# What's new in Grok Bot #1

Channel: Grok
Video: https://www.youtube.com/watch?v=i-7F2EpQOhc
Duration: 45 min
Language: English
Words: 10601
Transcript page: https://viewrankai.com/tools/youtube-transcript/i-7F2EpQOhc

---

[0:01] Thank you everyone for joining. Let's go ahead and kick off. Uh I'm Roshan and you've heard from Ari and Parker. We're all on the Grockbot team here at SpaceX. Uh we'll let Parker and Ari introduce themselves uh when they kick over to their respective sections. But super excited to be here. Uh when thank you all for joining us. Really, it's just amazing to see how many people are interested in and spending some time with us today. Um I'll start by introducing myself. I'm Roshan. I work on product. Uh and what we wanted to do today is just spend a few minutes going through what's new in Grockbot. And so

[0:29] we thought what we'd do is, you know, walk you through a couple of new features. So each of us will take some time demoing something that's new in Grockbot or walking through some new updates. Um, if you're using Grockbot already, hopefully you've seen some of these things. Um, we'll just show you some of our use cases around them and maybe inspire you with some ideas for how you can bring Grockbot more and more into your own workloads and, you know, spend more time with the tool. Um, maybe find ways to use it in ways that you hadn't thought of before. Um, and really what we want to also do is just create

[0:55] lots of time for some Q&A. And so, um, love to see all the energy in the chat. We'll please keep that energy coming. We'll, um, you know, pick a few questions and get everyone, uh, back on the horn to to go through them. So, jumping in here, um, just wanted to first call out that we, uh, are going to talk a little bit about future upcoming things. And so, just wanted to show you the safe harbor statement about, uh, what might be coming in the future and how to think about those things. And with that, we can go to our agenda. So, we're welcoming right now. We're going

[1:23] to talk about three new Grockbot features today. The first one is proactivity and the primary bot. Uh so bots but stars, which is really cool. Uh team bots, which I I guess I teased a little bit uh that that Ari will walk us through. Um I'll spend a few minutes talking about voice, which I mentioned is one of my favorite features. And then we'll leave some time at the end for some Q&A. Um we'll probably take about a half hour. I don't know if we'll need the full hour today, but excited to to spend the time with you all. And um yeah, I guess Parker, Ari, anything you

[1:52] guys wanted to add before we jump in? Let's do it. Okay, Parker, I think you are up first with uh productivity. So, I think you had some slides you wanted me to share, right? Yeah, let's do it. Um I can give a quick intro. Yeah, let's do that. I'm an engineer working on uh specifically productivity. So, I'm super excited to share this today. Um just kind of talk through a lot of the, you know, use cases we've been seeing and

[2:17] tips to make it better. Cool. And Parker, I guess um for folks who don't know or maybe haven't seen it, what what is proactivity and why did we build it or why did you build it? Yes, this is perfect. Uh yeah, so proactivity um in Primary Bot are these new concepts that we've added to Grockbot, which is um you know, we really want to build the best co-orker, the best teammate that you can have, right? And an important part of that is that you're not having to tell it what to do all of the time. Uh and so we kind of, you know, took a step back and said,

[2:43] what would be the most ideal experience here? and we landed on that Grockbot will reach out to you with, hey, I've drafted this. Hey, I've uh noticed this. Hey, I've prepped some meeting notes, etc. Um, and so we built the ability for Grockbot to do that and kind of wake on its own. Um, so, you know, we have routines and those types of things, but this allows Grockbot to say, "Hey, let me look at everything that you have connected and see what I can pick up for you." Um, and so we kind of we kind of

[3:08] landed here. Love it. Okay, let's walk us through um what's next. Yeah, I'd love to talk about a bit how it works because I think there's been a lot of questions around, you know, what is productivity, how is it different than a normal routine? Um, and you know, just h how can I see what's happening? Uh, and so I think an important thing to call out here is that you don't need routines for productivity to work. Uh, so you know, prior to productivity, you would a lot of people, I think probably most people here have, you know, set up a morning routine or these similar things where Grock bot

[3:38] will go off, it'll look at things and it'll come back to you. And so I think we kind of wanted to make that just the standard experience for everybody. Uh so there's no routines needed. Um and then the other thing here is that we have a bunch of different triggers which determine uh you know when we want to reach out to you. So you know it can be you got an email, you got you know calendar update, uh we just periodically are checking for work, those types of things. Uh and so that is what's kind of kicking off these proactive things. Um and then this is a background job and so

[4:06] we don't charge users for productivity which I think is uh one of the best things about it. So we will provide suggestions to you um you know for no cost uh and we are just spending a lot of time to you know look at everything you have connected uh look at what you've done in the past look at what you've toldot you want to do uh and point things out for you. So I think I have some examples on the next slide

[4:26] version I'd love to show. Oh I love this. Okay we've seen a lot of things on Twitter. So I just I I so I chose a bunch of them here and just kind of like popped them in. Uh one example that I really like is the top left. Uh and this is uh Matt. And I don't know I I should have made it so you could zoom in. But uh this is Matt. And so he had a flight and he had a scheduled Uber. Uh and his flight ended up getting delayed. And so he had to, you know, take a different flight and he hadn't changed his Uber. So

