MindStudio 2-Hour Free Workshop (10/28)

Uh

so again, not sure new folks that are

here uh whether you just stumbled across

Mind Studio because you know Ken told

you or or you know uh you've heard about

us before but the way I describe Mind

Studio in sort of a sentence is Mind

Studio is a platform for optimizing

business processes using AI. What kind

of business processes?

any kind of business processes,

sales, marketing, you know, the

executive, finance, IT, you name it, you

can optimize those processes using AI.

Uh, Mind Studio makes it easy to learn

to build AI agents. There are no

technical skills that are required at

all. anyone of any age, from any

background, can learn to build Mind

Studio AI agents in a relatively short

period of of time. And then once you've

learned to build these AI agents, you

can very rapidly build AI agents. The

average build time takes anywhere from a

few minutes to maybe an hour. It's rare

that you have to spend more than an hour

on building really powerful agents that

help you do what? automate things you do

manually now. Okay, what are those

things? Using the computer, using your

phone, you know, interacting with SAS.

Those are the things we do as knowledge

workers. We can now automate a bunch of

those things. Either completely automate

them so we don't have to think about

them anymore or partially automate them.

So, they're doing the heavy lifting and

we're just sort of sitting back and

saying, "Yeah, good job." You know,

change that. Okay. But it's driving and

doing the heavy lifting. Okay. Uh, Mine

Studio is quite popular. There's over

200,000 of these AI agents that have

been built and deployed in all kinds of

situations. There's like individuals

using it. There are small businesses,

enterprises, government. This is the

British tax authority, the British, you

know, internal revenue services we have

in the US and a bunch of others. Uh and

while these companies here are not yet

enterprise customers, there are mine

studio agents running in all of these

companies brought in by employees using

these tools. Okay. Uh we are

venturebacked. Uh if you want to see

who's invested, there's like a bunch of

interesting folks. I used to work at

Google. I was CTO of MySpace Music if

any of you remember that. Like you can

kind of look me up if you don't know me.

Uh it's my third venture back company as

a founder. So far here we've raised 36

million. Um the way I kind of again

describe things to like to lay persons

if I meet somebody at a party and say

like what is it that you do? I say well

listen whatever business you're in okay

does something and it's got these things

we call business processes and and they

work right like you've got a business

they work but you know business owners

and managers leaders etc are always

thinking like how can I make these

processes work better? How can I make

them more work more efficiently? How can

I get them to produce better positive

yield? Meaning, how can we make more

money?

Uh, and this is a constant thing that

businesses are asking, right? And now

they're starting to more and more ask,

how can I make my business run more

autonomously?

Why? Because of these new, relatively

new generative AI models that kind of

started with chat GPT and Claude and now

there's like lots of them. And indeed

these new technologies, these new large

models uh allow uh for the automation of

things that could have never been

automated before now. Okay. And and so

now is a great time uh to to start

thinking about how can I automate the

things I do manually to make myself or

my team or my company run dramatically

more efficiently. Uh the people who know

best where there are inefficiencies

in any company are the people that are

doing those inefficiencies. Meaning the

people that are doing the work manually

those are the people that know where

their inefficiencies. It's not some

consultant. It's not some SAS company

that approaches you and says, "Oh,

license my SAS and then you'll be more

efficient." Okay? Don't fall for that.

It is the people on the ground that are

doing the work that know where they

spend their time and and we believe that

if you simply train those people in this

case you yourselves are getting trained

train people to recognize what can be

automated easily

and and you can you can gain that lens

very quickly by understanding what mind

studio does how it works. You can very

quickly start to see, oh, that thing, I

can just automate that now. In the next

five minutes, I can just automate that.

Never think about it again. What? That's

crazy. But you get that lens, you know?

Or this thing, I can spend an hour and I

can build myself a little tool that when

then whenever I need to do this

repetitive thing that I do every day or

every week, whatever, I'll just push

this button and it will do it for me.

And that'll save me like hours a week of

mental anguish and and and time. Great.

You do that. Okay. So you can develop

that lens very quickly and then again

because mine studio is such an easy tool

to use uh you can just then fix the

problem. Not only can you detect the

inefficiency well you can patch the

inefficiency you see and and that's the

great value of it. Um and so yeah mine

studio is kind of this comprehensive

solution which is like you can learn to

build yourself and do it yourself. If

you don't want to do it yourself you can

hire some certified AI agent builders.

There's a bunch here with you right now.

Okay. Some are available, some are not

available. Some we've hired like Lewis

here who was an independent and we

poached them out of the community. And

so, uh, take a look at Lewis's LinkedIn

profile. Lewis, maybe you could post it

and and look at his posts and then you

will realize how he got poached. You can

get poached too by many many companies.

Uh, so do what Lewis does.

Um, okay. And I'm not going to spend any

time on this, but like there's places on

our website you can find this. Lots of

people from lots of important companies,

you know, think this platform is the

cat's meow and think our boot camps are

an awesome way to learn and like you're

in good company. Okay. Uh what you're

experiencing now is this 2-hour workshop

that we do. Most people learn to use

mind studio without any workshop. They

just look at our YouTube tutorials and

follow our documentation. They just

learn. So like, you can do it yourself.

You're doing this now. You can come back

to one of these. We'll do another one

next week. You can come back. They're

all a little different. I kind of don't

really have a a good script for them. Uh

so you'll learn new things hopefully

every time. Uh you can also join these

boot camps that we have. Uh these are

paid uh things, but you get a tremendous

amount of value. The next one's coming

up uh this next Monday and Tuesday.

They're full days uh 8:00 a.m. Pacific

to 9 to 9 to 5:00 p.m. Pacific. You can

take breaks if you've got like meetings

to take. You can go in and out, but

generally try to like allocate that

time. It's really valuable in 2 days. We

can make you an expert. There's still a

bunch more to learn, but like you can

get level one expert. Some folks here

are level three experts. again can show

a small show of hands like a bunch of

level three experts obviously the list

okay um and we do custom team training

so if you're part of a team if you're a

leader of a team these things are

absurdly valuable because we can take a

team and in a very short period of time

you know half a day one day two day you

pick level up the entire team and then

the rest of the week the team just like

transforms the orc and and so these are

just an insane value that you can just

get now. Okay. Uh and then yeah, let's

just start playing with the platform.

