She automated hours of her workday | Krista Letz (SpaceXAI)

Greg Kamradt· 16 min· 2,761 words· 13 min read· English ·Watch on YouTube

This is the full transcript of She automated hours of her workday | Krista Letz (SpaceXAI), published on YouTube by Greg Kamradt. Every paragraph carries the moment it was spoken, so you can click any line to jump straight to that point in the video, search the whole thing for a word, or copy it out.

0:00I trust Scrockbot to go off [music] and do autonomous tasks on its own. I'm in calls all day every day. So, what I've kind of set up is a series of routines. [music] Every time I finish a call, it will update the deck of the customer I just talked to. My name is Christa. I'm on the go to market team at SpaceX AI. Was previously at cursor where I was on that early go to market team. Uh so something I know very well is uh high high volume. Uh but thankfully I work with a couple of customers now. Uh I was an early user of Grockbot. We were dog fooding it

0:37internally. So I used it the first day and couldn't live without it. I think actually the first time I used it all my bots crashed. It was really sad and I realized how much it I depended on it. Uh so glad we've made a lot of improvements since then. But I'm just going to walk you through uh my workflows. Uh, again, I think Grockbot is a very personal product. It's a little bit different to how we use most AI tools. I use other AI tools before, but they felt kind of as Matt mentioned, very like synchronous. It's like a thinking partner, but it's actually not going off and doing. Uh, I kind of trust

1:14Scrockbot to go off and do autonomous tasks on its own. I'm in calls all day every day. So, uh, what I've set it kind of set up is a series of routines. Every time I finish a call, it will update the deck of the customer I just talked to or even live on the call if I run it. Uh, it will do routines for, you know, uh, account research. Uh, I can kind of walk you through how it's, uh, helping me stay up to date on customers, uh, and reaching out. So, uh, also plug, I'll use the the new X plugin. uh it's helped me especially with the uh influx of X

1:52traffic we've had. But I think something that's cool is uh Grockbot can actually consume content. So if you want to watch like maybe like a YouTube video um I could you know say uh one it'll like run these on a schedule but find the most recent Amplitude launch video on YouTube. watch the video and give me a recap and it'll go off, watch the video and come back to me. So, I can let that work in the

2:26meantime. I also have a chief of staff. Uh, this is my dog, Olive. Uh, so she does a lot of work for me during the day. Uh, I think some people use the chief of staff to orchestrate other agents, which I do occasionally, but I kind of use it as my admin. Uh, so

2:43it'll, you know, prep me for meetings. it'll monitor my inbox. Uh so it uh you know kind of I can you know jump in like an Uber and then on the way to work it will prep me for my meetings. Uh I also I think this one was pretty popular but when I'm on a call if I run echo it will update the most recent uh granola transcript I have and uh translate those customer notes on the deck. A lot of times I can run it like midway through the call. So if I'm doing discovery, you know, I'm talking about, for example, Amplitude's engineering initiatives, how

3:20they're building an AI software factory, it will update the deck on what their top initiatives are. Uh for this example, it talked about how uh uh this, you know, sample company I talked about, uh that new hires were taking a lot of weeks to ramp. Um they wanted agents that work overnight like Rockbot and they as a next step they wanted 30 minutes with engineering uh actually demoing the product. So uh I can run that live on a call. Um looks like I just ran it now and I can also use that as a follow-up. Uh now we have templates. Uh so if you have an amazing

4:02bot, you can share it with others. But this is an example of how I shared my bot template Echo. Uh it'll answer a series of questions to get that bot ready and personalized for you. And then when it's done, uh you can run that run that bot for you personally. Um and then I can come back. Uh it's watching no launch video. It found it. So I'll wait for it to finish. Uh I have a customer expert. So I work with a lot of customers at uh SpaceXi. And how I work with them is I help them on board cursor and our cloud agents platform. So for

4:39building a autonomous software factory. So similar to how companies use bot uh engineering teams use cloud agents to run overnight and complete large projects. So it'll monitor customers uh slack channels. Uh it will go through next steps. I also have it connected to notion as a database. So I can click into the account plan and uh after every call it'll update this account plan uh based on what we heard on the call um any action items uh just a snapshot about the account why it matters it'll plug all the relevant stakeholders I've been talking to so instead of me having to update this manually maybe pull information from

