← Back to videos
This is actually insane. SpaceXAI engineer Nick Miller: "I haven't opened a single bot chat in 3 mo

Waunty
@0xWaunty
November 18, 2026· 27:24
This is actually insane. SpaceXAI engineer Nick Miller: "I haven't opened a single bot chat in 3 months. My Chief of Staff bot does it for me It runs a PM, a QA and 25+ workers. I only see what passed QA" In 27 minutes, he breaks down how to turn a single Grok Bot into a team that hands off work and checks it without you Grok Bot → Chief of Staff → PM → QA → 25+ Workers Adding more bots doesn't make a team. Handoffs and verification do Worth more than most $1,500 agent engineering courses Save it and watch it today Then check out the full setup: Grok Bot team ↓
Transcript
- My name's Nick Miller. I am a
- 0:02
- lead field engineer at SpaceX AI
- 0:05
- still getting used to that
- 0:07
- I, I
- 0:07
- worked at Cursor and came over from the acquisition
- 0:11
- and, you know, when I first started putting together
- 0:14
- the talk for the session
- 0:17
- several weeks ago
- 0:18
- it was a very different talk
- 0:20
- but some things
- 0:21
- very exciting things have happened here at SpaceX
- 0:24
- AI in the last month or so
- 0:27
- with the release of GrokBot
- 0:29
- And so what I really wanted to focus
- 0:30
- in on today was GrokBot and
- 0:34
- how it is really changing the way that we work
- 0:38
- whether you are a software engineer
- 0:40
- or an executive or on go-to-market
- 0:44
- the problem we're solving with
- 0:45
- with GrokBot is that most meaningful work
- 0:49
- I think, is bottlenecked by Wrote reactionary work
- 0:54
- in our days, I think that we are regularly getting emails
- 0:59
- we're getting Slack messages
- 1:01
- there are all these things that we have
- 1:02
- to react to and act on
- 1:05
- and that often carves into
- 1:07
- the time that we can really do the meaningful work
- 1:11
- the special work that really moves the needle
- 1:14
- forward
- 1:16
- I was an early employee
- 1:17
- at Cursor. I was employee number 26
- 1:20
- before that, I founded a SaaS company
- 1:23
- that I ran for about 12 years
- 1:25
- And
- 1:26
- any of you here in the room that
- 1:28
- have worked at an early stage startup
- 1:30
- I think you
- 1:31
- know the feeling that I had in the
- 1:33
- early days of Cursor
- 1:34
- in the early days of my SaaS company
- 1:36
- where you are just overwhelmed with work
- 1:38
- There
- 1:39
- the ratio of the amount of work that
- 1:41
- you have
- 1:42
- to the number
- 1:43
- of people that you have working is way off balance
- 1:47
- and
- 1:49
- what Grokbot does is it
- 1:51
- acts as a knowledge worker
- 1:53
- It allows you to offload a lot of that reactive work
- 1:58
- that repetitive work
- 1:59
- and frees you up to
- 2:01
- really do the meaningful work that is most enjoyable
- 2:05
- that is leveraging what you are great at
- 2:07
- and again, is
- 2:07
- is what really moves the ball forward
- 2:10
- Grokbot was released, on August 11th
- 2:13
- so about a month ago
- 2:15
- we actually at Cursor
- 2:17
- we built this product in about four weeks
- 2:20
- it started out as just an experiment
- 2:23
- and I think after about ten days
- 2:25
- we realized that we had something really special
- 2:28
- I personally have been using it for about
- 2:30
- two months. It was in an internal beta
- 2:33
- at, at Cursor, at SpaceX AI
- 2:36
- and I will tell you
- 2:36
- if they took it away from me today
- 2:38
- I don't know what I would do with myself
- 2:41
- I, I've started referring to it as crackbot
- 2:44
- actually. When I talk to an enterprise
- 2:46
- I'm just like, get the executives, crackbot
- 2:48
- We don't even need to do a session
- 2:50
