HomeVideos

11 Grok Bot Use Cases That Feel Like Cheating

Now Playing

11 Grok Bot Use Cases That Feel Like Cheating

Transcript

565 segments

0:00

I've been using Grockbot every day since

0:02

it came out, and it feels more personal

0:05

and more capable than any other agent

0:08

that I've used. And to be honest, I

0:10

wasn't expecting it. The UI is built to

0:12

give every bot the feeling of a

0:14

persistent character rather than a

0:16

thread that you can just throw away. And

0:18

I really like it. So, in this video, I'm

0:21

going to show you all of the incredible

0:22

use cases that I'm using Grockbot for.

0:24

Plus, I'm going to show you some of the

0:26

best practices that I've learned in how

0:28

to get the most out of Grockbot. So,

0:30

let's get right into the first use case.

0:32

Email management. If you get a ton of

0:34

email like me, if you have multiple

0:36

email addresses like me, you probably

0:38

spend so much time trying to get to

0:41

inbox zero. And now it's much easier

0:43

with Grockbot. I have a specific bot

0:46

agent that I called the email agent. And

0:49

every single day at 7:30 a.m., it looks

0:51

through all of my emails from overnight

0:54

and from the day before, any that are

0:56

left over. And the first thing it does

0:58

is look for emails that can be easily

1:00

archived without me having to look at

1:02

it. I have dozens of emails that are

1:05

just, you know, like one-off

1:06

notifications and things that I don't

1:08

really need to pay attention to. And so

1:10

it puts those together for me and ask me

1:12

if I want to archive them. So here it

1:14

says overnight leftovers that look easy

1:16

to archive. 14 out of office auto

1:19

replies from my newsletter that goes out

1:21

and then it gives me a bunch of other

1:22

emails that look like they don't need my

1:25

attention directly. It gives me a little

1:28

button I can push. I click that and it

1:30

autoarchchives it. Now, to actually

1:32

accomplish this, you're going to go into

1:34

plugins and you're just going to make

1:36

sure your Gmail plugin is added. You

1:38

click add. A little prompt is going to

1:40

pop up in the thread just asking you to

1:42

connect. You click the button, you

1:44

authorize Gmail, and you're done. It now

1:46

has access to read your Gmail, add

1:49

labels, draft replies, archive

1:52

everything. Then what I ask it to do is

1:54

after those easily archable emails, I

1:57

ask it to look for kind of the loweffort

1:59

ones that I need to actually pay

2:01

attention to and it'll batch them

2:02

together. These are emails that I

2:04

actually need to look at or get some

2:06

information from, but I don't

2:08

necessarily need to reply to them. I

2:10

don't need to spend a bunch of time on

2:12

them. And so I have to blur these out

2:13

because these are my personal emails.

2:15

But that's what is showing right here.

2:18

It's just a few sentence summary of a

2:20

number of different emails all batched

2:21

together. I can take a very quick look,

2:24

understand it, and then just say

2:26

archive. And the cool thing is Grockbot

2:29

will actually learn over time and you

2:31

can tell it, hey, anytime you get an

2:33

email like this, go ahead, don't even

2:35

ask me. You can archive it or put it in

2:37

kind of the lowest effort email review

2:40

batch. Now, once I'm done with that, the

2:42

way that I get through the rest of my

2:44

emails is super interesting. I have it

2:47

go through each one that requires my

2:50

attention, requires some kind of action,

2:52

and I have it read all of the context.