I'm going to kind of skip through a

bunch of these here quickly. So, we'll

use this M studio uh agent builder. Uh

we sit in a bunch of environments that

are security, privacy, compliance,

sensitive. Again, the British tax

authority uses us in HR. For some

reason, I I think that's like the the

best punchline I've got on like getting

people to say to think like this is like

I've got this. If it's good enough for

them, it's good enough for them. I don't

know. Tell me if that makes sense to

you, resonates. They very, very much

care about privacy, security,

compliance. It's all over HR. If they're

fine with it, I suspect you don't have

to worry too much about it. But if you

want to worry, you can visit our trust

center and you can look at case studies

and all that stuff. You know, I don't

want to No, I do want to get you guys to

stop worrying. Actually, I was going to

say I don't want you guys to stop, but I

do. I don't want you to worry. Okay. Uh,

Mine Studio integrates with over 200

models and and you can just start using

them. You don't you don't need to go

create your own accounts. You don't need

to go get your own API keys. You can

just start using models from all kinds

of providers. These are just small list

of them. There's many more. And you can

mix and match and you can compare them

with one another in real time against

one another. What's the best model for

this part of the workflow and all of

that, right? Oops. And uh uh I'm going

to skip this vibe coding stuff. Uh but

recently uh thank you Lewis and Luis and

I think other folks. It's a lot of

people creating content. Thank you. I'm

I'm excited to always watch it that are

kind of going into this and showing

people just how crazy powerful a bunch

of these things that I'm not going to

teach you today are. like the ability to

vibe code completely custom frontends

for your AI agents and completely

custom, you know, outputs and and

artifacts they generate. And so there's

a lot of really awesome stuff in here.

Um, but we're not going to get to it. So

sorry, TE's, you can find it. Maybe

later Lewis can send you some links. Uh,

but this I do want to spend a couple of

moments on. So, and we'll demo these um

today here. Not all of them, but I'll

show you a bunch of them. So, with um

Mine Studio, without writing any code,

zero code, you can easily build in

minutes

uh full stack applications

that you can use from your phone or your

computer or your tablet. They've got

screens and forms, completely custom.

They can be pixel perfect, look like

anything you've ever seen online or on

applook interfaces, front end

interfaces. The back end is AI,

a bunch of different AI models working

together, doing whatever it is you want

them to do. Extraordinarily powerful,

countless use cases. You have the

ability to build applications that run

on mobile and desktop that do anything

leveraging AI with custom interfaces.

It's hard to explain what that means.

But imagine that you're understanding

that there's countless use cases for now

you being able to build these

applications. It is dramatically faster

and easier to build these applications

using mine studio than let's say if you

played around with like lovable or

replet or bolt. Those are for building

potentially different types of

applications in certain instances. But

most things that you might want to build

would be much faster and easier to build

with mine studio than those other

platforms. Although we do get frequently

integrated with those and so we pair

well together. Well, people will will

use lovable replet or bolt to build some

kind of a thingy and then they're going

to use mine studio as the AI backend to

do it. Okay. So, um so so that's that.

This is really powerful. We're going to

build one of these. We're going to build

a simple one, but you'll get see just

how easy it is to do it. Uh it's also

really easy to build applications that

don't have a front end. They're just a

back end. Well, why would you do that?

Well, because they can be run

autonomously on a schedule. And so they

can, you know, wake up at a certain time

or a certain uh uh date or, you know,

every minute or whenever you want a

schedule and then do something and then

they might trigger something else or

they might send you something or they

might simply check if something is

happening and if it is then they will do

something. like tons of applications for

building these uh autonomous agents that

are driven by schedule. Uh the first

agents we're going to I'm going to show

you as these demos coming up here in a

moment and then we're going to build the

first couple are going to be these

browser extension agents. These are

extraordinarily powerful. We spend most

of our time as knowledge workers in the

browser, you know, using different SAS

products through our browser and

therefore having the ability to click a

button and and then take that

information that's in the browser and

send it to an agent and have it do

something. Countless applications here.

These things are um I I call all of

these things, by the way, superpowers

because I kind of have no other better

sort of name to call them. That's what

they feel like. Um you can also again uh

as easily as you can build these things

you can build applications that are

email triggered and so when they're like

that they automatically get an email

address and then for example uh again I

get into these situations all the time

perhaps it's cuz I'm CEO and I get see a

bunch of things perhaps it's I mean I

used to get into I guess in other roles

that I've had where I get these like

threads that are like 20 messages 30

messages and I I just can't get my head

around what has been happening in this

thread. I can simply forward this to an

email a uh yeah an email agent, email

triggered agent and it can sort of

unpack the whole thing for me and tell

me the most important question. Am I

blocking anything? Is there anything for

me to do? What are my action items? And

if the answer is you have no action

items, that is amazing. That is a

massive unlock. I would have spent way

too much time trying to figure out that

I've got no action items. Now I don't

have to spend any time doing it. I just

like copy my agent says don't worry

about it, dude.

or andor it can show me all the action

items for everybody else that's on the

threat and so I can see the people that

I work with what are their action items

what things have been agreed upon what

things might be you know uh potential

issues like it can do kind of thinking

it can do analysis what's happening in

this thread are you dropping the ball

are you again countless applications

that unlock like massive value and then

finally you can build AI I agents that

are programmatically driven. They can

expose APIs. Uh they can expose web

hooks. They can be deployed as MCP

servers. Even if you don't know what an

MCP server is, you can deploy MCP

servers. Okay. And and that's awesome

because you know MCP servers are all the

rage and there is no easier way to build

and deploy MCP servers than with M

studio. Uh last slide. uh we when we get

compared to other companies nine out of

10en times we get compared to a company

called N8N

okay the other 10% we get compared to a

slew of companies these are the dominant

ones the general consensus comparison is

if you've used these platforms mine

studio is easier to learn and then

easier and faster to build and

specifically more instrumented

for building AI powered applications

These platforms have been around much

longer than we they bolted on AI. NAD

has done a great job bolting on AI.

Uh but we are newer. We came out of AI.

And so we just do AI powered automation.

So if you want to do AI powered

automation, Mind Studio is a no-brainer.

