I told my [16:31] bot like, "Never give me a link again. Just like go and watch it for me." Um, but this is an an account plan I created. It'll auto update after every call. Um, and I can update it via Grock, uh, via my Grockbot. Uh, something like kind of interesting is like I don't really use any of my goto market tools anymore. I have bot use them for me. So, I tend to choose like tools like granola or tools like notion that work really well with agents. Like I never update this manually in Notion. My bot just like does it for me. It'll like look at stakeholders. Uh it'll prep agendas for [17:09] me. Um look at usage, any like open Slack questions, uh risks, uh meeting logs. So you can just create routines to update this. And this is kind of my knowledge base of my customer and is helpful for me to like share internally. If you know maybe like I have a leader that's meeting with their CEO, I can give them this account plan and help uh prep for the call. Um [17:38] a few things to finish up here. I can also book travel. So I have Non hooked up to this. I like to ask like can you find a flight uh with Starlink? Uh which happens pretty often. And then Echo is my bot where uh I have Granola, so I use Granola um as well as Gong, but I can actually have it like create a deck live. So if you're talking to a customer about like a mutual success plan or maybe just like recapping at the end of discovery, it can uh update the deck in the middle of the call and then you have the deck to share at the end of [18:16] the call. Be like, "Okay, this is what we talked about. these are your pain points and these are next steps and you can share that live. I think that's kind of like a wow moment for customers. But you can even have this like run uh on a routine after every call. So here's kind of like the output when it's done. But this like was some of the discovery notes that they talked about. Um I can even be like uh translate this deck. I work with a lot of customers in Japan. Uh, so you can like [18:46] translate this stuff to Japanese. Um, I think that's about it. This is kind of like my bot that watches videos for me. Um, and this is my Xbot. So again, uh, helps me just like stay up to date on what's happening on X. Uh, and you can incorporate this research into your customers and like scan CEO's X accounts. But I think I finished right [19:10] on time. Amazing. Yeah, this is awesome. Um, yeah. The the other thing in my experience is like this. It it seems to just keep going. Like it's very tenacious. It's sort of like a vibe thing, but in my experience, it's like I'll give it a task and it'll just keep going until it does the thing. Uh, and yeah, it's like sort of cheesy, but I texted a friend. I was like, this is it's sort of broken my mental model of like instead of using AI as like a to-do list, it's like a to done list. It's a it's a done list. It's like instead of giving it like I want help with these [19:41] things, help me prioritize it or do research on it, it's like can actually just go do the things in many cases. Um, and I think the virtual machine is like one of the big things that unlocks that. Uh, and yeah, it's like kind of digital co-workers as well. Feels like Slack or iMessage. The mobile app is magical. Uh, [19:59] it's just like very very easy to use. So, um, yeah, it's very cool. Um, amazing, Christa. Thank you. This is awesome. Of course, I can try to answer some questions in the chat, but I will, you know, let you move on to the next topic. Okay, cool. Yeah, if uh if feel free to to chime in with any uh or or hop in the chat if there's anything that you want to answer uh in your experience. Um the other thing I forgot we were going to talk about this, Kristo um there are templates, right? So, this is this is uh like you can create a [20:29] template. Did you share any or do you have any you can share here? Drop them in the chat. Amazing. Give me a second. Um okay so bot bot creating a bot is very simple. Uh again like just you know if you sign up you'll see but um templates basically just you can like Crystal will share one as an example. Um you can just take it and use it as your own. Uh so the other thing I was going to ask is if anybody has done that put them in chat here if you want. Um and then I have uh the SpaceX AI team gave me some codes. Uh, so if you drop a [21:06] template in here for go to market, I'll send you a code uh for some credits for Grockbot. Um, but yeah, templates are a good way to sort of like just get up to speed very quickly. All right, sweet. Yeah, Christa put in the Echobbot. Amazing. Was that right? Did I say this right, Christa? Around templates. Okay. Yes. Yes. Yeah. I'll drop mine, but feel [21:30] free to share yours, too. Amazing. Awesome. Um, thank you, Christa. I appreciate it. Eric, what's going on, man? Thank you for being here. Hey, thanks for having me, Brandon. Um, I I like that we were also on X today talking about how go to market X needs to become a thing. Um, so we're going to be posting everything on X from now on. I'm [laughter] starting to hate LinkedIn. We got to go all in on [21:52] X. I love it. Um, I do I do spend a lot of time on X, but not as much posting. So, we can make that happen. I'm excited for it. Um, all right, Eric, take it away. I, uh, I