- You just need to start using it
- 2:52
- and you will not be able to put
- 2:53
- it down
- 2:55
- so over the last three years
- 2:58
- you know, no one has really cracked the code
- 3:00
- I think
- 3:01
- on making a
- 3:02
- singular interface that is really useful
- 3:05
- across a wide variety of domains
- 3:08
- we first started with chats
- 3:09
- right
- 3:10
- where you were able to
- 3:11
- essentially ask the chat bot to do something
- 3:14
- it would give you some information back
- 3:16
- and then you would act on it
- 3:19
- and then we got into the co-pilot area
- 3:21
- era
- 3:21
- cursor was a part of that
- 3:23
- as far as software goes
- 3:25
- you had other companies that were releasing legal agents
- 3:28
- finance agents, teaching agents, all of these different agents
- 3:32
- but I think
- 3:33
- the idea of this general agent framework was still
- 3:37
- left unsolved
- 3:39
- think about Grokbot is really a new
- 3:42
- agent primitive
- 3:44
- these bots are not limited to
- 3:46
- just retrieving information or just one knowledge area
- 3:52
- bots can really function as a team
- 3:54
- Team
- 3:54
- of knowledge workers with the ability to take
- 3:57
- an initial directive
- 3:58
- and turn it into routines that accomplish actual
- 4:02
- units of work for your organization
- 4:05
- the bots can complete really complicated
- 4:09
- long-form tasks that these early primitives were not
- 4:12
- capable of. I think that is one real difference here
- 4:15
- is that you can really give a bot
- 4:18
- something that is many
- 4:19
- many turns of work
- 4:21
- and it's able to finish it end-to-end
- 4:23
- And I'll give you a quick example
- 4:24
- I'm gonna talk a lot about using GrokBot in a professional
- 4:27
- context
- 4:28
- but I use it
- 4:29
- personally as much as I use it professionally
- 4:31
- I'll give you a quick example just to
- 4:33
- kind of whet your appetite
- 4:35
- I travel around the world for Cursor
- 4:38
- for SpaceX AI
- 4:39
- I'm currently actually living out in Singapore for about three months
- 4:44
- helping us spin up our APJ operations there
- 4:47
- And
- 4:47
- my wife and my kids are out there
- 4:49
- with me
- 4:50
- they joined me a few weeks ago
- 4:52
- and, it's about a week before I
- 4:55
- they were arriving in Singapore to join me
- 4:57
- and we still didn't have a place to
- 4:59
- live. And my wife was freaking out
- 5:01
- understandably so
- 5:03
- I was super busy with work
- 5:04
- and I
- 5:05
- I needed to find us a short-term rental
- 5:07
- So I had Grokbot do it
- 5:09
- I told Grokbot, this is what we need
- 5:11
- We, we need a three-bedroom furnished place
- 5:14
- We need it for three months
- 5:17
- I have two daughters. One is 15
- 5:20
- her boyfriend was coming to visit for a
- 5:21
- week. There was no way in hell they
- 5:23
- were going to be in the same bedroom
- 5:24
- so I needed an arrangement where
- 5:26
- he could be siloed away and locked up
- 5:28
- at night
- 5:29
- I gave Grokbot all this information
- 5:31
- I said
- 5:31
- I want you to go out and research and find
- 5:35
- short-term rentals for us
- 5:36
- and I want you to create a
- 5:38
- Google sheet of all of those rentals
- 5:40
- So I gave it access to my Google
- 5:42
- workspace
- 5:43
- It went out, it researched
- 5:45
- it came up with about 35 different properties
- 5:47
- that met my requirements
- 5:49
- put together this sheet
- 5:50
- Then I said, email this to my wife
- 5:52
- Now
- 5:53
- this is one of the first Lessons I
- 5:54
- learned is that it emailed it to my wife
- 5:57
- in my name
- 5:58
- so it signed kind regards Nick at the
- 6:01
- end. My wife was like
- 6:02
- what the hell is this?