If you are already using these

platforms, that's typically when people

start to think, well, I'm already using

these platforms. Should I just stay with

them? Should I add mine studio? Should I

replace them with mine studio? Then you

get a bunch of questions. Again, we

frequently get paired with these

platforms. We in fact even have

integration blocks directly in mind

studio. It makes it easier to pair with

these specific platforms. So, you know,

this is a great partnership thing as

well. Okay. And then you can see online

people talking about how Mind Studio is

sort of dramatically easier than N.

Spent 95% less time debugging errors.

Works without wanting me to pull my hair

out. If any DS Linux, Mind Studios, Mac

OS, this sort of a common thing. Okay,

enough of that. I really hate

presentations.

Uh, I'm not sure what I hate more,

giving them or getting them. Actually,

think I hate getting them more than

giving them. So, I apologize. No more.

Uh, let me show you some rapid fire

demos of things, okay? And then I'm

going to show you some superpowers.

So, this is a a long read. Now, this is

about the movie industry, but that's

irrelevant. What I want to point you to

is we I proposed, tell me if you think

I'm wrong, we spend our lives as

knowledge workers reading a ton of

things trying to extract

data points and and meaning and and sort

of cluster them and bucket them and do

and that's what we do. We spend our time

parsing this kind of stuff using our

mind, our mental energy, our time.

That's a lot of work. Okay, I don't do

that anymore.

I have a little button. In fact, I've

got a bunch of little buttons, right?

I've got a little thing here that's a

button holder, Chrome extension. I can

open this thing up. It's got a bunch of

little buttons. These are my favorites.

We call these pinned. I also have

hundreds of other little buttons

available for me here, but for now I'm

going to stick with the one I've got 10.

Uh so whenever I find something where

before I would have had to like read it

and spend time on it, my new approach is

ain't nobody got time for that. And I

just push this button and then I do

whatever it is I do. You know, I sip my

my coffee. Cheers.

I um I talk to my friends.

I I I I don't waste my mental energy

parsing this junk. I have my agent spend

the time doing it. And then when it's

done, I parse all of these things.

And it's still going. It's a long read.

Might take it 30 seconds, might take it

45, might take it a minute. It's done. I

can then copy this. By the way, if I

want to put it somewhere, I can share

it. I can also continue if I wanted to.

And I could sort of say uh extract all

entities

mentioned

and now it will go and it like give me a

list of people and you know companies,

organizations, films, whatever. That's a

superpower. See? Okay. And so that's

cool. So I can do this on any page on

the internet. Uh minus one. There are

some pages that sort of intentionally do

all kinds of things to sort of obfiscate

stuff where you might have problems. But

the vast majority of pages you can do it

uh some of those pages include YouTube.

YouTube is the second largest search

engine. I would argue that YouTube is

the video web. It's an entire web. Uh

Stan, yes, you can extract emails from

LinkedIn, but not directly from

LinkedIn. Uh YouTube is the And I'll

show you guys later how that's done.

YouTube is an entire web. It's an

amazing source of information. Okay? And

this is a conversation with Jeff Bezos.

It's two hours. I've run this on 10-hour

videos. I'm going to click the same

button. So, in the past, if I wanted to

understand what these guys were talking

about, I would try to watch this at two

times speed. And so, that would be an

hour of my time and attention. The

biggest thing is attention, mental

energy. I can't be distracted. I have to

pay attention. They're speaking quickly

at two times speed. I've got it

captured. I got to pause it. Maybe I got

to take a note. I got to remember this.

Like it's a lot of effort to try to get

like all this awesome

data out of this video. So on one hand,

video is an amazing an amazing medium.

On the other hand, it's got this like

fundamental flaw because of its like

linear nature.

But now this this gets rid of that. And

so I can read YouTube videos in a matter

of a couple of minutes that would take

mere mortals hours to digest. So what

does that mean? Well, that means that if

I spend an hour wanting to learn about

what's happening in my industry via

YouTube, well uh in an hour I can

consume an order of magnitude,

multiple orders of magnitude more than

any of my competitors can. That's a

superpower. Okay. This same thing works

on any page. Again, I keep sort of

making my joke of I I don't encourage

anybody to go to Twitter and I don't uh

but I still I still haven't changed.

I've been too busy or too lazy. If you

did go to Twitter and you want to

understand like what's happening without

sort of getting traumatized, uh you can

push the same button. It doesn't care

what it's on. If it's in the browser

here, it's going to try to grab it.

Again, vast majority of time it can do

it. And now in this case, it will tell

you, hey, tariffs suck. Um the same

thing works with uh again in a different

way a similar kind of an approach is to

be able to do things like this. So again

this is on uh Twitter but I could just

as easily do this on LinkedIn. I could

just as easily do on somebody's page

where they are going to be speaking at

an event and it's got like a long bio

about them and it says something about

them. I have a special button. Again,

this is called Twitter profile analyzer.

Should really be called people analyzer.

But I can click this button and what it

does is it again kind of like looks at

all of this and helps me better

understand who is Gary, what does he

care about, who does he care about, who

does he communicate with about what

subjects, who might be the people that

influence Gary. So if I want to go and

build a relationship with Gary, I want

to sell Gary something, etc., I have an

easy way to be able to get that data so

that I can better understand. Yeah, I

can be more successful

at connecting with Gary success. Okay.

Uh the same thing works on PDFs again,

websites, YouTube PDFs, 182page

financial disclosure. If you if you told

me to go and and parse this and and do

something with it, my head would not

figuratively, but I think literally

explode. Uh but now it doesn't. I just

push a button. Got to wait 2 minutes.

Probably cost like 40 cents to run. 40

cents for having still having your mind.

That's pretty good. Uh okay. Uh two more

things I'm going to show you. Then we're

going to build some agents. So this

thing

keeps changing because the uh people

keep changing jobs but let's take this

one maybe. Yeah. Okay. So this is a job

uh again before this before running this

company. I was at Google on the product

side. I was CTO at MySpace. I built two

other companies. I'm technical and so

I'm certainly qualified to be an SVP of

engineering for this appolio company.

Uh, but the problem is how do I get them

to know that?