2:54

So, that's not only the current thread,

2:56

but previous interactions with the

2:58

person, and basically summarize the

3:00

whole thing and try to suggest what the

3:03

next action could be, whether that's a

3:05

reply or whether it just needs me to

3:08

make a decision on something. And it

3:10

actually does this incredibly well. And

3:13

again, the more context you give

3:14

Grockbot, the better it can help assess

3:16

each of these emails. For example, I

3:18

connected HubSpot. I gave it access to

3:21

my Google Drive where I take notes and I

3:24

store contracts and everything like

3:26

that. And now Grockbot knows to go check

3:29

those different sources. And then it

3:31

will better understand what actions it

3:33

should suggest for me on each individual

3:36

email. And that's how I do it. And a lot

3:38

of times it's actually just writing an

3:40

email for me, which is quite nice. And

3:43

it sounds like me too, which is also

3:45

quite nice. So with all of this, I'm

3:47

able to get through my email much more

3:50

quickly than I would otherwise. And I do

3:52

get a lot of emails. So this is a really

3:55

big timesaver for me, and it can be for

3:57

you as well. I have a prompt that you

4:00

can give to your Grockbot to accomplish

4:03

this exact workflow. I'm going to drop

4:04

that and all of the other use cases that

4:06

we're going over today down below. Next

4:09

is email classification. Not only do I

4:12

want help triaging my emails, but I want

4:15

Grockbox to actually label my emails to

4:18

help me sort through them and categorize

4:20

them for later. And so I make YouTube

4:23

videos. I get a lot of sponsorship

4:25

emails. And I do get a lot of spammy

4:27

sponsorship emails. And so I actually

4:29

have Grockbot give a score to each of

4:32

those sponsorship emails. And so I can

4:34

see the spammy ones automatically

4:37

archive those. And then for the rest,

4:39

it's basically like a lead scoring

4:41

system. And I did build this lead score

4:45

heristic up over time with OpenClaw, but

4:47

I basically just took that scoring

4:50

system and then put it in Grockbot and

4:52

now it just works. And so you might need

4:54

to work with Grockbot or whatever agent

4:56

you're using to help come up with that

4:58

scoring system, but you can do that too.

5:01

And not only just scoring for sponsors,

5:03

but you can label any types of emails.

5:05

So I get emails from PR companies. I can

5:08

easily just label those. I have internal

5:10

ops and about a dozen different labels

5:12

that I use, which again is just super

5:14

helpful if I ever want to say, "Okay,

5:16

give me all of the highquality sponsor

5:18

emails that have come through. Show it

5:20

to me." It can easily do that now. And

5:22

so you can see I have this set up as a

5:24

routine. And you don't need to set this

5:27

up manually. You can literally just tell

5:28

your Grockbot to set up a routine. And I

5:31

have it set up to run every day during

5:33

the week, every 30 minutes between 8:00

5:35

a.m. and 7:00 p.m. So it's not just

5:37

burning tokens when I don't really need

5:39

it. It'll do a batch job first thing in

5:41

the morning. So you can see all the

5:43

different runs down here. Another one of

5:45

my favorite use cases is actually

5:47

allowing Grockbot to publish things to

5:49

the internet really easily. so I can

5:51

share them with my team or with clients

5:53

or with you all. Here is sponsoring this

5:56

video. I love their product. I use it

5:59

all the time. In fact, I actually don't

6:01

use it. My agent uses it all the time

6:03

and it just knows anytime I need to

6:06

publish anything, it will use here. Now

6:08

automatically, they actually have a

6:10

native plugin now. So, if you just

6:12

search for here now in the plugins list,

6:14

here it is. I've actually had the skill

6:17

installed. So, I'm going to install the

6:18

here. now plug in the native one right

6:20

now. And just like that, you're live.

6:23

That's as easy as it gets. And so when

6:25

you need to publish something, it will

6:26

give you back a public or private URL

6:29

within seconds of needing to publish

6:31

something. So you can see right here, I

6:34

published this little picture of the 3D

6:36

boat that I printed. I said, "Publish

6:38

this." It did it. And then it dropped a

6:40

link right here. Here. Now is completely

6:43

free. You can publish almost anything to

6:46

it. And you don't even need an account.

6:48

And in fact, I'm going to publish all of

6:49

the prompts that I go over in this video

6:52

using here. Now, and I'm going to drop a

6:54

link down below for that. Thanks again

6:55

to here. Now, huge fan. Go install here.

6:58

Now right now into Grockbot. All right.

7:00

Next is calendar management. Another one

7:03

of those things that you'll save a few

7:06

seconds here and there, but over the

7:08

course of a year, this really starts to

7:10

add up. And so, all this bot does is

7:12

it's a bot to interact with my multiple

7:15

calendars. It knows about my calendars.