know you posted something, uh, on some of the use cases which you're going to walk us through. Um, I have a few questions that if we have time, I might grill you on, and then there will likely be some other folks in the chat with questions that I'll pull up and ask [22:18] you as we go. Well, and so, uh, this, if you don't know who I am, my name is Eric Noslowski. I run an agency for the last four years. I was employee number at 11 at Clay uh clay.com and then I went all in on this agency. We've worked with companies as large as Reddit, HubSpot, Intercom, Notion and Instantly.ai and then all the way to smaller startups and all these kind of things. I'm actually the complete opposite of Christa when it comes to go to market. I actually despise enterprise sales and I love really direct response kind of things. And so I've I've dug my hands into all of the AI tools. I've had [22:56] OpenClaw. I've had Hermes. I pay for all of the major subscriptions right now. And I as soon as I saw Grockbot, I was thinking to myself, man, I think this is solved. I think we can go all in on Grockbot and the the Gro SpaceX cursor ecosystem because the amount of problems that they've solved specifically for our business. And I'm going to go over how we did that. I already saw some transition or some questions about how did you transition from Hermes to Grockbot. You literally just ask Grockbot to do it for you. like it's actually so easy to be able to do these things. We're going to go over some [23:27] quick use cases that we found, why we think it's so gamechanging, how we in we doubled the response rate for one of our customers uh when we're sending emails for and then we'll maybe open it up for questions. So, that's kind of my plan. Incredible. By the way, sorry to make you do your own intro. I should have said this. Uh but I I just put this in the chat. I think you're the number one go to number one outbound person in the world. So, uh everyone should follow Eric. you're a legend and you graciously share lots and lots of [23:54] content that is very very high quality. So I appreciate you man. I try. Thank you. I appreciate that. And so the first thing that I like to always go over when there is a new technology is we need to think about what are the things that is different from the the past iterations. And so number one it having its own computer is a stroke of genius from the the SpaceX team. So, I have this clunky, super annoying second MacBook for a Hermes agent that's doing one thing that I want to close and get off of Grockbot, and it's just too annoying. And I just have to keep this [24:27] computer plugged in all the time cuz Hermes breaks all the time. I need to be able to log in. I just can't keep it in another part of the house. I have to keep this open for my Hermes agent. And the fact that it has its own computer is awesome. If you've ever stress test the internet on that thing, it's also like a terabyte. I have no clue how they did that. But it having its own computer so that it can uh work even when you're [24:48] disconnected. Great. The second point is the fact it has its own computer but can still work on my Mac. Amazing. So then the the work stream between both is super great on Hermes. Setting up the cron jobs. So Grockbot made this presentation for me and I think it's saying cron without the ceremony was so perfect. I tried to set up a crown job and I literally had to check in six times to see like why it didn't get it started. I've never had that problem with Grockbot. It just it it just runs them perfectly and it's so easy to change the schedule. Number four, which I I think people are not [25:23] talking about enough, especially with Cursor's announcement yesterday that you can run cloud agents on your own computers. You can deploy them on Cloudflare and Verscell and all these other things, but you can turn your own computer into a cloud agent. the fact that you can push off heavy lifting of big tasks to cursor cloud agents especially in go to market I think composer 2.5 is the most slept on model because of the speed and the uh efficiency on cost of the model if you have skill files to do go to market tasks for you composer is like all you you don't need frontier level intelligence to [25:59] write you know some emails to customers like composer as long as it has the skill files backing it up is just so Great. We also installed for all of my employees, we installed the Grock build CLI into their Grockbot computer. So really heavy lifting. It could also use the Grock build CLI right there. And then the way that you can securely hand off the API keys with that little UI thing that they did that maybe we'll we'll show when we get into Grockbot. I just thought on Hermes I was just like literally holding my breath giving it API keys that I thought I was I was just [26:28] going to like destroy my my security because of it. So those are this top six changes that I personally love and it changes how we've basically built everything into the business. And so I'm just going to go through this. So when we are running outbound campaigns for people and and when you're running any kind of go to market motion you need to be tracking the daily data of the campaign whether it is organic, it's paid media, it's outbound, whatever