Okay. And and in fact, not how do I get

them to know that, how do I get their AI

to know that? Because when I send an

email now with my resume to a company,

there's a meaningful chance that before

any human sees it, an AI is going to see

it. And the AI needs to know whether I'm

a good fit or not, does it let me

through or not. And so in order to

convince that AI or that human that I'm

the right candidate, what I need to do

is I need to take the things that they

care about, which is what they tried to

articulate in this thing called the job

description, and I need to map my talent

and experiences, my skills and

experiences to their thing. I don't need

to lob my resume at them and then have

them try to figure out how do those

things map. I need to do the work to map

them. Well, that's a lot of work. We

call those things cover letters. I've

always hated cover letters. Actually,

even more than PowerPoint presentations,

which I hate a lot. And so, again, if I

was to apply for jobs, again, I would

certainly use an agent like this. I can

simply give it my resume. Now, it knows

about the job description. It knows

about my resume. And it, while I'm

hanging out with my friends, it can

write a custom cover letter point by

point enumerating why I am the best fit.

I'm also the best fit for this one and

the best fit for that one and for this

one. But the problem is I would have

never applied to all these jobs because

writing cover letters takes a long time.

It's hard and I don't have the patience

for it. So, I wouldn't have done it. But

now I can. That's amazing. Yeah, it's a

superpower. Um, okay. One last thing I

want to show you before we jump in. Um,

so, uh, I've now raised, uh, just over

$140 million for these three companies

that I founded. uh this one so far 36

the last one was 70 the one before was

34 conveniently like just 140 uh so I

have a broad network of VC friends and

contacts but once in a while I like to

engage in a activity I call VC

collecting I collect VCs in this case in

an air table database but that doesn't

matter you can collect them into HubSpot

again these don't have to be VCs you can

be thinking about this as being like

leads for example right or whatever it

is and here I've got the name the photo

the current job, current company,

professional summary, their education,

their key experiences, their latest

post, in this case on LinkedIn, their

LinkedIn profile, and their email

address. The Stan was asking, can you

extract it from email? Yes, Stan, you

can extract it from email. Here it is.

Now, not always, you know, some don't

have it, but most of them, as you can

see, have it. This database gets built

in this way. Uh, so typically kind of

the most dominant way. Whoops.

that uh I do this is what is the

problem? links don't work anymore

is uh by being let's say on LinkedIn and

whatever being somewhere on some page

where I might not have collected yet

before and I could see this Chris

Skeller and again in the past my

workflow would have been like to kind of

read about this person and to like what

are they posting about and and see like

their perspective if you're going to

reach out to them everybody's reaching

out to them if you want to be successful

you've got to connect with what they

care about you got to come at them from

their angle. You can't just like lob

messages at them. That that that's

silly. And then so I would like spend

time trying to understand this person.

Why? So I could craft for them in 300

characters some message that would get

me a meeting with them or a connection

with them. Well, that's a lot of work

and a lot of time. And so I don't do

that anymore. I instead have a little

button that I press and that's the

button. And by the way, since this AI

agent actually doesn't return anything,

meaning it just puts him in my thing, I

can sort of continue to move on and I

can say,

"Yeah, sure. This person looks good to

me also."

And then, uh, you know, here's another

general partner at a family office. Why

not? Okay. Now, typically I'm not sort

of like this with VCs, you get the

point. this like for the demo purposes

I'm doing this like I I want to see do I

want to pollute my database with them or

not do I want to talk and anyway that's

it and then it just like gets all their

info everything comes from LinkedIn

except the email address the email

address gets enriched via another

service but again the mind studio agent

just does that for me

and now I get their email addresses and

then finally you might notice that right

here this education thing has these like

curly braces

What is all this junk? These like square

brackets here. Why? And here. It's

because this database actually isn't for

me. This database is for another agent I

have that when it's running connects to

this database, picks a person uh

understands them from all this data that

I've got about them and crafts a custom

email to them and sends it to them and

then waits for them to respond. And if

they don't, then it crafts an a

different email to them from a different

angle. Okay? And in a s I call I

lovingly call it my VC harassment agent.

It harasses VCs until they say yes,

let's chat. And then it hands it off to

me and does it. And so again, most of

the time it's not running because with

great power comes great responsibility.

And um uh but but you can do that kind

of stuff and that's amazing. That's a

superpower. Okay. Uh, so I'm going to

quickly

uh build a couple of agents, then I'll

take questions. Actually, I'll stop here

for a moment. Luis, is anybody

desperately needing to ask a question?

>> No, I think you're going to get to this,

but uh folks are asking if there's a uh

a place with like pre-made agents that

they can use right away.

>> Yep.

I will show you that place. And then um

I'll also show you that because it's so

fast to make agents with mine studio

that even if there are pre-made agents,

you might find yourself just making them

from scratch.

But but however you want to do it, we'll

show you both ways. Okay. So I'm going

to go to mine studio. I'm going to say

my workspace.

I'm going to say build new agent. And so

the agent we're going to build

again I'll just use continue use this as

a demo because it's just a giant thing.

We're going to build this agent right

here. Like when we click it, it's going

to grab all this content, it's going to

send it to this agent.

This agent is going to kind of go

through it with a fine tooth comb and

get us Come on, agent. Guys, still with

me? I've had some network issues here

today. Um, and uh let me try that again.

And and then it'll give us all of these

bullet points uh about any piece of

content. It's the one I demoed for on

YouTube and all. Okay, while it's doing

whatever this thing is doing, we'll just

make one and and we'll run it. Uh, I'm

going to say build new agent. I'm going

to give it a name. I'm going to call

mine TLDDR. I can also give it an icon.

I just need to choose an icon and and

upload it. And I can do that. I'm not

going to do it. Uh, I can also choose

all kinds of other things about it like

who has access to it. Is it just me or

is it people in my workspace, team

members? Is it specific people? Is it

public? For certain kinds of agents, you

can embed them. This one's going to be a

Chrome extension, so it doesn't get

embedded. So, like a bunch of things,

but here I'm just going to leave all

these things the same. And I just gave

it a name. So, so far all we've done is

we said new agent. We gave it a name.

Cool. Agents are made of workflows. You

can have one, you can have many. This

one's just going to have one. Workflows

have a start and an end, and they have

things in the middle. On the start, we

need to choose a trigger mode. Okay, run

mode. Uh, there are different trigger

modes or run modes. The one we're going

to choose is browser extension. You can

also choose these others. Again, we'll

play around with some of these others.

So, I'm going to say browser extension.