- 6:03
- She knows that every email I send her
- 6:05
- I never put any sort of signature in
- 6:07
- there
- 6:07
- That was one thing we had to correct
- 6:08
- with Grokbot
- 6:09
- But she then went ahead and she marked
- 6:11
- five different properties that she liked
- 6:13
- She highlighted them yellow. She told me when
- 6:15
- she was done
- 6:16
- I then went back to Grokbot and said
- 6:17
- okay
- 6:18
- now
- 6:18
- I want you to go ahead and I
- 6:20
- want you to schedule appointments for me
- 6:22
- to view these apartments. Now
- 6:24
- Grokbot is also connected to my calendar
- 6:26
- So it was able to see what days I was available
- 6:28
- when I had windows of a time
- 6:30
- to go see these places
- 6:32
- and then it actually contacted the properties and set up the
- 6:36
- visits
- 6:37
- I went out, I visited the properties
- 6:39
- we narrowed it down to two
- 6:41
- and then Grokbot actually negotiated the leases with
- 6:44
- both of these places
- 6:46
- and got us all the way to the
- 6:47
- finish line. So
- 6:48
- very different from the co-pilot era and the chat
- 6:51
- chatbot era
- 6:52
- in terms of how wide that
- 6:54
- And where it can really go
- 6:58
- ease of use, I think, is
- 6:59
- is really important
- 7:01
- when we look at a lot of the existing knowledge work
- 7:04
- products out there
- 7:06
- I think that there have been some challenges
- 7:08
- when we've tried to kind of find the
- 7:09
- perfect form factor
- 7:11
- they're really complicated to set up
- 7:14
- they run locally on your machine
- 7:17
- there are all sorts of limitations around connectors
- 7:20
- They're flaky, they're hard to configure
- 7:23
- and you're constantly having to babysit them
- 7:26
- and it's just not easy to share these
- 7:28
- things. It can take a lot of time
- 7:30
- to get them set up
- 7:31
- and then it's not easy to kind of democratize
- 7:34
- that access and provide it to other people
- 7:38
- Grokbot was really designed to be super simple
- 7:41
- In fact, you know
- 7:42
- one of the mantras we have at Cursor is that
- 7:45
- is delete the product
- 7:47
- we are, as you know
- 7:49
- if any of you use the Cursor platform
- 7:51
- we are shipping new features all the time
- 7:53
- This is a space that's moving extremely fast
- 7:56
- but we are deleting features as much as
- 7:59
- we are adding features. So deleting the product
- 8:01
- is really an important kind of pillar of how we work
- 8:04
- and we definitely took that approach with Grokbot
- 8:07
- If you were part of the internal beta
- 8:08
- that we had with Grokbot before we did an external release
- 8:12
- there was actually a lot more interfaces
- 8:14
- a lot more that you could do with it
- 8:16
- and we actually trimmed a lot of that away
- 8:19
- so it is really just dead simple
- 8:21
- It looks like an iMessages
- 8:24
- sort of interface. So
- 8:25
- it's designed to be very easy to use
- 8:28
- When you open up the application
- 8:30
- it just, it just works
- 8:32
- so when you define a new bot
- 8:35
- there are kind of
- 8:35
- three parts that you want to focus
- 8:38
- Context connectors and outcomes
- 8:42
- So as far as context goes
- 8:43
- when I'm designing a bot
- 8:45
- I always start by giving it a name and
- 8:49
- some sort of initial understanding of what it
- 8:51
- does. You can see in this example
- 8:52
- we have a bot that's called Sales Outbound
- 8:55
- and the bot will even
- 8:57
- look at the name and automatically make some
- 8:59
- assumptions about what its purpose is and then
- 9:01
- ask you sort of how you want to
- 9:03
- use it
- 9:04
- Then connectors. I think this is really kind
- 9:06
- of the magic of what makes these bots
- 9:08
- so powerful
- 9:09
- You are able to connect them to
- 9:12
- all of these outside sources
- 9:14
- So at Cursor, for example
- 9:16
- there are maybe about a half dozen kind
- 9:18
- of
- 9:19
- information sources that are really key to my
- 9:21
- job
- 9:22
- There is Slack
- 9:23
- which is really kind of the nerve center
- 9:25
- of our company
- 9:26
- there is Notion
- 9:27
- there is Google Gmail and Calendar and Docs
- 9:32
- there's Salesforce. I use Granola for all my
- 9:35
- meetings. So I have the bot connect