5:24data bricks or other uh tools it will uh take this data update update it daily for me and update after calls. So these are all things I was doing manually. Uh I know we have a lot of founders in the room and not that many sales and go to market but when you're a founder you're always in sales. Uh so I think this will be really helpful uh for anyone uh starting a company right now. Uh I can go through my PGbot. So I think what's different for me with Grockbot is uh normally when you use AI tools to do research. Uh it's it's kind of looking

6:01up like company initiatives but oftent times the data is really stale or maybe you're purchasing another tool. Uh but with Grockbot I've been able to run this PG plan and it'll automatically research uh however many companies you want to look up whether that's 50 or 100. You can run this overnight, but it'll check public LinkedIn. It'll look at ex posts. It'll actually listen to the podcasts and webinars about the account or about the persona for me. I think when with Grockbot, you have to think a little bit bigger. When I was first using it, I was using it how I use most AI tools. And then I used it for this use case and it

6:40you know it actually linked me all the all the podcasts and the uh webinars but uh I was like I'm not going to watch a 45minute webinar about you know 50 accounts. Uh so I just asked like hey can you watch that for me and you know give me back like what you got out of the uh you know podcast and how SpaceX AI could fit in. And then it actually

7:03just started watching all these for me. Uh so it creates kind of this uh spreadsheet and this is sample data of course but it'll give me like the their LinkedIn or their ex why I should reach out uh compelling event um and I found that personal hooks uh work really well. I think people love talking about themselves. So that's worked really well for me uh with outreach. Um can you adjust the resolution on your

7:32screen to be a little bigger? Oh yeah, really small. Sorry. Can you see it better? Okay, perfect. Uh, yeah. So, for this one, it watched a uh Amplitude's last agent analytics uh product release, gave me a recap uh what products Amplitude is sol is solving. Uh so, it's actually consuming the content. Um I also use uh Grabbot for like booking travel. So, I have it connected to Non. Um, I think something as cool is like I can be like, "Hey, can you find me a flight that has Starlink to, I don't know, play

8:18to San Diego. Look for flights that are likely to have Starlink." Um, and then from like a personal perspective, I also have a bot. We uh love to read at SpaceX. So, we have a library in our office. We also have like a books channel. Uh, so I have it monitoring this books channel. Uh, and then whenever I find something that's relevant, I can add it to cart and then you can um you can have grabbot purchase on your behalf. Um, and I can come back and look at uh the travel uh let's see next week and it'll come back and give me some flights. Uh, but these are some of the

9:09ways I'm using grabbot. Primarily I'm in go to market so my day is spent listening or working with customers. Uh, I need to track you know X and LinkedIn to see what they're focused on and see how SpaceX can help. Uh, but it's automated like hours of my workday. I used to like be on calls every day and uh come back and have to spend hours on prospecting or follow-ups. And now with all the routines I've set up, it will kind of autodraft emails for for me, automatically do customer research. Um, I have it tracking like linear tickets and uh even like um like product requests so that whenever our change log

9:49is released, it'll actually suggest new products that we launched that maybe a customer asked about a month ago. And like before I was kind of keeping track of that in threads of my brain. But um I think bot has just allowed me to like better uh reach my customers and provide more customized uh support and enablement as well as find uh net new uh customers as well which as a founder I'm sure everyone is focused on. But uh thanks so much for your time. Uh let me know if you have any questions and I

10:18will pass it off to the next demo. [applause] Thank you so much, Kristen. We're gonna take three questions. I saw you first.