As soon as I did that, this thing showed

up here. And what this means is because

this platform now knows that we're

building a browser extension. It says,

I'm also automatically

going to give you all of these things

called, you know, variables. So

everybody know what a variable is. A

variable is just a way to store data in

a label. Okay? And so it's going to give

us a label called variable called URL.

And in it, it's going to store what?

It's going to store the URL of the page

on which this thing was clicked. that's

going to be in that URL variable. Uh

it's going to give us the metadata which

lives in the document. It's going to

give us the page content. Page content

is all the visible text on the page.

That's going to be stored in page

content. We're actually going to use

that in a moment. And it's going to give

us other things. Okay? And so um and so

we have these automatically. Again, so

far all we've done, we said new agent,

we gave it a name, and we said browser

extension. Now we're going to click

plus. And in this case, I'm just going

to choose this first thing here,

generate text. Um, and another name for

this block. These are called blocks, by

the way. And as you can see, the way

they work is like when you click on a

block, you configure it on the right

hand side. You click on it, you

configure it. You click on it, you

configure. That's how it works. And so

this block, another name for it could

be, you know, interact with an AI model.

Yeah. Uh large language model. And I'm

going to give it a prompt. I assume

everybody is like prompted something

like chat GPT or claude. And so by the

way, you can choose what model you're

using here at the bottom. My default

model that I've configured in my

workspace setting as being default is

claude 37 sonnet. But as I mentioned, we

support over 200 models. If you wanted

to use another model and you simply say

use or you can click on it and read

about it and evaluate it and do all

that. I'll stick with claw 37 on it.

It's fine. And here I'm going to give it

a prompt. And the prompt I'm going to

give it is extract all learnings. You

can give it any kind of prompt. You can

say, you know, pull out all the key

points of this or enumerate all the

claims being made like countless things.

You play around with it. I really like

this prompt. It's got a bunch of in my

experimentation. It's the best way to

get like TLDDR of something. And then

I'm going to type these double curly

braces.

And then that's going to give me a drop

down of available things. So I can use

variables. I'm going to choose that page

content variable. Remember over here

because we said that this was going to

be triggered by browser extension. We

got this page content. The page content

is all the visible content on the page.

That's what lives inside the page

content variable. Oops. That's what

lives in that content variable. And so

when we type this in, when this runs,

the model is going to see the words

extract all learnings. And then it's

going to see all of those words that are

on

this page right here. That's what it's

going to get. And then the output's

going to be displayed to the user. And

and that's fine for us for now. And

that's it. We're done. So again, we gave

it a name. We chose browser extension.

We typed the command. And we just gave

it all the content. And now I'm just

going to say publish.

And now I have an agent. Now if I want

to, I can click this button to open the

agent. And it says, look, this agent is

a browser extension. You said it's

triggered with the browser. You can run

it, but it's better run from the browser

extension. And so I'm not going to run

it this way. And so I'm going to go back

here. And so these are my favorites.

They are pinned.

This one we just made, so it can't be my

favorite yet. So, I need to click on

this M and that's going to open up the

side drawer.

What is happening with my computer?

Um,

sorry guys. Let me uh let me close some

of these tabs. Uh, let me

here. Let me actually let me close

Chrome here for a second, I think. May

have been because I tried to upgrade

Chrome maybe and didn't do it. Uh, leave

site. You guys are not going to leave

me. Cool.

Uh Chrome

mute all. Thank you.

>> Okay, let's try that again. Apologies.

Uh, and so I'm going to go to my demo.

I'm going to find my uh man, my mouse is

doing awkward things, too. Uh and then

I'm going to click on this button

and

o

you by chance experiencing there it is,

man. I don't know what's going on. Okay,

sorry about that folks. Again, could be

internet traffic. Who knows? Okay,

there's our little TLDDR or I'm just

impatient, but typically these things

are much faster. Uh, here's that TLDDR

button we made. And so I can click run.

And then again, it's grabbing all of

this stuff. It's sending it to this

agent. And this agent is going to go

through it and it's going to extract all

of these bullet points. That's what we

get back. A bunch of bullet points.

Yeah. And again, the same agent that we

built also works on YouTube and works on

PDFs and works on Twitter and works on

whatever website you're on. Again, minus

one. Uh, and that's really powerful

because again, I took a couple of

minutes to build it, but we'll soon see

that you can build these things

obviously even faster by doing this. So,

let me build you another agent. Now,

instead of building it from scratch, I'm

going to go over here to build. And

build has all of my agents I've

published in my drafts. Okay. And so,

here's that TLDDR agent we just built.

Now, I'm going to click on these little

three buttons, and then I'm going to say

duplicate.

And so now we just made a copy of it.

Okay. And so I'm just going to give it a

name and I'm going to uh build an agent

called stock analyst. Okay. And so I

just changed its name. I'm not going to

change anything about this agent except

for this.

I said, "Do a fundamental analysis on

this stock. Now, I'm going to publish

it."

We just built a new agent. Well, let's

try it out.

We can be on whatever some page, Yahoo

Finance. Let's say it's got a bunch of

information about a stock. I now have a

new little button that I made.

Hopefully, we won't have to wait too

long for it.

There it is. I can say run.

going to grab this page and it's going

to do the things that I would need to do

manually

and she's going to do it for me. Okay?

And it's going to help me understand,

you know, some performance strengths and

trajectory and strengths and weaknesses

and, you know, that's freaking amazing.

Again, I don't know what to call this

thing other than a superpower. Like what

what did we just build in what 30

seconds build? It depends on who you

are. You'll build different buttons. And

again, even if this is all you do, you

build the TLDDR agent, you make a copy

of it, and you just change what this

thing says and publish it. You can make

countless superpowers for yourself, your

team, your company. You can wow people.

It's amazing. Okay? Do it. Think about

it for your own thing. Do that. That's

what you should do first. Okay? But why

stop there, right? We don't stop there.

Couple of things I want to show you

here. One is here I wrote all of this,

you know, manually. Uh, but we have a a

thing built into Mine Studio. So instead

of writing the prompt manually, I can

click right here at the bottom right of

this little text area, edit in full

screen. And so this just lets me edit

the prompt in full screen. Also here

you'll find this little button called

generate.