you're doing. There are numbers that you need to be looking at to make sure that the inputs are correct. You need to be making sure that you're hitting your [27:01] KPIs internally and then you need to have a gro or something needs to be able to run and fix when anything is having a problem. So the customer that we're going to go over today, we generate literally 100red leads every single day for this customer and we've been doing that for a year now. And because we've been doing it for a year, it is completely a math equation. And so then this was the one of the first things that I had my Grockbot take over. And it literally has to do four things. Every day it checks and it makes sure does the campaign have at minimum 100,000 leads [27:34] loaded into the campaign at any given time. Number two, at 12 noon, it does a send check to just make sure it's sending enough emails by noon. If it didn't send enough emails by noon, it automatically refills the campaign uh with more leads. It's constantly checking the deliverability every single day to see if there's any crazy drop off on the um inboxes that we need to be aware of. And then every Friday it looks at any domain that sent at least 200 emails. And then it cancels any domain with less than a 7% reply rate and then actually automatically buys them as well too once they did the the link [28:09] integration. I just gave it credit cards and it's just it's it's completely buying everything for me as well now too. And so um in Grockbot so my Grockbot and I'll show my Grockbot. I just think it's kind of boring to look at the bot for some of these things. So, my Grockbot for the lead floor, you'll see I have this database in Subabase that's updated all the time with all of the leads and everything that we have in here. You'll see at the bottom we have 950,000 leads ready to go at any given time for this uh for this customer. My Grockbot every day connects to this [28:39] superbase and connects to the Clay CLI and if we do not have 100,000 leads pumps this Clay CLI workflow which in this workflow every single email that we are sending is personalized. Every sentence of the email is personalized based on how long the company's been in business, what the company description is, how much news and articles are written about the company today in major outlets. And then we're also personalizing with like finding other contacts at the company and seeing if we should get a referral to them and all these other things. This was also completely set up by Grockbot as well too. I have a demo video for that at the [29:15] end. Um and yeah, and we literally just have our bots doing all of this set up for us and I'll show you the the bots that we have when I kind of get through this. Um, the one thing that I also think a lot of people are sleeping on in this like new go to market engineering world of AI tools is when people talk about go to market engineering a lot of times they're talking about like this stuff where it's like do we have enough leads in the campaign? Do we need to set up the campaigns etc etc. We've also used Grockbot to build completely custom [29:46] audits for some of our other customers. So, this customer is an accessibility um ADA compliance company and we have created with Grockbot, we've created 5,000 custom pages for this customer where we have completely done an accessibility audit of their website. And not only have we done the accessibility audit to tell them all of the problems that they have in their website, we've even screenshotted where the problems are and how they could literally fix the code. and Grock, well, I guess technically Grock build did this in the CLI and Grockbot orchestrated it, but this is also like you're not only setting up campaigns and just running things when you're using Grockbot, you [30:24] can also make highly tailored uh audits for customers as well too. And and this is completely all done with Grockbot. And I'll show the Grockbots in a second. Uh it starts with the workers. I'll show that in a little bit. The secrets. Okay. And so what it did to to double the reply rate for this customer is Grockbot just every day is pulling the smart lead API and quite literally pulling all of the analytics of all the domains and finding all of the replies and then it's killing anything that's below like I said that 7% thing and literally Grockbot came out it came out on a Tuesday. You see it's dated Tuesday [31:03] August 11th and then by Wednesday we completely fixed everything. And so, um, I don't know, this is just some real life of this actually working. And then we use it to build things in the Clay CLI, which would take too long. So, I just have this YouTube video, um, pulled up that we're just going to go through. But here, I literally just asked it, "Hey, could you build something for me [31:21] in the Clay CLI?" Let's see. And see, this is my prompt. Literally, you can see this is not a long prompt by any means. And then Grockbot just started getting to work. It said, "The live workflow is up." And then it literally just started doing everything for me. Adding all of the nodes, putting it all together. I didn't have to touch anything. And it's literally just making all of this for me. So, we're also setting up all of [31:48] our campaigns