[4:55] Rockbot reached out to him without him doing anything to say, "Hey, you have the scheduled Uber. Why don't you go change this?" Uh, and then, you know, gave a suggestion. And so this is the type of thing where we want it to, you know, feel like you're you're a personal assistant and and the best co-orker you've had. And so, uh, I've got a lot of them here, but, uh, yeah, I love that one. Um, I'm actually traveling to a friend's wedding this weekend and I've had my personal Grockbot reminding me every day about what I need to pack to make sure I don't forget. Uh, and it's like it's great cuz

[5:24] it's one of those things where it's like one less piece of mental burden that I have to be thinking about on my own and and you know like planning or or keeping track of or you know it's like I' I've stopped using my reminders app. Now it just feels like I can do everything with

[5:35] my bots. Exactly. Yeah. And I think you know one of the coolest things about this and I put this on the next slide uh is that you can kind of control what is happening with your proactive suggestion. So I don't know if you've noticed this this week but if you've been packing and it's you know suggesting things and you tell it that you enjoy that it will do that more. Uh and so I think that is one of the coolest things. Um I'd love to go to the next slide and just kind of talk through you know how folks can get the most out of productivity. I think there are two

[6:00] key things uh for me. The first is connectors. So we have, you know, this large marketplace of different plugins that you can add to Rockbot. And so this is things like your email or calendar or Slack, any of those types of things. Uh, and that is what is really empowering productivity to do the most. And so for people that are getting, you know, the most productive notifications, the best time, etc. Um, you know, it's because they give bot access to all of these things. Uh, and so this is probably the number one tip to increase productivity and just, you know, get the best

[6:32] notifications. One of my favorite one of my favorite use cases for productivity uh proactivity I should say at work is meeting prep. I don't know if you get these notifications or these these if you've trained your bot to be a meeting prep bot. Um but I find it amazing I don't know if you have one on the slide to just I go right when I'm getting ready for my meetings. My bot just knows that I have a meeting coming up. It knows the context of what's on my calendar. It knows notes that I've saved from previous meetings and it kind of just gives me this perfect meeting

[7:01] brief. And I find it to be like so powerful because not only is my bot connected to my calendar, but it's connected to like all of our company context, right? So I have notion information and so much more. And it just feels like the richest meeting brief product I've ever used. It's also good at at reminding me of things that I would not have otherwise remembered. Like I for frequently forget like, oh, I said I was going to do XYZ before this meeting. The meeting is coming up in two hours. And my bot will remind me like, hey, you know, last week you said you were going to do this

[7:28] before this meeting. It's in two hours. you might want to try doing it now. And it's usually good at pretty like it's pretty good at checking like, you know, notion or slides or whatever the the source of truth is for where I would actually be doing that work. Uh, and then telling me like, hey, you should

[7:42] probably do this. Yeah, that is a great that is a great use case that I think has been empowering for me because I I am a uh I'm somebody who will get a Slack message and save it for later and often do not come back to it later. And so I think this has been a good thing to be like, hey, you didn't reply to this person and you can come back. So, it's been awesome. Uh,

[8:01] okay. Yeah. And then I just think the other tip here, which which you know, Ro and I was kind of talking about a bit is, you know, bots use your memories and your chat history to inform like what to tell you versus not. So, when we when we are looking for, you know, the things to hand to you, the things we can take off your plate. We're looking at what you've told us that you want us to do. So, uh, I think this is an important thing where uh I think a lot of the time for my meeting prep, I said, "Hey, I want it to

[8:26] be in this format." and every time after that the product notifications, it's in this format and I'm really enjoying that. I think that's a really good experience. U I think Russian, you had something similar where you you have a particular meeting prep format that you like. And uh I do I prefer like context on why we're having the meeting, the goal and the outcome. And I actually sometimes will tell my bot like what do you or ask my bot to say what do you think the outcome of this meeting should be? Because it kind of knows what I'm working on and I've set up my team in a way where I

[8:51] have sort of bots for each of my work streams. Um, and then I I I'm a big uh, you know, the product manager in me is like I must have action items and like know why we're here and what we're going to do next. And so, um, I'm big on like what what are we going to track next? And one cool thing that you were mentioning earlier is that, you know, like bots can also just carry through on those action items, I think. Right. So, it's like it's not just that, you know, you're getting prepped for a meeting, but right after the meeting, I very often will get a summary from bot that

[9:19] says, "Hey, here are the four things that came up. like two of these were message this person in Slack. Do you want me to just do that? And I'll get a card right in bot to just hit send. Um, and so I love that loop of proactivity, but then also tying it to like helping me get my actions done. Um, and I guess Parker, that's what you were saying about bots being really great

[9:38] colleagues, right? Exactly. Exactly. I think a good example of of this specific case is somebody had posted last week that they and I don't know if I included it here but um somebody had a meeting which has a prep dock or a meeting dock attached to the meeting in on the calendar and so every person is you know filling out a piece of that doc and they hadn't done it yet and so it reached out to them and said hey this meeting's in 20 minutes you haven't done this yet here's a suggestion for what to put and they one clicked and their thing was filled out

[10:06] so I think it's it's it's really good at understanding you know the purpose of these things and and how to be a good teammate. Love it. Um well, for folks who are listening, I have not been following the chat. I'll pull it up. But if you have fa your favorite proactivity cases uh that you want to share in the chat, please let us know. Um would love to to chat through some of those and and see what folks are are seeing for their own setups. Uh Parker, anything else you wanted to add? Actually, you know, I had a question for you. So, you you built this feature. I'm curious. Could you