And here I can say like provided

with a stock

do a fundamental analysis. Yeah. Again,

you can be very dirty like I've got this

capitalization. I'm going to leave it

in. You can misspell everything. Doesn't

matter. It's not a very good prompt. Uh,

but what this thing is in charge of is

taking this crappy prompt or I call it

lazy prompt and putting in some effort

and making it kind of more structured so

that I can then take a look at it and

say, "Yeah, but I didn't mean that. I

don't want that. I want it to be a

different way, etc." And so it can take

the first pass or the first and last

pass at writing this prompt.

And now I've got a much more

sophisticated prompt that I can use. And

again, I can edit it, I can change it, I

can do whatever. And so this like

generate prompting thing is

extraordinarily powerful in being able

to do it. By the way, uh, as a novelty,

this thing right here is just a Mind

Studio AI agent.

It was built with Mind Studio

and integrated back into Mind Studio in

order to help people build things with

Mind Studio. We've got a bunch of these

kinds of capabilities in all kinds of

parts of Mind Studio where we've used

Mind Studio to build Mind Studio and and

you get kind of this recursive inception

thingy going. It's it's pretty crazy.

Okay, that's Stock Analyst. Uh

and then one last thing I'd like to

build for you quickly and then I'll take

questions. Okay. Um just one more.

This is uh how you get in trouble with

Mind Studio. You might be up at like 2

am. You need to go to sleep and you're

like, "I've got another idea. Just one

more. It happens to me all the time. Be

careful." Uh, I'm going to go ahead and

I'm going to uh remix that thing again.

I'm going to duplicate our TLDDR cuz why

not? And I'm going to call this thing uh

content to uh LinkedIn post.

Okay. And I That's it. So, I'm going to

do change its name. I'm going to leave

this the same. Going to leave this the

same. I'm going to change one thing on

this block right here. And that thing is

this behavior here. By default, we've

been displaying things back to the user.

But we're going to change that here.

We're going to say save to a variable.

And so whatever comes back from this

thing when we said extract all

learnings, what comes back to us?

Meaning what do we see in that right

hand pane when I click that TLDDR

button? We see a bunch of bullet points.

Okay. And so I'm going to call this

thing bullet points. I could have called

it summary. I could have called it I can

call it anything I want. I'm going to

call it bullets. Okay, there's a bunch

of bullet points. Call it what it is.

And then so now we've created a new

variable. So we called the model. We

gave it all the page content. It came

back to us with a bunch of bullet

points. We created a new variable called

bullets and we put those bullet points

in it. Woohoo. We got now a variable

called page content. We got a variable

called bullets. Got all these other

variables from the past. Now I'm going

to click plus.

And I'm going to call a model again. And

I'm going to ask this model to write the

LinkedIn post.

I'm also going to add some other things.

Use emojis. By the way, I don't have to

add these things, but I know you want

LinkedIn posts. LinkedIn loves emojis.

And then uh and that's it for now.

That's all I'm going to do. And then I'm

going to do this. I'm going to give it

those bullets.

Except I know that it's best practice

and again this has nothing to do with

mind studio. This has to do with prompt

engineering with prompting models that

when you give a bunch of content to a

model and then also a command it's good

to sort of wrap that content or

delineate it in some way. And I'm just

going to use the same thing and do this.

Okay. So again this might look like code

to you but this is not code. These are

just labels. These are just words. This

is a common format where to use these

angled brackets kind of looks like HTML.

This also looks like HTML. This means

end. So whatever is between this and

that the model is going to know that is

something that's some content we're

calling bullets. And here we say write a

LinkedIn post. The better way to more

precise way would be write a LinkedIn

post about these and then here

uh bullets and then this will be

ultimately clear to the model what I'm

trying to get it to do. I'm giving it

these things. I'm calling it bullets

that's actually going to get expanded

into like all of those bullet points to

this next model. And I'm going to ask it

to write a LinkedIn post about these

bullets. also asked it to use emojis

and then I can continue and I can click

plus. So far we've just used this

generate text block but there are many

other blocks. In fact there's 123 blocks

in this tab. Another 665 blocks in this

tab. One of the blocks I know in here to

be a create LinkedIn post block.

And here it says what account do you

want to use? If you haven't used any

account, you got to log into some

account oath it's called into it. And

then you have access to your accounts.

I've got three I can use. I'm just going

to use this one. And then I can say it

says, "What do you want me to post to

this Dmitri Shapiro account?" And here

I'm going to post what? Well, the post,

right? We just created a thing called

post. Or did we? No, we didn't. We

forgot. Sorry, jumping ahead of myself.

So, we called the model. Sarah, let's

step back. We trigger this with a

browser extension on any web page. We

get the page content. We send the page

content and the command extract

learnings to a model. Whatever comes

back, a bunch of bullet points. We save

it in a variable called bullets. We then

call a model again. We give it those

bullets we just got from this and say

write a LinkedIn post about this. Use

emojis. Here we want to save this to a

variable. And this variable I'm going to

call post because that's what this is

coming back with a LinkedIn post. So I'm

going to call it post. Now I can use

that variable in this field here when it

says what do you want me to post to

LinkedIn?

Well, whatever is in that variable post.

Okay. There's one other thing I want to

do because again I'm old enough to know

that I shouldn't post things publicly

that I haven't read cuz things can go

very wrong. And so here I'm going to

click plus after I create the post but

before we post it I'm going to click

plus and I'm going I'm going to use a

different type of block. This one's

called display content. Now some people

get confused between this generate text

block

because it's got this like big text area

and this display content block because

it's got this big text area. They are

nothing alike. This generate text block

calls a model and gets a reply back from

the model. This display content block

simply displays things back to the user.

It doesn't touch any model. It it's just

a display. It just displays things back

to the user. In this case, we're going

to display that post that we just

created in here in here. See that post?

We're just going to go ahead and display

it. And then we're going to pause

execution.

And then we're going to review it. And

if we say yes, then we're going to post

it to LinkedIn. Okay? And so that's it.

Like not not hard. And again, what I

want to show you here is like you can um

continue to add more blocks and use

variables to be able to do things in

those blocks. I can say publish and

again now I can go to this thing and I

can find this button

much faster. I can say run. By the way,

if I wanted to pin it, I could click

here and see I could pin it. And there's

like other things here obviously. And

then I can either sit here and watch it

run or we like to go to this debugger

over here and we can click on this and

we can watch it run in real time

and we can see what's happening, right?

Like it's it's it's instantiated itself.