in Grockbot as well, too. Do I need to pause? Was it correct? Oh, it was perfect. It one-shotted it. It oneshotted the entire thing. Oh, yeah. We We also live like my whole company has skill files on skill files on skill files on how I like these campaigns to be set up. But yeah, it oneshotted the whole thing. I think the only problem it it maybe did is in this uh in this node right here. I think it used like an expensive model that was overkill and we could have done something simpler, but that was that was [32:15] the main thing. Um and then let me see. I'll show you my Grockbot setup. I don't really think my setup matters that much, but yes, that's basically the end of the presentation here. My setup looks specifically like this. So I also didn't want to waste any time. So I even pulled up and I said, "Hey, show me that workflow of the overview that we did." And see, it literally is just it it reme and that's the other thing too. It's memory is so great. This happened 3 weeks ago, right? And it still remembers, oh yeah, we set this CLI in workspace 5561. If anyone has a [32:52] clay workspace lower than 5561, show me. Uh because I usually win that contest. But then we set up this table. We It set up all the CLI. It did everything for us. It has the real company name, the description, our article angle, the employee line, everything got completely done. A couple of other things, we have um a bot that completely knows how to audit smart lead replies and it knows [33:16] everything about the smart lead API. I've set up um customer literally like one thing that I think is great is you can set up bots for particular customers. So, one customer like this customer gets anywhere from 80 to 100 leads a day. another customer might get two leads a day. So I literally have this one bot for a customer that was having deliverability problems that it is trained purposely on just like for this customer these are the problems that we see for them. I want you to check these numbers every day. I want you to check these numbers every day and then give me an output of how this [33:47] customer is doing. So not only can we do systemwide things but for particular customers we can check on things. I have a Clay CLI expert that knows exactly how the Clay CLI works. We have a copywriting expert. We have a campaign strategy expert. I have an email deliverability expert. Um, I don't know how to pronounce their name, but Potato on X shared this Dr. Eggbot thing that I added as well too that I highly [34:08] recommend everyone joins in on as well. And so, uh, I mean, yeah, this is this is basically our setup. I guess I can answer any other questions because I I have a quick question for you. Can you like double click into one of these? So, like the copyrightiting one, I think this is another thing that that I'm still sort of trying to figure out myself is like, okay, there's not this concept of skills. So like a copywriting you're going to build in your own skills. How do you think about that? How do you kind of like work with think about this zero and I think people [34:34] over complicate it. So what I also love about Grockbot was that every single bot is sitting on every skill within the business. It's just great that you can have this orchestration that I talked to my taskmaster. I guess I missed the memo that I should have used my chief of staff, but I think I just like bulldozed right by him when I set this up. But I set up a taskmaster bot that it just knows, hey, go to this bot when you have a uh when you have a deliverability thing, go to this bot when we're setting up the clay CLI, etc., etc. So, what I [35:06] do love about them is that all of them are sitting on top of all of your skills. I think in my company, we have 352 skills the last time I looked at it. They are sitting on top of all of your skills, but then it's so easy to just specialize it. And literally the way I write these, which I think I've I think I stole this from Alli Miller. The way I write these is I will say uh hey Grockbot, like we need to make a [35:31] specialized agent for the clay CLI. Learn everything you know about the clay CLI, etc., etc. then it'll make that and then I'll say, "Okay, now stress test what you think you know about the Clay CLI and come up with 10 um examples of real world scenarios we're going to have to get through based on the past work that we've done and the things that you see in my GitHub repository. And then based on that, run those stress tests and make sure you know how to navigate them and let me know if you have any questions and then I'm done. I don't do anything else for the the description of [36:02] the bot from there." And it will sort of like maintain that over time as well. Or does that do you find that that deprecates? I guess it's been I don't think about the maintenance. I don't really think about the maintenance. I sometimes every week I will tell Grockbot make a markdown file of of everything that our setup does and then I'll have Grock 4.6 audit it to see if there's any redundancies and things that we're, you know, crossing paths with and and I definitely recommend that. But as far as me actually keeping up with things, it just does it all. And you have to remember like these the way [36:32] I think about it is like