7:17

it has authenticated into my calendars.

7:20

I don't usually interact directly with

7:22

my calendar bot. I'm usually talking to

7:24

my chief of staff who will then delegate

7:26

it out to the calendar bot. But I will

7:28

explain the chief of staff later. But

7:31

all you need to know is I use this for a

7:33

lot of really cool use cases. And so I

7:35

use this to help me negotiate meeting

7:38

times. So if I have somebody emailing me

7:40

who wants to meet, I don't have to go

7:42

manually look through my calendar and go

7:44

back and forth with them. I simply have

7:46

my calendar bot do that. It checks

7:48

automatically for overlapping meetings

7:50

and helps me reschedule if I ever have

7:52

that. And honestly, I get a lot of

7:54

emails that have multiple calendar dates

7:56

in them. And all I do is have Grock

7:59

check the email and then create calendar

8:02

events for everything in that email.

8:04

Super useful. And again, the way I want

8:06

you to think about Grockbot and each of

8:08

these bots is they should kind of have

8:11

one [clears throat] main task. All

8:13

right. And next is browser use. Don't

8:16

forget Grockbot has incredible browser

8:19

use and it all runs in the cloud. I'll

8:22

touch on the cloud aspect of Grockbot a

8:24

little bit later in the video, but just

8:26

remember anything that you can do in a

8:28

browser, Grockbot can do for you. And so

8:30

a few things that I use browser use for

8:32

are well first shopping on Amazon. I'll

8:35

just tell it to go find me something.

8:37

And it's kind of interesting if you

8:39

remember back, Amazon actually sued

8:41

Perplexity because they said Perplexity

8:43

agents should not be able to browse

8:45

Amazon and the judge threw it out. So

8:48

now agents can freely use Amazon. And

8:51

that's what we're going to be doing with

8:52

Grockbot. So right now I have it

8:54

searching for a 45B tungsten cube, which

8:59

Brian recommended. And if I actually

9:02

open up the computer use right here, you

9:05

can see it's a mere $20,000. All right,

9:09

so here we go. Closest thing on Amazon

9:11

is the 4-in tungsten cube from Midwest

9:14

Tungsten Service. Shout out Midwest

9:16

Tungsten Service. Listed at $20,000

9:19

in stock. Hey, I can get it within a

9:21

week. So there we go. And it's really

9:23

easy. It can find anything for you. It

9:25

can even complete the purchase once you

9:27

sign in with Amazon. And to do that, you

9:29

just go to the browser right here and

9:31

you click sign in, right? Just sign in

9:33

like normal. It works just like it would

9:36

in your own browser. And once you do it

9:38

once, it's going to persist. So you

9:40

don't need to always sign in each time

9:41

you want to use it. You can have it put

9:43

together grocery lists for you. You can

9:45

have it register your car at the DMV.

9:48

All of these tedious tasks that require

9:50

you clicking through websites that might

9:52

take just, you know, 10 minutes, 20

9:54

minutes each. You save so much time by

9:57

just having your bot do it for you. One

9:59

of my favorite use cases is having my

10:01

Grockbot return Amazon items for me. So

10:05

rather than clicking through the whole

10:07

return flow, it just goes and does it

10:09

for me and gives me the QR code for it.

10:12

It can book you gym times. It can book

10:14

you doctor's appointments. I'm telling

10:17

you, it really is incredible and it

10:19

saves a lot of time. All right, next.

10:21

And this came directly from the cursor

10:24

team. They told me about this. You

10:26

should actually be using Grockbot as

10:29

your coding agent, which kind of sounds

10:31

weird, but it makes sense once you

10:34

understand the logic. And so, here's the

10:37

gist of how it works. You have your

10:39

coding agent in Grockbot. It has all of

10:42

this context about you, how you like to

10:45

work, and it can actually call the

10:47

cursor agent CLI directly from Grockbot.

10:51

And again, the Cursor team told me, the

10:54

engineers at Cursor told me this is how

10:56

they're using Grockbot today. And so the

10:59

way that they told me they're doing

11:00

this, and I haven't done it myself yet,

11:02

but I want to explain how they told me.