It started. There it is. It loaded up a

bunch of those early variables. There's

the metadata variable. It also has all

the page content and all that. It's

called a model. Okay. And and it says

extract all learnings. And then it went

off and gave us all those bullet points.

Uh that cost us nine uh ten of a penny.

Uh then we uh uh called another model.

Okay. And now we're paused

on uh this display content block. And if

we look here, there is our LinkedIn

post. There's like our

um hashtags, there's some emojis, etc.

And if I say next, it's going to post

that to LinkedIn. Okay, I'm not going to

do it because I don't want to go delete

it, but you guys get it. This is a

really easy way to be able to chain all

of these things together, pipe them

together, and then make all kinds of

amazing agents. All of these I've shown

you are super simplistic,

but absurdly powerful. Again, even that

like one block pattern that I showed you

with the TLDDR, financial analyst,

there's countless superpowers you can

build right after this class for

yourself, for your friends. These make

great gifts. We've seen people doing

this knowing that their friend does

things repetitively and say, "Look, I

made you a gift. Here's an agent that

like will free you for, you know,

congratulations."

Uh, okay. Uh, any questions so far

before I build more stuff?

I think you're good to build more stuff.

Uh maybe one thing uh it it's not really

part of the builds you normally do, but

if you could just go over the logic

block and like human human in the loop

um interactions. I think there are

there's a question related related to

that and kind of conditional logic.

>> Perfect. Let's have you do it

and and uh I I can uh uh relax for a bit

and do is that cool?

>> Yeah. Yeah, that's cool. Awesome.

>> You guys are in for a treat. Uh I don't

normally do these these things, but um

>> What are you talking You don't normally

do.

>> I don't normally do the I don't normally

do the the the twohour ones. Um but

yeah,

>> you're not doing just doing the Just

Show them some of the

>> I got I got you. I got you.

>> I've never seen you to be shy. You guys

see like What is this?

>> Now Not now that Louis is married, he's

gotten shy.

Okay, can you all see my screen?

Okay, so uh I saw a question about um

you know pause commands and if then and

basically all this revolves around human

in the what we call human in the loop

interactions

and

you guys can all see my um my screen.

Yeah, thumbs up. Yeah. Okay, great. and

you see like uh okay great. So all of

these uh human in the loop interactions

are just different blocks that you add

to your workflows. So it's really easy

to uh incorporate some of these things.

We saw uh Dimmitri showcase um the user

input block. Did he showcase the user

input block? Right. Where you can create

these Google forms.

>> Not yet. That's what I was going to

build next. But again, feel free and

build these things like run with it.

>> Okay.

>> Yeah, I'm I'm handing this off to you.

I'm I'm losing my voice anyway.

>> You can always hand it back to me, but

>> Sure thing. Yeah. So, um you can use

these user input blocks to uh

incorporate these Google Forms style

form fields, right? And you simply add

the block. And then there's, as you've

seen, there's this configuration

interface on the right hand side. And

you can add new types of form fields by

clicking on the plus button and then

clicking on uh create new. And this is

going to show you uh a new input inside

of this folder called user inputs. Um,

now we learned about variables uh just a

moment ago where Dimmitri was showcasing

that page content variable. And in this

case, you can actually name your own

variable for what whatever you'd like to

name it. And the value from what that

what that person enters is saved uh into

uh this variable. So you can use it

later in your workflows. So uh for

example uh I can say something like what

uh kind of content

do you want to make right and there's

different uh user input types and you

can see a preview of it on the right

hand side. So this one's a long text

there's a bunch there's short text

there's uh the ability to upload files

and images and audio. Um, and what we're

going to be using is this uh input

called text choice. It's like a multiple

choice question. Um, oops, I uh

actually added the label in the wrong

place. And we're going to name the

variable content

uh type.

And you can add different options to

these inputs. So, I'm going to say

something like blog post or

uh social or maybe we can say LinkedIn

post. Maybe we add one more that is a

tweet storm.

Right? So now we have three options and

this is uh a point in the workflow where

uh the user or where we're going to

pause in the workflow and wait for a

user's response. So, if we set this AI

agent up in the same way that uh

Dimmitri set his AI agent up by clicking

on the start block and enabling it in

the browser extension, we now have

access to that same page content

variable. And then we can use that same

generate text block

in order to uh create the the content

that we want to make. Uh so I can say uh

uh we'll include page content at the

top.

There we go. And then we can say

based on the content above

create the following.

And then we can include the content type

underneath.

We maybe we want to say something like

the following new

type of content

reply only

with the new content

and nothing else.

Okay. So, we're taking in that same page

content. This is the content that's on

any web page. And then we're saying,

hey, use that to uh generate uh this new

thing that we're going to share

potentially on, you know, LinkedIn, just

like how Dimmitri had it set up, right?

So, let's go ahead and for now, we can

use this display content block or

actually we don't need that. We're going

to display this to the user and we're

just going to publish this and see what

happens.

The user input. There we go.

And now uh let's open up this agent. I

need to give it a name really quick.

Maybe I'll call this um multi-

content

generator.

And when we republish, we can go to any

website. So let's say uh the Verge,

there's this thing called Warner

Brothers mergers. We can open this up.

And then inside of my um Chrome

extension side panel,

this might have to do with um screen

sharing

on Zoom actually.

Okay.

Let's try this maybe on a different

site. Let's go to YouTube

and we'll go to uh podcast

and we can uh

here we go. We've got I don't know

something something here. Doesn't really

matter what it is.

We're going to use the Chrome extension.

There we go. And then we have this uh

new agent that we've just published.

Let me make sure I'm in

Oh, I think I'm I was testing stuff and

signed out of the

Chrome extension. There we go.

There we go. Okay, let's run this uh one

more time.