- 9:38
- To all of these
- 9:39
- and then it has eyes and ears and
- 9:41
- fingers and all of these tools
- 9:43
- And then outcomes
- 9:45
- The agent needs to know what it should
- 9:46
- do
- 9:47
- So in the same way that we might
- 9:48
- measure a human and how many meetings they
- 9:51
- book or how many PRs they create
- 9:53
- bots need to have a similar sort of
- 9:55
- goal in what they want to accomplish
- 9:58
- So when you deploy a Grok bot for your team
- 10:01
- you know
- 10:02
- it might look like
- 10:03
- you're working on iMessage all day
- 10:05
- I mean, this is
- 10:05
- I can't tell you how much my day
- 10:08
- how much my work
- 10:09
- how much my personal life has transformed
- 10:12
- I spend so much of my day now
- 10:14
- in this interface as opposed to in Slack or in Notion
- 10:18
- in these other tools
- 10:20
- But I think
- 10:20
- one of the things that really makes
- 10:23
- Grok bot special
- 10:24
- is that it learns from you
- 10:27
- So most of the other tools in this space
- 10:29
- when you ask a bot to do something
- 10:32
- it is essentially kind of launching this ephemeral
- 10:34
- session that is there just for that task
- 10:38
- These bots persist
- 10:40
- So what that means is that they continue
- 10:42
- to get smarter and smarter
- 10:44
- And as you guide them
- 10:45
- as you correct them
- 10:46
- as you give them more information
- 10:48
- as you teach them
- 10:49
- they get better and better at the work
- 10:54
- this is a good graph of
- 10:55
- of what I was talking about at the beginning
- 10:57
- of the talk here
- 10:58
- I think on the left side
- 11:00
- this is really a lot of our day
- 11:03
- I can tell you
- 11:03
- this is what my day is
- 11:05
- All that red stuff
- 11:06
- and honestly
- 11:07
- I've had plenty of days where most of
- 11:08
- those blocks are red
- 11:10
- I get so many messages in Slack
- 11:13
- email, to-dos
- 11:14
- I could spend all day
- 11:16
- honestly, just in Slack
- 11:18
- and never get to the really strategic high-value work
- 11:21
- which is why the company hired me
- 11:24
- And the idea with Grokbot is really to
- 11:27
- get the bots to take on as much
- 11:29
- of that repetitive work
- 11:32
- so it can really free you up to
- 11:34
- do the magical work
- 11:35
- that is most fulfilling and
- 11:37
- and why you were hired
- 11:39
- Or why you founded your company or why
- 11:41
- you do your work
- 11:43
- I think another core differentiator
- 11:46
- with Grokbot is that all of these bots
- 11:49
- share a computer in the cloud
- 11:52
- So a lot of the other
- 11:54
- knowledge work tools out there
- 11:56
- they're running locally on your computer
- 11:59
- but all of your bots have a
- 12:02
- computer in the cloud with its own browser
- 12:05
- that persists as it works
- 12:07
- So for the average user
- 12:09
- it means that these agents just
- 12:11
- just work. You don't need to
- 12:12
- manage the runtime environment on your local machine
- 12:15
- You're not competing for resources
- 12:18
- if a
- 12:19
- let's say a non-technical user wants to develop a prototype
- 12:22
- the agent has a place to operate that
- 12:24
- isn't constrained by the user making the request
- 12:28
- and this also solves the problem of system
- 12:30
- integration
- 12:31
- giving the bots access to a browser means
- 12:34
- that they can interface with systems that may
- 12:37
- not necessarily have an MCP server
- 12:39
- Or an API
- 12:41
- So a bot can utilize the computer to
- 12:43
- use technology to navigate through a browser
- 12:46
- in the same way a human would
- 12:48
- and this really opens up a world of
- 12:50
- possibilities
- 12:52
- Grokbot is the
- 12:53
- fastest growing product that the cursor team has
- 12:56
- ever developed. I was
- 12:57
- looking at the numbers here for this week
- 13:00
- 418,000 weekly users
- 13:02
- hopefully that number will tick up a little
- 13:04
- bit after this session
- 13:05
- and more than half of those users are
- 13:08
- using Grokbot for more than four days a
- 13:11
- week
- 13:12
- I use it every single day
- 13:14
- morning, noon, and night. In fact
- 13:17
- at 5 a.m. every morning
- 13:19
- my chief of staff sends me an initial message