rock 4.6 is made to code enterprise software. I am sending emails to people. Like it's not that it's like really not that hard. Uh in comparison to the the intelligence that it has. And so uh yeah, I just let it just audit all of its own things. It looks at my calls. Oh, well, another thing that we did that I highly recommend everybody does is I So, we're doing all of our work today through Grockbot as the surface and then we get, you know, we send off cursor cloud agents or Grock build or whatever it's [37:05] going to be. Keeping this like multiplayer situation across the team is really difficult cuz if I finish something and then it's on my Grockbot or it's it's saved on my computer or something, that gets really hard. So, we did make a ClickUp skill that forces any finished products to be pushed to Google Drive or to be pushed to Slack or to be pushed to Superbase or whatever the output of the finished product is. And then we use the ClickUp plugin to then like literally if there's no task associated with what we're trying to do, it'll make a task and then it'll literally run through like, hey, [37:39] this was the plan that I started with. Here's all the work that I did and all the updates that I have. And then as it finds roadblocks, it'll log that roadblock and how it fixed that roadblock. And then the finished product gets attached to it and then it gets moved to like in review inside of [37:52] ClickUp. That way we solve two problems. One, anybody on the team can then look at like, oh, on August 20th, Eric did this work. Here was the finished product of that work. Now I can just pick up from where he left off. that then two the benefit you have of doing it in ClickUp and not GitHub is like anybody on my team can read ClickUp comments but it's really kind of annoying to dig through GitHub if you're not a technical person and so uh that is also how we're maintaining work across the team and like having until Grockbot allows for this like multiplayer like my bots can [38:25] talk to other people's bots that's how we're like solving for it right now. Yeah. And one of the the questions here is do the bots know to work with each other? Yes. The answer is for sure. Like if I give it a task, I have a chief of staff and so I I'll ask my chief of staff to do something and then it'll sort of delegate to one or multiple bots. It's kind of specialized bots. So again, I I think about this as like digital co-workers. I don't know if that's the right way to think about it, [38:47] but that's like sort of my mental model. Um, I do think that it'll be wild and interesting if uh at some point like I can ask a bot something and then loop in you, Eric, and you can interact with that bot and your bot can interact with my bot, you know, if we're on the same team, for instance, or not. Um, that'll be really crazy. Um, but today, yeah, [39:07] totally. The thing that I love about all these systems is like don't wait for the SpaceX team to put that together. Like that exist like that you can make that happen because the plugins are already so good and because you have all of your bots and you can share templates and all these other things. They even just added the ability to send a web hook to kick off a bot as well too. So like don't wait for that to happen. We're doing all [39:27] of it in ClickUp today right now. Yeah. Amazing. Okay. I had one question for you. This is sort of like a meta question, Eric. I know like you and I talked maybe a year ago when you were in Cursor. I don't know. You were like buying G renting GPUs, doing crazy stuff. You're like obviously on the [39:41] bleeding edge of a lot of this stuff. How do you think about and for me like I did like I I never messed with open call. I was like I'm sure I could figure it out, but like I I didn't go that deep. Uh you know, and so how do you think about but Grockbots to me is like okay this is capture me and I'm like okay this is very very interesting. How do you think about when new things pop up? like what is how do you kind of decide how w which new tools you use and even within Grockbot like are you going to build something on top of it or you [40:12] going to wait for them to build something like what is your mental model as you actually so even the conversation that we had that like a year ago was you were like hey like everyone's talking about MCPS why aren't you doing anything with MCPS or whatever that was and I was like look like the models just aren't intelligent enough that it's it's not pushing the boundaries on anything so I I I highly recommend everybody needs to create an internal benchmark for getting work done in their business because like these these companies, they can say anything they want about these benchmarks. You have no clue what's actually going on [40:41] with these with these benchmarks. So when something new comes out, how do you know that it actually can help you? And so our benchmarks are one. So in our business, we need to be able to order inboxes. We need to be able to set up clay tables. We need to be able to make really sophisticated lead lists. And we need to be able to set up simple [41:00] campaigns and very difficult campaigns. So I have basically three scenarios for all of those. So like having an AI on the current skills that