[10:34] tell us a little bit more about how it works under the under the hood and kind of what's actually going on to make these bots really great? cuz I I find this experience to just be so delightful every time it happens. Absolutely. Yeah. I think it's important to kind of talk about the process we took here. Uh you you all will remember that we launched this uh a couple of weeks ago and there was some negative feedback from the group which is hey you know we think it would work better in these ways and so we actually took some iteration cycles to say hey we really

[11:00] want this to land. Um and we changed a bit about how it works where now it's a background job. Now it is you know not taking from your usage and then now it can do specific things like drafting etc you know super obviously and and shows up in your main chat. Uh and then once we had done that we were able to you know relaunch internally and I think the result that we all felt was that it was a lot better. kind of felt that it like met the PMF bar for us uh a bit and so

[11:29] after that we we were able to relaunch. Um so yeah some of the the core like functionality things that we changed were like I said the background job giving it access to more things I think was important at the beginning we had kind of limited what it had access to and now we we give it access to everything. Uh and then the primary bot the number one piece of feedback that we got was that people wanted to be able to customize it. So we have allowed that uh you know folks can make their existing chief of staff, a new chief of staff, whatever their primary bot and and go

[11:56] from there. Um but that that sort of like process I think was what was super important to us feeling good about launching this is that we we got a lot of feedback before we we did it. Yeah, I love that. This this one definitely had a ton of internal back and forth, but like you said, got to some really good internal PMF and I think that was a great sign. Um and if you didn't hear what Parker just said, I'll just mention it again. um proactivity messages do not count against your usage and so this is just something that comes batteries included with Grockbot uh which is really great

[12:24] and so you should rest assured you know like go forth and let your bots be proactive and uh that's just part of the product for us which is which is really cool. Um a lot of great stuff in the chat too so hopefully folks who are watching are also just following along there uh and kind of seeing uh some of some of the other ideas here and getting inspired. I'm seeing some really great examples around bots helping with meetings and um helping yeah prep for for life tasks uh being able few people talking about coaching your primary bot and then how to modify it and teach it

[12:56] to be the most effective colleague for you. And so Parker kind of what you were saying earlier and just kind of training your bots to you know respond in ways that you want use the tools that you want um and rely on every day and just making sure all those things are connected. Exactly. Cool. Love it. This is awesome. I'll hand it back to

[13:12] you. Cool. All right. Well, thanks Parker. That that was proactivity. Um definitely we'll spend some time on that in Q&A if folks have questions. Next, Ari, you're up. We're going to talk a little bit about Team Bots. Yeah. Hello. Hello. Uh my name is Ari. I'm a PM at uh originally at Cursor and now at SpaceX. I work on a combination of Grockbot and Cloud Agents. So, if you've ever used Grockbot for coding tasks, uh what gets kicked off under the hood is actually a cloud agent that does the work. Um, and I'm here to talk about team bots. So, we like to think of Grock

[13:43] bots as kind of these autonomous teammates that you can actually give real work to. Uh, and of course, one important thing about a teammate is that multiple people need to be able to talk to it, right? No teammate, if you have a teammate that only you can talk to, it's not really a teammate. And that's actually why we built team bots. So, the idea of a team bot is that one person can go and create a bot. they can load it up with all the skills and secrets and plugins and you know whatever information that they want to make it really really useful and then everyone

[14:09] at the team can actually DM it. Uh they can actually talk to it in Slack. Uh they can put it in group chats uh and basically work with it like it's a real teammate. And so instead of doing slides, I'm going to do the scariest thing for a PM, which is do a live demo. So we're going to create a Team bot live uh on this on this webinar. We're going to add it to Slack. We're going to talk to it in Slack. And you know, if we're lucky, maybe we'll have Roshan talk to it, too, just to show you that everyone

[14:32] can talk to this. Love it. May the demo gods be graceful. May the demo gods be be on our side uh for this demo. So, here is my Grockbot. What we're going to do actually is go to this little plus menu, and you'll see that there's a section for team bots. Uh you can see a bunch of team bots that we use across SpaceX AI uh in real life. Um but instead, what we're going to do is we're actually going to create a new team bot. And uh while we do this, um the first step of what it's going to actually ask me to do is uh ask me what

[15:04] I actually want to use this for. And here we go. Uh so you can see it's it's it's letting me know that it's a team bot. Uh and it's going to ask me what we what it actually should be used to help the team with. And so what we're going to build right now actually is like a a source of truth and data quer for our entire GTM team. So we have a lot of folks on our GTM team. uh they ask a lot of questions um you know about various different customers, various different usage sources uh you know who are the most power who are the biggest

[15:32] power users on this particular account. They also have a lot of uh inbuilt skills that they use to do things like prospect right to figure out like oh for this account you know who is the best person for me to reach out to. And so we're going to build a bot that helps them do all of that work. Um and so to give it a quick oneliner I'll basically tell it hey while you're talking Oh go ahead. Go

[15:51] ahead. I was gonna try to cover for you. No, go for it. Uh, while I type, uh, go for it. Yeah, I I'm never good at typing and talking. Maybe maybe you're better [laughter] than that. But yeah, I guess while Ari was talking there, just um, you know, this is a real use case, right, that we saw internally at SpaceX when building team bots. And it's actually one of, if I'm not mistaken, this was like one of the the target use cases we were building for internally, right, was our go to market team that had set up a bunch of custom tooling

[16:12] already on their own. Yeah, that's exactly this is exactly this is actually the first team bot that was ever built was uh was a version of this uh was a version of this spot. Awesome. Awesome. Okay. So, you're seeing something real that we use every day within SpaceX AAI, not not a contrived example. Cool. So, I can give it some basic information about what it's actually going to do. Um, the first thing it should hopefully do is uh rename itself on that basis. Uh, so you can see it it decided to give itself a name based on what we what we said it was going to do,