- 13:21
- which gives me a readout of my day
- 13:24
- It's looking at my calendar and telling me
- 13:26
- what I'm doing that day
- 13:27
- It's pulling in information from other sources to
- 13:31
- help educate me on those things I've got
- 13:33
- to do
- 13:34
- It's looking at my inbox
- 13:35
- it's looking at my Slack
- 13:37
- it's telling me what are the high priority
- 13:38
- things. Things I need to kick
- 13:40
- tick off
- 13:41
- and then I'm using it throughout the day
- 13:43
- So the goal here with Grokbot is really
- 13:45
- to have it used wherever you do knowledge
- 13:47
- work
- 13:49
- Grokbot is a generic agent primitive
- 13:52
- bots are not constrained to be
- 13:54
- only good at one particular domain
- 13:57
- they can be a
- 13:58
- a team of autonomous
- 14:00
- knowledge workers
- 14:01
- that can transcend all different types of jobs
- 14:05
- at SpaceX AI
- 14:06
- users across all parts of the company
- 14:09
- are leveraging bots to increase their productivity
- 14:13
- single engineers are having the impact of ten engineers
- 14:16
- the 10x number that we always hear
- 14:18
- Individual go-to-market members are accomplishing what previously looked
- 14:22
- took an entire team
- 14:24
- And so far, feedback has been really positive
- 14:26
- Any of you that are on Twitter
- 14:29
- I know there's always kind of a flavor
- 14:30
- of the day or a flavor of the
- 14:32
- week
- 14:33
- it is very much a horse race
- 14:34
- I think, in, in terms of
- 14:36
- Twitter with all of these different companies working
- 14:38
- in AI
- 14:39
- but we've had a really great moment with Grokbot
- 14:41
- and I think the reception has been
- 14:44
- really positive. I think we really have landed
- 14:47
- on the proper form factor
- 14:49
- for this type of work
- 14:51
- So let's take a look at a few specific use cases
- 14:54
- just to kind of give you an idea
- 14:55
- of how you can use this tool professionally
- 14:59
- executive use case is a really good example
- 15:01
- I think that can apply to anybody
- 15:02
- in this room
- 15:04
- I talked a little bit about the morning
- 15:05
- briefing that I run
- 15:07
- with my chief of staff bot
- 15:09
- where it is giving me an update
- 15:11
- Each, each morning briefing me on my day
- 15:15
- And again
- 15:15
- it's bringing in information from all of these different connectors
- 15:19
- that I've hooked it up to
- 15:21
- Meeting prep is another big one
- 15:23
- that I use this for
- 15:25
- So
- 15:25
- I am meeting with customers throughout the day
- 15:28
- I'll be honest with you
- 15:29
- with my aging brain
- 15:31
- any of you that have worked with customers
- 15:33
- I don't know, you know
- 15:34
- you can have these great meetings with a
- 15:35
- customer and have a very meaningful experience
- 15:38
- And then by the end of the day
- 15:39
- you have met with six other customers
- 15:41
- and it's all just kind of muddled in
- 15:43
- your head. So I'll go into a meeting
- 15:45
- and not quite remember what we had talked
- 15:47
- about in the previous meetings
- 15:50
- I
- 15:50
- the bot is able to surface all that
- 15:52
- up for me
- 15:52
- and brief me on what we talked about
- 15:54
- Again, it's tying into the granola call transcripts
- 15:58
- It's tying into
- 15:59
- Slack messages that I had with my colleagues about
- 16:03
- this customer. It's looking at Salesforce
- 16:05
- So all of this information
- 16:07
- And for me, one of the
- 16:08
- the first kind of light bulbs that went
- 16:10
- off with Grokbot
- 16:11
- Is how good it is at searching all
- 16:14
- of these data sources. I'll tell you personally
- 16:16
- I
- 16:17
- I find the Slack search to be very frustrating
- 16:21
- our notion, there is so much information in there
- 16:23
- it can be hard to find things
- 16:26
- I found that the first thing that was
- 16:27
- really powerful that I discovered in Grokbot
- 16:30
- was that it's really good at search
- 16:32
- not just from one source
- 16:34
- but from all these different sources
- 16:35
- So the fact that
- 16:37
- I can walk into a meeting and have
- 16:38
- a brief there from my Grokbot
- 16:40
- that has synthesized data from all these different