we have and the current data tools that we have build a list of peptide manufacturers is like my favorite list building test because building a list of peptide manufacturers is impossible with like current filters, right? And so one, you need to have internal benchmarks for yourself so that you actually know when something's worth it. when something drops I I mean we also pay for everything so like I do just like it like I was just already on the cursor team subscription so it was [41:35] just easy for us to try I like I started the presentation with I always start with what is the current tech that we're using right now and what has changed about this and like does it solve any of the current problems that I have so for example uh the the situation of like my computer always has to be on and it always has to be connected to the internet. Where I live in New Jersey, my power goes out once a month and then my computer's down. Now, doesn't matter cuz my Grockbot is just connected to the internet and I'm all good to go. And it solved that problem for me. Also, having [42:08] the experience with Hermes and knowing how hard it was to set up cron jobs, the minute the cron job worked correctly to pick up my Fathom call, I I was almost sold there. I also think like the another gamechanging thing is the idea of having multiple bots. I'm not sure if that's like maybe overkill and and leading into some of the usage problems that people talk about online. Um, but it was truly novel to be able to have like yes, this bot is trained on all of your skills because Hermes you would make a like a a second Hermes and then it was like trained on that one [42:39] particular skill. It is nice that I could go ask my Gmail assistant bot to go build a clay table because it still has the knowledge of the rest of rest of the company. But the way I could just go to one taskmaster and it just passes out all of the tasks and we know that it's getting done versus like in other tools you just had to kind of hope that everything was getting done. But now you have one specialized agent. The other thing too is um Grockbot just has a really good sense for what the finish line is as well too. So then we also can [43:08] just set it and forget it to just do things because it just finishes tasks better. Actually, I'm not sure. I I I prepped this. I'm not really sure like if I can share something from Codeex, but like this is I mean it's a bad thing against Codex, so I'm just going to share it. But like I was while you pull this up, I also have found that like when I ask Grockbot to do a task and it it's like, you know, has access to to certain tools or like Fathom call recordings or something, it will also then loop in like, okay, [43:37] here's the thing. I just did it for you. By the way, there's a thing that I saw you're supposed to email this person and you haven't emailed them. You might want to follow up with them. Like that literally happened two days ago. I was like, "Whoa, yeah, you're right. I didn't follow up with that person." Yeah. Which is Well, so here in Codeex, see how like I was like, "All right, can we add leads to this campaign?" And it was like, "Yes, we added 1500 leads and we deleted the 9,000 leads and so we're ready to put more leads into the campaign." I was like, "Okay, but how can we put leads [44:02] into the campaign?" And it was like, "Oh, we we can totally do this for you." I'm like, "All right, so let's add them in." Like I just I like had it was a super simple task that I had to like keep pushing it to do. Um, and I've never had that problem with Crockpot. It's like run. It's finished condition [44:16] is way better. Yeah, it's so tenacious. It just doesn't stop. Tenacious is a good way to put it that doesn't stop. It just keeps going. It's a bulldozer. Um, amazing. Okay. A couple people, Eric, I don't know if you're comfortable sharing. A couple people have asked like how much you spend. Um, is that something [44:32] as a company or me personally? Uh, they didn't specify. Um, so it kind of depends. I'll totally share. I have uh the whole team has a cursor team plan. I have the Grock build super Grock heavy. Um, which like while Christa's here, nobody's been able to switch my account from cursor team usage to super gro heavy. [44:52] That'd be great. Emailing me. [laughter] Yeah, actually I DM'd you too. It's like I I'm like what? Like why can I not like switch my usage pool? Cuz I like Yeah. Anyway, we Anyway, so then we then have the Claude subscriptions which we're pretty much looking at canceling Claude by this point. And then we have codec subscriptions for everybody. So I don't know, I think on just subscriptions maybe we're spending anywhere from $3,000 to $4,000 a month on that. And then on the inference for like the copywriting for the emails because all of them are so personalized, we spend like $6 to $8,000 [45:26] a month on on that. Amazing. Um and then are you also like monitoring? Um, again, I just think about like maybe this is like a bad fear that I have, but like some sort of like tech debt or bloat. Just like me personally, like, okay, I'm going to spin up a new bot. Do you have a bot to clean up for you once [45:45] a month? You just don't think about it. It's like whatever deprecates over time, it