[16:44] which is uh, you know, answer GTM questions. And the first thing we're going to actually do is add plugins. So plugins are the ways that your Grockbot actually connect to external data sources. Uh for a teambot there are two kinds of plugins actually. One is a plugin that is personal to you. So for example, Salesforce is a personal plugin. When the my this teambot uses Salesforce, uh it will actually use the the user who's talking to its Salesforce account. Uh there's also team shared plugins. And so data bricks is one example of a team shared plugin which is you know everyone who uses this data bricks plugin uses a shared key uh that

[17:17] only has readonly access. So you know everyone on your team can query data but not everyone can write to this data store. Uh we also probably wanted to have a few more uh you know granola is very common uh for for sales folks. Uh we probably also wanted to have gong which is what you know a lot of our our our sales team uses to actually track calls. Um and I think that's probably good for now. Uh we can always go back and add more later if we need. Um and let's go ahead and add Slack too because we're going to add it to Slack later. Uh

[17:45] cool. Um then we can go and add it give it secrets. I actually don't need to give it a secret right now. Um so I'll skip this step. But you can give it a secret uh such that it can go call an API uh or do work uh you know that you want to be shared across an entire team. But for now we're going to skip this. Uh and then we can add some skills. So you can see I already have some skills uh loaded from my personal bots. This goes in and reads all the skills I have and tries to pick the ones that are like

[18:08] relevant based on the the you know the task we gave it. Uh so I actually think the ones that it shows are actually pretty relevant. Uh this is going to tell it how to pull information from data bricks which is what we connected earlier. Um this is a you know a style guide for how uh we want it to actually talk to to folks on GTM. Um and I think we I think these are probably pretty good. Uh we can actually go in and add I can go in and add any skills that I've used across any of my bots. So, I'm actually going to add uh one more

[18:37] um which is on account research. Cool. And let's save that. Um I can also add files. So, if you have like PDFs or documents or text files that are PRDS or sources of truth um for for various systems that exist at your company, you can add those. I've already added those as skills for me, so I actually don't need to add more files. And um cool. So, what it's going to then ask me to do is ask if I'm ready to publish it to my team. Uh, I'm going to say yes. Uh, and what happens now is actually this GTM helper will show up on all of my

[19:08] teammates uh, accounts. Uh, when they go and search team bots here, they'll see GTM helper um, right here. Uh, and Roshan right now could look it up and see it on his uh, his Grockbot app as well. Awesome. Um, cool. Uh, so what we can do is ask it something. You know, it already has it's we we loaded it with a bunch of

[19:26] skills and a bunch of uh, data already. So, we can start asking it some questions. Um, so I'm going to ask it sample questions so you guys don't actually see our company information, but we'll still show some cool stuff. Uh, so can you tell me from the sample industries that you have data on, what

[19:45] is the most popular? So it has under the hood, this is using the connectors you set up and the skills you set up. Yep. So what it's going to do is it's actually going to look up we have a skill that actually tells it how to query this data source and we gave it access to data bricks and so it's going to hopefully use that skill look it up in data bricks and actually get an

[20:04] answer for us. Cool. While that's running what are some examples of other really great team bots that you've seen? I mean I have my favorites but I'm curious uh what are some of yours? Yeah so I think the coolest thing we actually had is that most of the team bots feature was built by a team bot. Um and so yeah, so you can see here uh we have this kind of test data. Uh it's going to give us some information about which which accounts are are most popular and across which sources. Um yeah, so going back to that we actually built most of this feature with a

[20:35] teambot itself. Um so the team bot is where engineering, product and design all collaborated. It built up a huge knowledge store of all the things we had worked on, a bunch of product decisions we had made, design decisions we had gone back and forth on. Uh and then it actually kicked off cloud agents uh through a cursor project to actually implement most of the features. And one of the coolest things we had it do is every night it would actually dog food the product. Uh you know we gave it a test account uh on on Grockbot so that it could go around and play with the UI

[21:02] and and and and try different things out. And every morning it would put up a like you know 15 to 20 PRs fixing various nits, copy things that were confusing, loading states that were bad. um you know all this stuff that we probably would have you know it would take us a few days to go out into the field get feedback from real customers or internal users instead rockbot did that every night uh by itself while just you know and just put up PRs for us to approve it allowed us to ship way faster than I think we would have been able to do otherwise that's awesome so software development

[21:35] teambot to go and build the team bots feature yeah exactly uh we called it we we lovingly called it team botsbot we're not that creative with our naming um But yeah, it was very cool. I love it. I love it. Okay, cool. Sorry I took you off your demo track. No, no, no. Love it. Um, what we're going to now do is actually add this bot to Slack. Uh, so we can so, you know,

[21:54] multiple people can actually talk to it. Um, and you can see I can actually just prompt it to do it and it'll be able to do it. But just to show you what this looks like, I'm going to do it via the UI. Um, so once I've built it, you can see, uh, I can always go back and change what plugins it has access to. I can add new secrets. I can add new files. I can add new skills. Um, but for now, what we're going to do is actually bring this to our Slack. Um, and so I'm going to add this to our our our test Slack