- 16:43
- sources and got it set up for me
- 16:45
- can be really, really powerful
- 16:49
- let's talk about engineering
- 16:51
- so I know a lot of you are
- 16:52
- engineers in the room
- 16:53
- and the original description of this talk was
- 16:55
- really talking about the software factory
- 16:58
- We have a
- 16:58
- lot of engineers at Cursor now
- 17:00
- that are using Grokbot as their primary surface
- 17:03
- for building
- 17:05
- so when you ask
- 17:06
- Grokbot to build something
- 17:08
- it is going to launch a cloud agent
- 17:11
- so that's how it does all of its development work
- 17:13
- and, any of you that have played around with Origin
- 17:16
- which is our new entry into the source control management space
- 17:20
- with Origin you can actually start working without
- 17:23
- even creating a repo
- 17:24
- but you of course can use it with GitHub
- 17:26
- GitLab, any of the other SEMs
- 17:29
- but I think something else that's interesting about
- 17:31
- Grokbot is that you can actually create
- 17:34
- a team of bots that can cover
- 17:37
- a lot of different personas
- 17:40
- in an engineering team
- 17:42
- So you might have a front-end bot
- 17:44
- you might have a QA bot
- 17:47
- you might have a documentation bot
- 17:49
- you might have a research bot
- 17:51
- and so you can work with
- 17:52
- with all of these bots
- 17:55
- The engineers that are working on the bot team
- 17:57
- they are so bot-pilled
- 17:59
- that this is really the primary tool they're using now to
- 18:03
- to build
- 18:04
- they have moved off of
- 18:05
- like, using the cursor desktop app and
- 18:09
- and using, using Grokbot a ton
- 18:11
- so, so pretty, pretty interesting
- 18:13
- It's just such a simple interface
- 18:15
- and again, the fact that these bots can run end-to-end
- 18:19
- really can be quite magical
- 18:21
- Go to market
- 18:22
- any of you that are here
- 18:24
- working in an early-stage
- 18:26
- startup, founder
- 18:28
- even those of us that are engineers have
- 18:29
- to talk to customers
- 18:31
- we have to sell
- 18:32
- and I just
- 18:32
- I can't tell you how powerful Grokbot can be
- 18:36
- around GTM tasks
- 18:38
- and, and as I'm talking through these
- 18:40
- I want you to know that there is
- 18:42
- we have, a guide section on the XAI site
- 18:46
- that has guides on how to best use
- 18:48
- Grokbot for a lot of these different use
- 18:50
- cases that go really deep
- 18:52
- whether it's engineering, product man- Schmidt, GTM
- 18:56
- so if any of this is of interest to you
- 18:58
- that can be a nice way to kind
- 18:59
- of get inspired. I'll tell you
- 19:00
- for me, when I first started using GrokBot
- 19:03
- it took me a minute to kind of
- 19:04
- wrap my head around how to best use
- 19:06
- it
- 19:06
- And we had a session at Cursor
- 19:09
- within the first week or two of getting
- 19:11
- the tool where people started sharing what they
- 19:14
- were doing
- 19:15
- we have an account executive at Cursor named Krista Letts
- 19:18
- who actually just did a
- 19:20
- a, a
- 19:22
- a workshop that was on Zoom where she
- 19:24
- talked about how she uses GoToMarket
- 19:26
- When I heard her talk about how she was using it
- 19:29
- like, I
- 19:29
- I cracked the code and it completely changed
- 19:32
- the way I use GrokBot
- 19:33
- So
- 19:33
- I would encourage you to check out some
- 19:35
- of these guides. I think
- 19:36
- once you kind of wrap your head around
- 19:37
- some of the ways to use this
- 19:39
- it can be quite magical
- 19:41
- One other thing
- 19:41
- I would say the next big light bulb
- 19:43
- that went off for me
- 19:45
- Is that the bots can talk to each other
- 19:48
- which can actually be kind of funny when
- 19:50
- you see them messaging each other
- 19:52
- So I'll give you another example
- 19:54
- I, as I mentioned
- 19:55
- I do a ton of travel
- 19:57
- I'm traveling
- 19:57
- I've
- 19:57
- been on the road basically for a year
- 19:59
- traveling around the world
- 20:02
- I spend
- 20:02
- hours and hours every month booking my travel
- 20:07
- I'm a bit of a hotel snob
- 20:08
- I'm kind of picky on where I stay
- 20:10
- I've got, you know, certain airlines that I want to fly