11:04

They said they're creating a new bot for

11:06

every project that they're working on

11:09

and even bots for the work streams. So

11:13

if you have multiple work streams within

11:14

a single project, you want to create a

11:16

Grockbot for each of those. Then they

11:19

actually initiate it from Slack. So they

11:22

all chat in Slack and they will actually

11:24

kick off different tasks to Grockbot

11:27

which then delegates to the cursor CLI

11:30

all from Slack. And again it has access

11:33

to notion. It has access to email,

11:35

Slack, GitHub, everything it needs. And

11:38

it can actually collect and persist that

11:40

context apparently better than cursor

11:42

can directly. Then if they need to

11:45

generate more notion docs or plans or

11:47

anything, they do that straight from

11:49

Grockbot. It can kick off cloud agents,

11:52

cursor cloud agents directly from

11:54

Grockbot. Grockbot will follow up with

11:57

those cursor cloud agents and make sure

11:59

that it's working and it completes the

12:01

task end to end. And so an example of

12:03

kind of the endto-end workflow, also

12:06

maybe known as a loop, the Grockbot will

12:08

look at a PR and if the CI isn't green,

12:12

it will continue to follow up and fix it

12:14

until it is. And it just knows continue

12:16

doing it until you get a green CI so

12:19

that we can merge. And so instead of

12:20

seeing all of the context, all of the

12:23

text really that you usually get in

12:25

cursor, you get very brief summaries in

12:27

Grockbot. And you can always ask for

12:29

more detail because it is technically

12:31

just cursor running within Grockbot. It

12:34

is very interesting. And next, if this

12:37

sounds like a joke, I am actually

12:39

serious. I have Grockbot ordering food

12:42

for me from Door Dash. Door Dash

12:45

announced its CLI product a few months

12:48

ago, and there is a wait list for it.

12:50

So, if you don't have access to it yet,

12:52

go sign up for the wait list. I'm still

12:54

on the wait list, so I haven't actually

12:56

done this myself, but I know a lot of

12:58

other people who have. And basically

13:00

what they're doing is they're allowing

13:01

their bot to order food for them. Now,

13:04

you can do this through browser control

13:06

with Grockbot today, even if you don't

13:08

have access to the CLI. It's just a lot

13:10

slower. And so, I just typed open Door

13:13

Dash and order me a sandwich any from

13:15

Mendescino Farms. And you can see I

13:17

already set up my Grockbot to use the

13:19

CLI. I'm just waiting to get access to

13:21

it. but it knew, okay, you don't have

13:23

the CLI yet. So, I'm opening Door Dash

13:25

in the browser instead. And so, in a

13:28

second, we're going to see the browser

13:29

open. It's going to navigate. Here we

13:31

go. You can see it already. It's going

13:32

to navigate to Door Dash. It's going to

13:35

look for Menescino Farms. And there it

13:37

is. It's going to order me a sandwich

13:38

and get it ready. Once I sign in, once I

13:40

give it my address, it will have all the

13:44

information, and I'll just be able to

13:45

order it with a quick voice note or with

13:47

a quick prompt. It's so easy. All right.

13:50

Next, this is another one of those super

13:54

powerful use cases that I use every

13:57

single day, multiple times a day. This

13:59

is the meeting summary workflow. And

14:01

basically how it works is I have the

14:03

Fathom notetaker join all of my

14:06

meetings. And so it will record the

14:09

meeting. It will then transcribe the

14:11

meeting. I've connected it to Grockbot.

14:14

Grockbot every 30 minutes will check

14:17

Fathom's API to see if a new meeting has

14:20

been recorded. It will then ingest the

14:23

transcript. It'll write a summary of the

14:26

meeting. It will write up action items.

14:29

So, it will detect different things that

14:31

I committed to during the meeting or

14:33

that the other person in the meeting

14:35

committed to me. It will write those up

14:38

and then it will send that to me either

14:40

in Grockbot, it can send it to you in

14:42

Slack. And I've even hooked up Telegram,

14:45

which I'll show you in a moment. And

14:46

then the next thing, and I haven't done

14:48

this yet, but I think this is really

14:50

going to unlock the power of Grockbot.