happens. So, the other thing too that you have to think about is these tools have context windows into like the 500,000 tokens to a million tokens. A million tokens is the Bible. Do you really think that you have more context than the Bible? I just don't think so. I So, I just I just literally [46:08] just don't think about this at all. Yeah. Nice. Um, amazing. Um, let's see if there's any other questions that people have. Um, yeah, this is great. Um, amazing. Uh, cool. Anything else? Oh, this is a good one. Either of you feel free to answer. How do you think about like channels versus like going directly to a [46:40] bot? Eric, do you use channels? I never use channels. I talk to my taskmaster for everything. I I like actually don't use the taskmaster that much or like a chief of staff. Sometimes I'll use the chief of staff if it's like a random question and I don't I'm not sure where to fit or I'm just like I don't know want to quickly ask a question and I know it'll reach out to another bot. But I do like working with the like teammate bots like the specialists. Uh I'll use like the chief of staff once in a while to like orchestrate, but some people I've seen uh like I went to a demo Monday, [47:20] somebody has like a they run like a big meeting between their bots like every day, like a rundown and I guess uh they all talk to each other and make sure they're like staying on their it's like a daily standup, but he's not involved. It's just the bots like running the standup. and he said it keeps everyone on task and like he gives them like what their job is and what they need to report back. So, people have gotten really creative, but I think I'm pretty [47:49] simple. That's wild. Yeah, daily standup across your digital co-workers. Uh yeah, just report back even though that's wild. Uh yeah, I've done something that's just like what is what are the top two or three things that I should be focusing on today? Like make sure that I you don't you know that I don't drop any balls. um because they have like all this context now. It's it's kind of crazy. Um yeah, can it be counterproductive to create bots with slightly overlapping functions? I agree. That one's sort of like I have a research bot and then like an editing bot. How do you think about this? You don't think about it, Eric, [48:22] because you're just talking to your the the main. Yeah, I really because it like remember because like I literally could go to my Gmail assistant bot and have it create tables like so I think the only Okay. I like it's inefficient because I think and Kristen knows this, but I think you're limited to 50 bots, right? [48:41] Yeah. So maybe that's the inefficiency. Yeah. Yeah. So I guess you're inefficient by that. But literally any any question you have about these things. That's what I always remind people about AI. It's like if you have a question and you think you're being inefficient, the perfect person to ask is Grockbot. Like like you [48:56] could just go ask it itself. Yeah. Uh Eric, somebody asked uh Tom, what's up, Tom? Are you using Grockbot for your own crawlers or something else? So yesterday I needed to build a list of every used car dealership in the United States and then get a custom site map of their inventory page and then uh we needed to get the average price and the count of the amount of cars that they had on their inventory page. So we went to Grockbot and we said, you know, hey, this is this is the goal. Use this API to to crawl Google maps. Um, there's something on something on rapid API [49:29] called maps data that if you want to crawl Google maps is gigantically cheap in my repo called outbound skills. It's a public open- source one and I have every zip code in the United States. So, it just connected to that and just crawled the United States for used car dealerships. Then uh it so it did that and it saved it on the computer and then I sent it over to Grock build where I was like okay obviously the easiest way to do this is check the sitemap and see if there's any obvious like usecard dealer.com/inventory used car dealer back slash used cars or whatever and then it did all those site [50:02] maps. Then Grock build scraped whatever it could and then we passed it to firecrawl for whatever it still couldn't get past because then it had like custom scraping and we just should just download the whole website at that point. Um so yeah that that's like a perfect example of something we did there. Amazing. What is build blasphemy? What do you [50:22] mean? That's just their version of like the clawed codec. Um, okay. Maybe we have time for one or two more questions. As you run more go to market motions through Grock bots, how do you decide which learning should persist and change the next decision versus staying specific to one campaign account or moment in time? [50:46] Chris or me? Yeah, Eric, what do you think? Oh, sure. H, you want to be really careful. So, we have built a bot that I talked actually with the sell better guys. Shout out to the Heathan on you. I saw Chris was in here too. We did something yesterday yesterday about using AI to fully autonomously [51:04] take over your uh your campaigns. I would still say that when you're using AI to optimize any marketing campaigns, you still want to check what its suggestion