- 20:13
- I want an aisle seat if I'm in premium economy
- 20:16
- I want to be in the front row
- 20:18
- I've gotten kind of picky about it
- 20:20
- I have a travel bot
- 20:22
- I have a flight bot
- 20:24
- and I have a hotel bot
- 20:25
- My flight bot knows all of my preferences
- 20:28
- around what airlines I want to travel
- 20:30
- what my, what class I can fly
- 20:32
- and what my seat preferences are
- 20:33
- I've got the same thing for hotel
- 20:35
- what lines of hotels I like
- 20:37
- what's important to me in a hotel
- 20:39
- And then I just talk to the travel bot
- 20:41
- I say, hey, I have a meeting
- 20:43
- next week in Amsterdam. Am
- 20:45
- you know, I'm going to be in London
- 20:47
- I want you to book my travel
- 20:49
- That bot will then look at my calendar
- 20:51
- it'll see when that meeting is
- 20:52
- it'll know the timing of when I need to fly in
- 20:55
- when I'm going to fly out
- 20:56
- and then it's going to talk to the
- 20:58
- flight bot and talk to the
- 21:01
- to the hotel bot
- 21:02
- and both of those bots are going to
- 21:04
- work away at getting the flights and the
- 21:05
- the hotels booked
- 21:07
- and then they're connecting to Navon
- 21:09
- which is the travel platform that we use
- 21:12
- either through the MCP
- 21:13
- server or through the web browser
- 21:15
- So the fact that the bots can talk
- 21:16
- to each other can be really
- 21:17
- really powerful
- 21:19
- Anyway, as far as go-to-market goes
- 21:21
- so many different things you can do
- 21:23
- here to
- 21:24
- automate away a lot of the repetitive tasks
- 21:27
- that you have to do
- 21:28
- in go-to-market, so you really can focus
- 21:31
- on how to best be strategic with your accounts and
- 21:35
- support your customers
- 21:38
- talked a little bit about how bots work
- 21:39
- together here
- 21:41
- you know, a limiting factor, I think
- 21:42
- with a lot of the existing products is
- 21:44
- that they are very isolated in their context
- 21:47
- and in the way that we work in real life
- 21:49
- you know, it's very collaborative
- 21:51
- where we are sharing amongst each other
- 21:53
- And as I, in the example I gave before
- 21:56
- this is really how the bots can
- 21:57
- can work together
- 22:00
- and bots aren't just limited by your laptop
- 22:02
- We have an iOS app
- 22:03
- we have an Android app
- 22:04
- you know, I can kick something off from my laptop
- 22:07
- and then I can leave my hotel
- 22:08
- and while I'm in the Uber
- 22:09
- I can check in on my phone and watch it
- 22:11
- Again, every bot has its own computer
- 22:14
- so I can close my laptop
- 22:15
- and it will keep running
- 22:17
- you can also have routines with bots
- 22:19
- so they can check things or do things
- 22:21
- on a recurring basis that is not contingent on
- 22:24
- your laptop or, Or your, your device
- 22:30
- Bots, as I mentioned before
- 22:31
- are meant to be democratized
- 22:32
- So Grokbot allows for very easy sharing
- 22:36
- Once a bot has been created
- 22:38
- we have a bot marketplace now where you
- 22:40
- can install these bot templates
- 22:43
- I created a closer bot
- 22:45
- So I am, as I said before
- 22:47
- I'm in all of these meetings all the time with customers
- 22:51
- and I'm also often promising them something
- 22:54
- or multiple things in those meetings
- 22:55
- Maybe we're talking about cloud agents and they're
- 22:57
- asking me about security
- 22:58
- and I say, well
- 22:59
- we've got a couple of good security docs
- 23:01
- that outline our security posture with cloud agents
- 23:05
- But then I've got to remember to send
- 23:07
- all those things. I've got to write up
- 23:08
- the emails. So I have a closer bot
- 23:10
- At the end of every meeting
- 23:12
- Granola sends a web hook to
- 23:14
- my Grokbot
- 23:15
- to tell it to
- 23:16
- analyze those meeting notes
- 23:18
- and find any deliverables that I promised
- 23:21
- That Grokbot is also looking at my email
- 23:24
- it's looking at my Slack
- 23:25
- and it is creating a Queue
- 23:28
- stack ranked of all the things that I
- 23:30
- need to do
- 23:31
- And then I can open up that closer
- 23:33
- bot at any time
- 23:34