10:36Is this Mike? It's Mike. [laughter] Hi. I really love the demo. Thank you so much. Uh, in terms of like the usage and your day-to-day, how do you sort of optimize it? So it's like to the best performance in terms of your GTM strategies is one and two uh in terms of reaching out to prospects and everything. Um like I still have some

10:57trust issues. Yeah. So um how much of that are you like capable of automating so that you don't really have to hit on roadblocks on the way? Yeah, I think I'm trusting it more and more. Uh it's uh I still have it draft only, but I think things that help is the first week that you use Grockbot, it's going to be the worst it's ever been because it's still like you're onboarding Grockbot like a co-orker. So what I did was I gave uh Grockbot access to my Gmail, so it scanned my writing style. I also gave it a few drafts I was proud of and uh I don't I think I was

11:32just kind of shocked that when it delivered messaging it uh was exactly how I would write and better in my opinion uh because it could like create these really in-depth emails and pull from like my notion and my Slack and uh even have like an engineer that has access to the code base so it can pull like more technical uh data. So I'm trusting it more and more. I still have it draft only and I just have all the drafts in my Gmail. So whenever uh I have like a moment between meetings, I'll just hit send. Uh but there hasn't uh it's pretty much oneshotted everything for me and I haven't really

12:10made any changes which is shocking to me because I think other AI tools I'd have to like no m dashes till they come to time or there's no smoking gun. Uh, so, uh, because it knows like me and my writing style and each week it just gets better and, uh, it feels like a co-orker I can trust, I think I'm almost ready to, you know, just autonomously send them out. But a lot of people do at uh, SpaceX. I just work with like a select group of customers, so I want to take

12:36ownership. Usage, credits, how do you optimize that? Uh, so routines. Uh I like I say I use routines here, but I don't have routines that like run a couple times daily. Most of my routines I run in the morning uh when I'm at my laptop like preparing for meetings I run at night. So I would say optimize your routines. I also have a chief of staff agent, but I don't use that for all of my threads. So I still like having different specialized agents for different like colleagues. I think of them like an engineer uh you know a slide designer um you know prospecting like a BDR. So I think just like keeping

13:16the threads separate and not just like in this like huge uh cha staff agent where it's like a lot of context and you know tokens. So those are you know some ways but we are optimizing uh planning to optimize token usage in grabbot as well. All right other questions you're too close. Who else? Over here. [laughter]

13:38Thank you. Interesting. Thank you. Um uh very fascinating to see the evolution of your workday. One question that I had was uh with respect to using with LinkedIn um the what is uh I'm very concerned that LinkedIn will block me and uh for me it is really important. So any sense of of how it is behaving when there are

14:07sensitive uh uh activities? Yeah. Uh I haven't had issues. Um but you know I think the how I would think about it is uh you can or you can operate Grockbot like a human. So I don't have Grockbot like I don't know like uh go on LinkedIn and pull up like a hundred different profiles. like I use it very selectively for prospect research. Um but I think there's also like some like browser tools like in plugins you can use too if that would be

14:39helpful. Uh but uh there is an X plugin. I do think most people are on X now. I think everyone in technology, everyone in the room is probably on X. Uh, so I do think that's the new platform and a lot of my prospecting I reach out to people on X or I see what they're posting and I think that's just much more uh personal than uh LinkedIn and I don't know everyone's like shipping their products in X now. So would recommend the X plugin. You get some credits. Uh

15:10but yeah. All right, last question. All right, you got it. Hello. Hi. Um, you mentioned about calendar and then after the call it does something. Can you actually set up triggers or is it just routines that scans your calendars every few minutes or such? I think it's a routine. Maybe there's like a better way for me to set that up. But I think it just like looks at my calendar and every external looks at if it's an external meeting. So I don't want to run the routine if it's an internal meeting. And then after any external meeting it will um like start working on the deck. But sometimes I

15:48trigger it like I'll just like run like echo um and just have it run it for me. But um maybe ways are going to optimize that. Cool. Thanks. That's a feature request if you can. Perfect. We will track that. Yeah. Thanks. I believe so. Give a hand to Christa. Thank you. [applause]

Where these words come from. This is the caption track YouTube holds for this video, written automatically by YouTube rather than by the creator. We read it, tidied the line breaks and laid it out so it can be read. The plain text version is at https://viewrankai.com/tools/youtube-transcript/VXyqyb8giyA.txt.

All rights in this video belong to Greg Kamradt. Watch it on YouTube. If this is your video and you would rather this page did not exist, tell us and we will remove it.