[22:15] workspace which is called Fly Low Airlines. Um, if you have a if you're uh Slack Workspace does not require actually any admin approval. Um, oh, we actually have added too many bots to our Slack workspace. [laughter] Second while I uh too many demos go and remove one. Um, but uh you'll basically see that uh if you don't have an admin approval step in your Slack workspace, which most people typically do not, uh you can just add it directly to Slack. Um and otherwise uh you know you'll actually submit a request for your admin to approve uh this bot being

[22:48] added to Slack. Yeah, super great point. So like if you're you know many Slack organizations of course have you know rules around bots, but we've made it really easy to just be able to add and prompt to you know create a new bot in Slack. And once that's I believe it's just once the Slack integration or connector is added to Grockbot that sort of automatically makes that connection. Um but then yeah if your admin needs to approve um you can you know ask your your Slack admin to jump in and approve the bot. Um while you're uh you know while you're cleaning that one up Ari the other thing I was

[23:18] going to mention is that um we've seen a massive proliferation of bots um within SpaceX uh as soon as this feature was launched internally. uh everything from the go to market bot that we were talking about earlier to a legal ops bot to IT support uh really like our IT team using bots to help direct traffic on inbound help desk questions. Um seen a few others that are coming in coming into the into the chat here as well. Um, but really the way I'd think about team bots is any case where you would have multiple people coordinating to do an activity uh and you want to add some

[23:58] intelligence and maybe memory uh to all that work uh bots can be a great fit. What do you think? Yeah. Uh I think another use case we've seen be super useful is as like a first point of triaging. Uh so typically most teams and companies have like many many channels that are devoted to just asking questions, right? where anyone in the company can go ask a question, you know, help legal, help uh people ops, help revops, help finance, help, you know, this product area. Uh and it turns out if you have a team bot that actually, you know, is just like listening in that channel all the time, around 70 to 80%

[24:30] of the questions just like get answered by the bot and you actually don't need a human to be involved. It's a really really great way to just like free up time that like otherwise is not super useful. uh you know most times when someone asks a question like that uh especially in channels where I'm a person who's responding I will just go ask my agents anyways for what the answer to that question is instead you can kind of cut me out as the meat proxy here and just have the bot answer directly based on you know sources of truth that I can vet and I can monitor

[24:56] and I can and eval over time that it looks like your Slack integration worked yeah so we uh so you can see we were able to actually add this to Slack and what I'm going to do is actually switch my screen sharing over to Slack back. So, we can actually try this live. Um, so here we are. Uh, this is our our test little Slack thing. Oh, and looks like Roshan's already

[25:17] He beat you to it. He beat me to it. So, he's going to be able to ask GTM Helper. Hey, what's up? And if the demo gods are working with me, we should hopefully see uh Oh, so first you can see uh it gives me a little message when it's added to Slack. Hi, I'm GTM Helper. The team bot you added to Flyo. Uh, and Roshan asked it a question. It's able to. So, oh, Roshan has to link his account here, but I think my account should already be

[25:40] linked. Okay, my demo is not as good as yours. I shouldn't I shouldn't have tried. [laughter] Uh, and if I ask it, hi, who are you? Uh, it'll give me this little nice thinking state while it's working. And then what I love about this is how native it feels to Slack, too. It just like feels like you're literally chatting with someone you work with in Slack, which Yeah, exactly. It it it works. you know, it tags me like a real human being would. Like, hey, Ari, uh, you know, I'm GTM helper. And it's able to answer questions for me. And the cool thing is, uh, you know, if Roshan linked his

[26:10] account, he's able to respond in this thread. Uh, when he talks to it, it uses his credentials and, you know, is able to do things on his behalf. When I talk to it in the thread, it's able to use my credentials and do things on my behalf. So, it's a very nice combination of, you know, it's its own teammate that has its own credentials, but can also do things on your behalf when you uh when you ask

[26:28] it to. Um, Final thing I'll talk about which I didn't talk about enough is that this actually learns over time. So it has a shared team memory store uh that actually allows it to to learn as you keep using it. So an example we have is like we have this data bot which is like the first point for if any you know any

[26:45] data analysis questions you need to ask. Sometimes the data bot gets things wrong. Uh you know and our our Sally our amazing data analyst will come in and be like hey actually no this answer is wrong like here's the like correct way to do it. Uh and the good thing is like the bot learns from her fixing it. So when she tells it, hey, you looked at the wrong table here, it stores that as a memory typically and so that the next time someone asks about it, it actually

[27:05] is able to pull the right information. So it's a real teammate that like learns with you over time just like a real teammate would and you can delegate real work to them. Amazing. Okay, the biggest question in the chat is who can use team bots? Who is it for? Um like can anyone turn this on? Can you share a bit more about how to get how to get started? What type of account you might need? Yeah. So right now this works if you are using a cursor teams or enterprise linked account uh for Grockbot. This will in a few weeks be available to super grock teams users

[27:35] as well. Um and we are working right now on bringing some of this functionality to just like individual users as well. Uh so right now you need to be on the same team which is what you know allows these like very permissive expressions of team bots. Uh we will probably have another version of this coming out soon that will be probably more restricted in its permissions but will still be very cool uh that you'll be able to use even if you're not on the same team. Okay, very excited for all that. A lot a lot of great stuff to come there. But um yeah, I guess like I can't say enough

[28:03] how much this has changed our internal workflows at SpaceX. Just amazing to see everyone creating bots. Data bot is one of my favorites. Forgot to mention that one earlier. Just being able to ask data science questions. um get status on how things are going, track revenue, track product usage. Um it's just really great to have that collaborative surface. And honestly, I think what makes it even better is the fact that I know it's learning, right, like you said. And so when a data scientist comes in who corrects the bot, it's just Yeah, I I I