14:53

I'm actually going to have it look at my

14:55

list of to-do items, look at all the

14:57

context it has about me, and try to

15:00

figure out if it can actually go

15:01

accomplish one of those items for me

15:04

without me doing anything. And don't

15:06

forget, during the meeting, you can

15:08

actually talk to Grockbot. I know it

15:10

might sound a little awkward if you're

15:12

meeting somebody for the first time, but

15:13

certainly if you're talking to your

15:15

team, you can just say like, "Oh, hey

15:17

Grockbot, don't forget to remind us to

15:19

email that person next week or anything

15:21

like that, and it will read that

15:24

transcript." And remember to have action

15:26

items that you specifically told

15:28

Grockbot to do. So, this is probably

15:30

between email triage and the meeting

15:34

summary. These are my two favorite, two

15:36

most powerful use cases that I do with

15:38

Grockbot today. Okay, so this next one

15:40

isn't really a use case. It's more how I

15:42

use Grockbot. And I set up a bot called

15:45

Chief of Staff. And let me show it to

15:47

you. So I have Chief of Staff as a

15:49

pinned bot. And if you want to pin any

15:51

of these, you basically just rightclick

15:53

and then click this pin button right

15:55

here. And then it's going to show up in

15:56

the top anytime that you need it. And so

15:58

I have this chief of staff that I

16:00

basically use as my main bot that I

16:03

communicate with mostly. So, if I'm

16:06

using the desktop application of

16:08

Grockbot, I'll probably click into one

16:10

of these individual bots, but sometimes

16:13

I just use Chief of Staff directly.

16:16

However, if I have Grockbot plugged into

16:19

anywhere else like Slack or Telegram, I

16:22

am only communicating through my chief

16:25

of staff. My chief of staff can then

16:27

delegate to any other bot that it needs

16:30

to. So, I can tell my chief of staff,

16:32

order me dinner for tonight. It will

16:34

delegate out to my food bot. I can say,

16:36

"Hey, let's triage the emails." It will

16:39

communicate directly with my email

16:41

triage bot, but I'll communicate

16:43

entirely through the chief of staff. It

16:46

is really neat. And so, this is just a

16:47

workflow that I use, but you can figure

16:49

out if you like it or not. So, the next

16:52

pin bot that I use, and really the only

16:54

other pinned bot that I use is the

16:55

personal bot. I really tried to separate

16:57

out my work life and my personal life,

17:00

and that's how I did it. chief of staff

17:03

is everything workrelated, but if it's

17:05

anything to do with my personal life, I

17:07

talk to my personal bot for that. All

17:09

right, the next one, and this is a kind

17:12

of a sub agent or a subbot of my

17:15

personal bot. This is the schoolbot. And

17:18

again, this is another one of those

17:20

super high-V value use cases. If you're

17:22

a parent, you know how many emails you

17:24

get from your kids' school. It seems

17:27

like a huge essay every other day with

17:30

calendar events and things I need to

17:32

bring and remember and it's so

17:34

overwhelming. I now have a schoolbot to

17:38

manage it all for me. It scans my

17:40

personal email inbox once a day at 700

17:43

p.m. It looks for emails from my kids

17:47

schools. It will look for important

17:49

information in them that I definitely

17:51

need to know. It will quickly summarize

17:53

the rest of the email and then it will

17:55

send me a message in Telegram or in

17:58

Grockbot and it will even look for

18:01

calendar events that I need to know. It

18:04

will create a calendar event. It will

18:06

invite my wife to it and it all happens

18:09

automatically now. And so instead of

18:12

reading through these massive emails

18:15

once every other day from my kids

18:17

schools, I now just have Grockbot doing

18:19

it for me. All right. Next is a very

18:22

practical bot that I just started using.