is going to be. There's a marketer, and I forget who exactly said this, but there's a marketer that like that said if if we overoptimize every website, every website would turn into an adult film website essentially if we like overoptimized on conversion. And so we still will have it suggest the changes for conversion and then we make the So what we presented on yesterday is we're currently some sub affiliates for some data for companies that are procuring business data to then sell to [51:45] the frontier labs and like it went through and it suggested like all these ways that we can improve the campaign and we check every way that it wants to improve that campaign and then say yes it's a good idea that we should reach out to this company about their animated film content because they're movie studios, etc. But this other idea of like asking plumbers for all of their [52:07] work orders is like, don't even try. Like, they're not going to have it. It's probably not digitized. How many of these guys understand AI enough that they're even going to want to hand it over? Like, yeah, kill that campaign, right? So, when you're optimizing marketing with AI, you definitely should be because it can come up with way [52:21] better ideas than you can, way faster. But, you still just want to check them to make sure you're locked in. Yeah. Um, okay. Somebody earlier was asking, are is this the singularity? Are we being replaced? I'm curious, Eric, in this moment in time, how you would answer this kind of idea that we keep thinking about. I do think it's, you know, sort of like a meme at this point, but like is AI going to replace sales people is like kind of pedantic, but how [52:43] do you think about this? Right. So, here's one thing. I've never worked more in my life because it's like always at any given time like my wife has started to watch my wife is a genius. She can speak Albanian, Greek, English, and Spanish. and she just like watches these Spanish nollas like and I don't I don't understand anything that's going on. So then I'll just be sitting there have an idea and then just like whisper to my Grockbot and then just keep working while she's watching these nollas. So, on one side, I've never worked hard in my life, but I will say that we were keeping underperforming [53:12] team members on the team for a long time because if there was an emergency and I needed a lot of hands to just attack the emergency, I needed them to to be on the team. And uh we let them go because these tools replace all of that button pushing and things like that. And so we've definitely seen it reduce our headcount and our and we're pushing more campaigns with more emails with more customization with more segmentation than we ever have before. But I just feel like it's just filling the vacuum of like well now that we could have now that we have more time back we're just [53:45] going to do more overall work instead. So I don't really see it yet. But who like even Elon's I I think Elon's um thought process on this from the Joe Rogan podcast like you should see it. He was like like when you think about it like if we had unlimited robots with unlimited intelligence and unlimited energy sounds like actually a pretty good world. So I'm not I'm not really that [54:06] worried about it. Yeah. Yeah. I think that's a good uh for what it's worth technoptimist take. Uh I'm the same for what it's worth. Like I am working way more and feel very exhausted and very energized and excited by AI but it's a lot because there's new stuff that's happening every day. Uh and keeping up is tough. Um, [54:27] okay. Uh, what's a better subscription to get, Super Grock or Cursor? I don't know if you can answer that, Christo. I don't understand the differences. All of the like different companies subscriptions is confusing to me. Yeah. Didn't they post though that like Cursor Ultra has the same usage pool as Super Gro Heavy or something like isn't it all kind of make all these simpler? Uh, yeah, of course. Uh, you know, I think there's like a bunch of applications. So, I think our team will work on making all these subscriptions simpler. uh scrobbot [54:56] for the latest there. Um all right, Christa, anything else before we close out here? Thank you again for being here. Thanks for jumping into the chat as well. I see you. Yeah, thanks everyone for joining. Um definitely share all your templates on X. Um would be great to see what everyone has built so you can share with [55:13] others. Yeah, please tag me Christa Eric on X. Uh when you share those, I'm also trying to steal your best templates. Uh so I'll do the same. Um, Eric, always fun to have you. Thanks for hanging out. And Christa, thanks for making the time. I appreciate you all. And we'll see you next time. [55:32] Bye, everyone. Okay. See you. --- About this transcript Read from YouTube's own caption track and laid out by ViewRank AI (https://viewrankai.com). ViewRank AI finds the videos already beating a creator's own average on Instagram, TikTok and YouTube Shorts, transcribes them from the audio itself in more than 60 languages, and turns what worked into new ideas and scripts. Free transcript tools, no account needed: https://viewrankai.com/tools How to read any video this way: https://viewrankai.com/llms.txt