[28:29] it's such a such a cool use case. Yeah, 100%. Um I think some other questions I see in the chat uh is can your team bots talk to each other? The answer is yes. uh especially in Slack, it's a very common thing that we have people do. Uh so for example, another example we have internally is like we have a deal desk bot uh that you know it takes in like triages all incoming requests from GTM. Sometimes if it's a legal question, it farms it out to legal

[28:56] bot. So it tags in in Slack at legalbot. Hey, can you take a look at this? Legalbot will come in and answer that question. Uh so the bots can communicate with each other just like real people can. Um so that is uh is very very useful. Um the other question I saw is um if one if one user asks the bot to do a very long task does that block the bot for everyone else? The answer is no. Uh so the bots share like context, they share memories, they share a lot of things, but they have their own sessions. Uh so you can just because you're talking to it and it's taking a

[29:28] long time to think does not mean you're stopping someone else from talking to it. The bot will be able to respond in both places at the same time. In that way, it is a little bit better than our our human teammates which have to typically do things one at a time. And Ari, remind me, you can chat with team bots in Grockbot or in Slack, right? Once you've added them, it's kind of wherever you want to interact with

[29:46] them. Is that right? Exactly. Uh you can talk to them in Grockbot just like you would talk to any other bot, but you can also talk to them in Slack. I also saw some questions for this. Uh we are going to bring it to Microsoft Teams soon. So you'll soon be able to do this in in in Teams as well. Amazing. Okay. Well, we'll definitely take some more questions on team bots, uh, if there are any more in a little while, but I want to move to the third section for today, which is going to be a little bit about voice mode. And so, um, I'm going to follow in AR's

[30:17] footsteps and hopefully do as good of a job. Um, but we'll also do a bit of a live demo here. And so, um, for those who haven't used it yet or seen it, voice mode is a really exciting feature that we built into Grabbots a little while ago. Um, this little icon down here that looks like you can talk to the bot. Does exactly what you'd expect it to. Um, I mentioned earlier when we were setting up and kind of gathering for for this morning that, um, this is one of my favorite use cases when I'm driving into work each morning, just to be able to

[30:46] hop on a quick call. And really the inspiration here was um sort of like a slack huddle or a way that you'd you know get someone you work with on the phone for a quick question. Um and so really these can be shortlived conversations, they can be longive conversations. But what I really enjoy here is that these are two-way

[31:03] birectional voice chats with your bots. Um you can customize the, you know, the the voice that the bots speak in. Uh you can, you know, change how quickly they talk. I'll show you some of those features. Um, but really the the idea here is that you can basically talk to your bots and get a quick sort of high bandwidth conversation when you don't want to type or you can't type. And actually, I'm going to reshare my screen here so I can make sure um that we are

[31:27] sharing the audio. Looks like we are. Okay. So, we'll see if this works. Maybe Ari or Parker, one of you guys can give me a thumbs up if it if it does. Um, what I'm looking at here is my team bot setup for Fly Low Airlines. So, that shared team that we just talked about. Um, this is my PM org. So, I have PM Pete. I have Cora, my chief of staff. I

[31:45] have Ashley who's my data scientist. Pixel, who's my designer, Emily. So, I have a whole team of bots here, a whole engineering team that's working on features. And we've been doing some work together to um understand what's going on in our mobile ticket funnel. So, Flyo is a fictional airline company that sells airline tickets and operates an airline. So, we've been doing some work into the into the funnel, right? Like we all everyone cares about funnels and, you know, making sure that customers who want to buy something can get through that funnel and buy something and and do that in a really effective way. But what

[32:11] I'm going to do is just check in with Cora to see how things are going. So, let's see if this works. Hey, Cora, can you hear me? Loud and clear. Awesome. Hey, can you um just do a quick run through on Slack and see if anything requires my attention right now? Pulling Slack now. Cool. I'm going to mute my voice out while I'm talking here. So, Cora like is connected to all the same connectors here that I normally use. So, voice mode supports all the connectors that you've already connected into Grockbot. Um, and what's really cool about this is that you can just do anything you normally

[32:43] would by typing into Grockbot, but with voice instead of having to type. And so, looks like Cora is getting back to us now. I think my favorite part of this is actually how fast it responds. It's like it's almost faster than like if I was actually in a huddle with a real person. Um, it doesn't take time to think. It basically responds immediately after you

[33:00] stop talking. Yeah. Hey, Cora, what's going on? Still waiting on that Slack sweep. want me to widen it to email and calendar, too, while we're at it? Nah, let's just stick with Slack. And in fact, why don't you just uh tell me what you have so far? Nothing's landed yet. Give it another minute and I'll have the rundown. Okay. Okay, I can wait. Um, while we're waiting, you can see here that uh we can change a lot of settings for how the voice behaves. Uh, so this is kind of what I was talking about earlier. We have a ton of voices. This is powered by our Grock voice APIs uh uh that give you

[33:31] all kinds of options, all kinds of voices. You can also change the speed. I'm a 2x podcast listener. Usually, I won't make you all have to do that with me. Um, you can also change the language, which is really cool. And so, we had so many international folks joining from around the world. Um, you can choose your language and the bot will speak to you in whatever language

[33:47] you want, which I think is really cool. Um, it looks like Kora might have gotten us stuck on this particular query. So, let's see if we can ask Kora to do something different. Hey, Cora, this seems like it's taking a bit longer than I expected. Why don't you just let me know um about the last few things we talked about in our chat here? from this chat. LHR sky serve ground handling is this week's focus. Brian Hazel's Saturday post in station ops and fly low ask untagged Dana's one pager pre-eread