18:24

It is called computer cleanup and it is

18:26

exactly what it sounds like. Grockbot

18:28

not only is a cloud-based agent, but it

18:31

also has access to your computer. And so

18:34

here's what I do. So I basically told it

18:37

to look through all of the files on my

18:39

computer. Look for any that it thinks

18:42

that I should delete. Don't actually

18:44

delete it. This is really important. I

18:46

told it do not delete anything. and

18:49

categorize everything that it thinks I

18:52

should delete into low risk, medium

18:54

risk, and high- risk. And so, I'm going

18:56

to create a routine to run this once a

18:59

week. So, I'll call it weekly cleanup.

19:01

And I'll run it every week. And there we

19:03

go. It's done. And so, every week, it's

19:05

going to look through all of the files

19:07

on my computer and just keep my hard

19:08

drive clean, which is very nice. Here's

19:12

a few things. So, here's 90 GB it found

19:14

that is low risk. 26 GB of cache. 17 GB

19:19

of superhuman cache. Here's cursor with

19:23

a copy of my database locally. More

19:26

cache, npm, unused docker images. It's

19:29

all here and I can just say go ahead and

19:31

delete it. Then we have some medium risk

19:33

380 gigabytes. I have 100 gigabytes of

19:37

work trees which is nuts. And so then it

19:40

just asks me if I want to delete some of

19:42

them and I can go through it with

19:45

Grockbot together. very very useful to

19:48

keep your computer clean. All right, so

19:50

next, if you use OpenClaw, if you use

19:51

Hermes, you've probably grown to like

19:54

doing that through Telegram, and I

19:56

definitely have. Now, Grockbot today has

20:00

an iOS app, but it does not have an

20:02

Android app, which you know, fine. The

20:05

Android app, I have heard, is coming

20:07

soon, but if you want it in Telegram,

20:10

it's not built in natively. It's not

20:12

that hard to set up, but it isn't done

20:15

natively. So you do have to set that up

20:17

manually. Now once again I've created a

20:20

prompt. You can just copy paste it

20:22

directly into your Grockbot. It will set

20:24

it up exactly how you need it. Once you

20:27

copy paste this prompt into Grockbot. It

20:29

will walk you through what you need to

20:30

do in Telegram. All you need to do is

20:33

open up what's called the bot father.

20:35

It's a contact that is in everyone's

20:38

Telegram. You create a new bot. You give

20:40

it a name. It gives you the bot token.

20:42

And then you give that token to

20:43

Grockbot. It sets up everything for you.

20:46

It works quite well. It just took a few

20:48

minutes of massaging to get it to work

20:50

for me. But all of the issues that I ran

20:52

into, I have solved and it is included

20:54

in the prompts that I'm going to give

20:56

you down below. And so now I can access

20:58

my chief of staff Grockbot anywhere that

21:01

I am from Telegram really easily. Plus,

21:04

if you're on the go, you can drop a

21:07

voice note in your Grockbot Telegram and

21:10

it'll behave just like a voice note to

21:12

any assistant. it'll just work. So, here

21:14

are my thoughts on Grockbot as a whole.

21:17

One, it is a bridge between just asking

21:21

Q&A with chat GPT, so you're just asking

21:23

a question, getting an answer, and more

21:25

complicated coding agents like Codeex

21:27

and Claude Code. If you're not technical

21:30

and you don't want to think about using

21:33

some of these coding agents, Grockbot is

21:35

such a nice middle ground. It really

21:37

simplifies the interface. A lot of the

21:40

stuff just works. Sometimes it gets a

21:42

little bit complicated, but for the most

21:44

part it does just work. So, it is very

21:47

powerful and I've been really pleased

21:49

with it. It is definitely a big part of

21:52

my daily workflow. Now, I reviewed

21:54

Grockbot in full. Go check out that

21:57

video right

Interactive Summary

The video highlights the powerful capabilities of Grockbot as a personal assistant by sharing various practical use cases. The creator demonstrates how to automate daily tasks such as email management, calendar coordination, browser-based shopping and administrative work, coding assistance, and even school-related communications. Additionally, the video provides insights on how to set up specialized bots like a 'Chief of Staff' for delegation and integration with platforms like Telegram, emphasizing that Grockbot acts as a versatile bridge between simple chatbots and advanced coding agents.

Suggested questions

4 ready-made prompts