- and let's say that security document
- 23:36
- I can tap on that or click on that
- 23:39
- and then it will go ahead and draft
- 23:41
- the email to the customer with the links
- 23:43
- It's tied into all of our documentation
- 23:46
- So I don't have to manually write that
- 23:48
- email. I can go in
- 23:48
- make a few tweaks
- 23:49
- hit send, and it can go
- 23:51
- So using the sharing capabilities
- 23:54
- I was able to share this closer bot
- 23:55
- with the rest of my field engineering team
- 23:58
- It's been a big hit
- 23:59
- So this is a great way that you
- 24:01
- can kind of seed your bots
- 24:03
- Like there's a great bot called HaggleBot on
- 24:05
- the marketplace
- 24:06
- right now that you can use to go
- 24:08
- and negotiate. We used it
- 24:10
- with renegotiating our Zoom contract at Cursor
- 24:14
- and it saved us like tens of thousands
- 24:16
- of dollars completely done by the bot
- 24:19
- So sharing can be
- 24:20
- a great way to kind of seed your
- 24:21
- bots and get ideas
- 24:22
- but also share them across your organization
- 24:26
- The key
- 24:28
- to really building a great foundation with your bots is
- 24:31
- a few different things
- 24:33
- giving them access
- 24:35
- so
- 24:36
- using those connectors and hooking them up to
- 24:38
- all the different things that you do
- 24:41
- investing in teaching your bots
- 24:43
- again, just like when you're using agents
- 24:45
- let's say, in cursor
- 24:47
- setting up sort of guardrails
- 24:49
- setting up instructions to supplement the model's knowledge
- 24:53
- to make them smarter and smarter
- 24:55
- so you can just talk to your bot
- 24:57
- and give it information. Bots can create skills
- 25:00
- You can ask a bot to create a
- 25:01
- skill or update a skill
- 25:03
- also in computer use
- 25:05
- you can click a button and it will
- 25:06
- learn based on the way you're working around
- 25:08
- a browser
- 25:09
- So again, these bots get smarter and smarter
- 25:12
- and then delegating between bots
- 25:14
- the idea that these bots can talk to
- 25:16
- each other. You can even create chat rooms with multiple bots
- 25:20
- inside. A lot of our engineering teams are
- 25:22
- doing that
- 25:23
- So in summary
- 25:26
- why
- 25:27
- I think RockBot is really special and worth
- 25:29
- taking a look at
- 25:30
- it is dead simple to use
- 25:33
- I onboarded my 76-year-old mother last weekend
- 25:37
- and this is someone who
- 25:39
- you know
- 25:39
- you look at her
- 25:40
- iPhone and like the text is this big
- 25:42
- She's been asking me technical questions my entire
- 25:45
- life
- 25:46
- She is using Grokbot
- 25:48
- They're always on, they have their own computer
- 25:52
- they use tools just like you do through a web browser
- 25:55
- you don't have to have an MCP or
- 25:56
- an API
- 25:57
- They're really good at finishing complex work end to end
- 26:01
- like my Singapore housing example
- 26:03
- and you've got shareable templates
- 26:06
- so my hope is that out of this talk
- 26:08
- if you all haven't tried out Grokbot
- 26:09
- that you'll be inspired to download it
- 26:12
- check out some of the guides
- 26:14
- give yourself a little bit of time to
- 26:15
- kind of wrap your head around how they work
- 26:19
- and, let me know what you think
- 26:20
- My email's nick at cursor dot com
- 26:24
- I, I, I, I, you know
- 26:26
- I say this in all honesty
- 26:27
- this tool, just like cursor
- 26:29
- totally changed my life a couple years ago
- 26:32
- This has completely changed the way I work
- 26:35
- and it's just improved the quality of my life
- 26:37
- I mean
- 26:37
- it has just freed up my time
- 26:39
- to be a better professional at my work
- 26:42
- to be a better dad
- 26:44
- to be a better husband
- 26:46
- I just, the quality of my life has really
- 26:49
- improved, in every aspect. So
- 26:51
- hope this is eye-opening
- 26:53
- inspirational for you
- 26:54
- have a great conference. Thanks so much for having me
- 26:56
- And, this was my Star Trek moment
- 26:58
- for sure. Thank you
- 26:59
- everybody
Watch these videos on your TV
XTTV brings video content from X to your living room. Download now and start watching on the big screen.