[34:08] for the airport authority is due today. Actually, um what's that one pager for the pre-eread that you mentioned? Dana wants it as a pre-eread for tomorrow's airport authority meeting. Due today EOD, can you um tell PMP Pete to help to start working on that pre-eread for me actually? And then I also want you to ask PM Pete uh what's going on on our funnel analysis work that we recently did. I think we were working on some features to optimize our mobile checkout funnel. I'm just not quite sure where

[34:34] that landed. Messaging Pete on both. Cool. And so you'll notice, I'm going to mute myself again here. Um, the bots can also talk to each other just like they would uh in Grockbot normally, but using voice mode. And I find this to be really cool. So if you have different bots who are playing different roles, I have like my engineering team. PM Pete is my product manager here. And you'll see that these bots are going to talk to

[34:54] each other while we're on this call. He's on it. Both asks are in. One correction worth knowing, Brian's post actually put Dana's pre-eread due last Monday for a sink on Tuesday the 6th. Not That's okay. We can just get it done today. Good call. Pete also flagged the three topics Dana wants covered. T2 winter stands, the 25minut turn versus the

[35:15] published slot mix, and IRS desks. Awesome. That that all sounds great. Can you just summarize the to-dos that we have right now in our chat just so I can track that three open one the Slack sweep is still running two Pete's drafting Dana's one pager for today EOD T2 winter stands 25minut turn versus slot mix IRS desks three same thread Pete's reporting back on where the mobile checkout funnel work landed cool that sounds great and then um you said that there was nothing new in Slack but actually I think my teammate Ari might have added a new go to market bot to Slack. Can you check to see if that's

[35:53] live? Looking for Ari's bot. You see if this works. Uh Cora seemed to have some trouble getting to Slack before. [laughter] I guess while we're waiting for Cora here. Uh Ari's bot is live. It's called GTM Helper. Posted first in number all-fly low-ines at 10:29 this morning. You DM'd it hey what's up and it said it's waiting for you to link your account. Ari asked who it was and it listed four things it can do. research an account, prep for meetings, surface recent signals, and find reference customers. You can appmention it anywhere, or DM it, but linking your

[36:28] account is the gate. Awesome. Thanks. Okay. And this is like super easy, right? I just hung up hang up the call. I'm done with what I need. I move on with my day. Um, and you can see that in this case, Cora was able to talk to other bots. We got some insight about to-dos on our list. And it'll actually give me a summary after the call as well, which I find super helpful. So, I get a sense of like, hey, here are the main things we covered. And then, you know, because this was also my main bots, my proactive bot, um, this bots now has even more context in the

[36:54] things that I'm, you know, curious about when I choose to call it and hop on a voice call. So, really, really fun feature. Um, but also just really incredibly useful. There's a bunch of folks internally at SpaceXi who are using these bots for morning briefings. They're using these bots to quickly take notes on things. Um, we did a demo a few weeks ago where we actually showed how you can use voice mode for engineering tasks where you can very quickly ask one of your bot to spin up cloud agents, create PRs, make changes to code, um, and all and much much more really. The sky's is really the limit here. And so I

[37:23] guess with that, Ari Parker curious, do you guys have do you all have any favorite uh team or uh, voice mode use cases? I can talk through one. Go for it. Go for it. Yeah, I can talk through one that's been that's been pretty crazy for me. I uh I have set my action button on my phone to

[37:42] open a voice chat with my primary bot. And so I will often do that in the morning, you know, as I'm kind of getting around or or anything like that where I will just press the action button immediately. I'm in a call with my primary bot and we're just talking about, you know, the start of my day,

[37:55] getting the things I need to work on. And I think that has been that has been pretty incredible. It it gets me into this experience incredibly fast and it's it's a great way to get all the information. I love that. Yes, great point. The voice mode does work on mobile as well. And so if you don't have the Grockbot mobile app, you're missing out. It's really cool. You can do all the things that we've shown you today from from mobile as well. And Ari, did you have a

[38:14] favorite voice mode use case? Uh I feel like I just I use it just to talk through like things that I'm not 100% done thinking through. Uh it's really good as like a brainstorming partner of sorts. Uh especially when I'm thinking through like, hey, like what are the next things we should build? I usually like first stream of consciousness talking them out to a to a to a person. Uh, and now I can basically do that with the bot and and it's a really useful way to like clarify thinking because it'll, you know, after I I give this whole stream of consciousness thing, it'll be like,

[38:42] "Hey, Ari, some of the things you said were smart. Some of the things you said were not smart and we probably shouldn't ship those." Uh, it's very useful for for for keeping me honest like that. I love it. I love it. Cool. Okay. Well, that takes us through what we wanted to cover today. Hopefully, you enjoyed kind

[38:56] of just seeing what's new in Grockbot. Um, we did want to save some time for Q&A and so maybe take a couple of questions from the chat. So, I'm going to scroll through the scroll back a little bit, but if folks who are joining have a question, please feel free to drop it into the chat. We'll pick our favorites. Uh, and maybe I'll tag in the

[39:13] team here to help answer some of them. So, uh, okay, let's see. Got some comments on live demos. Thank you for the kudos there. Um, someone's asking if it's possible to add a voice in the future. Not currently planned. We have a a set of voices that are expo that are available via rock voices. We're really focused on those for now, but definitely

[39:32] a good idea. Someone's talking about projects. Well, one person's asking about team bots for team accounts and I think are you call you called earlier teams and enterprises and then soon for uh for cursor accounts and then coming soon to super grock teams. Okay. Uh I saw a few questions about uh the like X use case that we talked about at the beginning. That's pretty easy to set up. Uh if you go to connect apps in the bottom left of your Grockbot application, you should be able to look up the X plugin. Uh that'll it'll prompt you to authorize with uh with X uh and then you should just be able to it

[40:07] should be able to query X for you. It's actually even better. As of yesterday, no plugin needed. We added a new feature yesterday afternoon I want to say where you no longer need to add the plugin. So if you want to connect Grockbot to X in readonly mode, uh that is now just included in Grockpot. So, um, yeah, go forth and go forth and doom scroll. Uh, someone was asking, Parker, maybe this one's for you. Uh, is proactivity in the marketplace? Do I

[40:32] have to turn it on? Yeah, great question. So, you do not have to turn anything on for productivity? This is something that's on by default comes uh kind of included. I think one thing to mention is, you know, you do want to turn on a good chunk of connectors to make it the best experience. Um, but yeah, as soon as you sign up for Rockbot, you you have

[40:49] productivity enabled. The proactivity is also included. Um, I see a question. Go ahead. Go ahead. Can we set who in our team can use which team bots and restrict some with sensitive skills and abilities to only certain members? Uh, as of I think a day and a half ago, the answer is yes. So, you can actually limit uh you can actually create two kinds of roles for a teambot. Uh, managers and you know users. Uh, so managers are people who can actually go in and edit skills. They can edit the plugins. They can edit the memories. They can go and actually manually edit stuff. Uh and you can also

[41:25] set users who can who are the people who are allowed to use your app and who it shows up in the app for. Uh that can be your entire team. It actually can be just a skim group. So you can you know define it for a finance function or an EPD function uh or just individual people. So you have you have a pretty granular uh arbback controls on team bots. Now love it. Awesome. There's a question here about the Microsoft suite and how you can use Teams or Microsoft products with Grockbot. We do have connectors for Teams. I think I think SharePoint is live uh or coming soon. Um we have a

[41:57] bunch of other Microsoft connectors. We're working on bringing team bots to Teams very soon. Um but yeah, you c you can already hook up uh a bunch of connectors for Microsoft services as well if you browse that marketplace. Uh all right, let's see what else. any any other good ones we're seeing here. Lots of great use cases in the chat and I really do think we should we should mine through a bunch of these. Um some really great feedback for the team. Um there was a question about what's the best or safest way to give Grockbot passwords to store. Uh I don't know if Parker you want to talk through the one

[42:30] password uh integration. Yeah, absolutely. Uh so yeah, we do have a one password integration that you can find in the marketplace uh by searching for one password. And so what you'll do is we'll have you connect uh your one password account and then you'll choose the credentials that you actually want to share with Crockpot and uh it'll have access to them. So you'll be asked, you know, can I use these credentials? Then you can always allow or just allow for that case. So it's a great way to share credentials and um I think we have more

[42:55] coming here soon. So uh yeah. Yeah, I'm seeing lots of stuff coming into the chat for more and other password manager integrations. So great feedback for the team. Also, some kudos that the one password integration uh seems to be working pretty well for some folks. So, I love to hear that, too. Um, cool. Well, I know we're almost at the hour. We thought we'd take a half hour and we ended up going almost the whole time. So, why don't we wrap up here? I guess Ari, any parting thoughts from you, Parker? Any parting thoughts from you uh before we wrap up? Uh, use Grockpot. Give us feedback. We are

[43:25] actively shipping every day as fast as we can. So, uh, please send us feedback on X uh either through inapp feedback, any channel you can get it from. We are monitoring all of them uh and we'd love to make their product better for you. Parker, same same thought. He he took the words out of our mouth. So yeah, I think that

[43:42] the feedback piece is super important. Hopefully folks have seen that we we are trying to react super quickly to feedback. So keep it coming. Yeah, hopefully all are seeing that. Um definitely the best way to stay up to date on what's literally shipping every day by is just by following atbot onx. We are posting at least once a day with something new that the team has cooked up. And so if you really want the latest

[44:02] and greatest, that's the place to go. Follow atbot onx. And then um you know, we're we're all posting from our personal accounts, too. We read all of the feedback that comes in to replies at bot and for replies on our personal accounts, too. And so, please just keep the feedback coming. It truly is a source of inspiration and feature requests for the team. And a lot of the stuff that we've built have been directly in response to things we're seeing. So bug reports, what you want next, what's working, what's not working, any and all that stuff is just incredibly valuable for us and we truly just want to build a product that you

[44:32] all are really excited to use all the time. Uh, and so I know we have a bunch of other workshops coming up in this series too. Um, Batist put a link in the chat a little bit earlier. Maybe we can repost that just to share. So, if you're interested in joining more of these sessions, really just want to give you all ways that you can continue to grow how you use Grockbot, think of new ways to use it that you might not have, and also just give us a chance to hear from you, like we said earlier, on on all the great things that you all are doing,

[44:54] too. So, with that, um, thanks so much for spending some time with us. We're really excited for all the things you're building with Grockbot. We can't wait to see what you guys build next, and we can't wait to ship a bunch of really cool stuff soon. Uh, thanks Ari, thanks Parker for joining, and we'll see you

[45:07] all in the next one.

---

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