So now we have this multicontent

generator. I can click on run

and as it's running it's going to take

that page content and then we should be

presented with an option here. So you

see now we have this human in the loop

interaction. There are many kinds that

we can add. Um, and we can say I want to

create a tweet storm about the content

in this video. So, let me go ahead and

click on tweetstorm. And then you can

see here that uh based on the content,

we get each of these uh tweets. We could

obviously format this uh much nicer, but

this is awesome. We can now begin to use

uh this piece of content. So, that's one

human in the loop interaction that's

really easy. One of the most basic

types. The other type

um is well this is not human in the loop

but we are talking about uh if then

statements and sort of creating

conditional logic right and we actually

have a block for that

where we can uh if we click on view all

blocks and we scroll down we can find

this called the logic block and this

logic block is going to

uh It's going to route create two

different branching routes in your

workflow. So you can give it the con any

sort of uh context and you you create

these conditions and if either of these

whichever conditions met it will route

in that direction. So the uh a good

example of this I'm going to just uh

make another workflow here just uh to

show you how it works. We're going to

create a user input and this user input

is going to have a variable called

comment

and we can say what did you think

right so whatever they type in is going

to be saved as the comment and in our

workflow we can use a logic block

and we can uh include the

um the context and say the following is

a comment from a user

and we can showcase the comment, right?

And then we can put conditions it and in

this case in our example we're going to

say the comment is positive

and in the other one we're going to say

the comment is negative.

And then based off of these conditions,

we can add other blocks and route this

to different blocks. So for example,

if I uh have these two uh display

content blocks in this case or maybe we

have two generate text blocks.

One is going to show the comment

uh and say

reply to this comment

uh

thanking

the user, right?

And

we can give it the comment, right? So if

this is positive, we want it to uh say

thank you. And if it's

negative, let's say we want it to

uh we want it to ask, you know, why or

or what what the issue was. Um

so we'll do the same thing. We'll have

the comment and we'll say

based on

the comment above

uh infer

what

upset the user.

Okay. And so let's take a look at what a

workflow like this would look like. So

we're taking in a comment. We're using

AI to uh determine uh using kind of like

this conditional logic to determine

whether or not that comment was positive

or negative. And then we're doing

different things based on uh where the

AI routes this workflow.

So let's double check logic block must

not be empty. Oh, that doesn't matter.

That's in our other workflow.

Okay. And we will uh go ahead and and

open the draft.

Oops. We need to make this the entry

workflow. You can rightclick and find

all sorts of menus. Let's make this our

entry workflow

and open the draft agent.

And we can let's start by giving it a um

a positive comment. Wow, that was such a

great example.

Good job.

And its response was uh a thank you.

Right

now, let's go ahead and try this one

more time.

And we can say your example sucks.

You should

uh never do that again.

And in this case,

it uh tried to infer why the user was

upset. Right? So you have this sort of

conditional logic that you're able to

add in uh these various uh AI agents

that you end up building. And it's super

easy, right? It's just a couple of

blocks. And if you can determine the

path that you want your AI agents to

take, becomes really easy to create

these um these different workflows that

have uh all sorts of of different

outcomes based on, you know, things that

you view online, things that you end up

generating, you know, any sort of

information that you input into these

things.

So, I hope that answered the question.

Um Dimmitri, are you still there? Maybe

maybe we can uh open up the floor for

some questions.

>> I am if there's anything else you'd like

to cover uh regarding this uh this

feature.

>> I think questions sounds great.

>> Okay, great.

>> Yeah.

Any questions?

>> Hi. Uh this is Ellen Black. Um, can

do you have a like thumbs up thumbs down

capability available

uh so that you can record user feedback

on your agents?

Uh, you could certainly build one with

mine studio, meaning we have a a

multiplechoice user input that would

allow you to ask the user for a thumbs

up, thumbs down on things. So it just

depends on what it is you you are

building. Yeah.

>> And a rating input.

>> Yeah. Or you know again real thumbs up

thumbs down with vibe coded stuff. So

short answer is many yeses.

>> Okay. So if I created an agent uh or

assistant

>> I could build some way into that

with mine studio to capture feedback on

the response.

>> Yes. Now, your question is a bit

ambiguous, but I I'm certain whatever

you're trying to do, which is if you're

interacting with if somebody's

interacting with your agent is what I'm

hearing,

>> can you capture the response and get

their feedback? The answer is yes.

>> Perfect. Thank you. Even even better

than that um if you want to you know

edit that response in you know real time

and check in and say hey is this

something that is okay for example is

okay to post on LinkedIn. We have a

block called a checkpoint block which

will actually check in with the user

present the generated content that you

specify and you can chat with the agent

to make updates to that piece of

generated content.

>> Okay. Excellent. Thank you.

>> Yeah, generally you will find that uh

there are multiple ways if not many ways

to do things with M studio kind of the

way you want to work.

>> Great. One more question. Um

where where where do you stand in terms

of um your platform security? Like is it

sock 2 compliant now or uh

Excellent.

>> Yeah, we you can see on our homepage a

thing we call Trust Center. Uh we are

SOCK 2 type 2 compliant and audited.

We're GDPR compliant. We're finishing up

ISO 271

compliance. So

>> excellent. Thank you.

>> Uh hi, this is Jason Goldberg. I asked

the question in the chat, but I don't

know if anyone saw it. Do you is are you

can you build

>> can you build like non-web based agents

for or not I'm sure you can but how

would you do something if you wanted to

let's say you know I think the examples

I said is organize files on your

computer or like rele label them for you

or categorize them for you or

>> I don't know just things that aren't

web- based I guess.

>> Mhm. Well, so again, there's like web-

based uh uh

so again, let's take your example of on

your computer. Okay, it doesn't matter

if the thing is web- based or not. It

just needs access to your computer and

then it could be on your computer and if

it has access through some kind of a

proxy, then you could leverage that data

on the computer, use the web-based thing

to be able to analyze it and be able to

drive things on your computer. There's

nothing at this moment that we have that

installs on your computer that gives the

agents access to your file system.

And so this would need to be a

third-party thing that you know you you

would need to find. But if you did, then

it's really easy to get the agent to

connect to that thing. you know, that

thing more than likely will expose an

API or will export things out of your

file system and then it itself will call

an API that the AI agent can expose. So,

there's ways to integrate this running

in the cloud, but integrate it with

things that are sitting uh not in the

cloud, sitting inside of your perimeter

or on your computer or whatever

somewhere else.

Does that make sense?

>> Okay. So, you're saying at the moment

you can't do it locally?

>> Not Not with just Mine Studio. That's

right.

>> Got it.

>> Uh-huh.

>> Okay. Thank you.

>> Yeah. By the way, I I don't know.

Somebody else has asked me about this

before. I should probably do some

research. Offhand. I don't know like

what thing is available to run on your

computer to expose it. There's like a

bunch of obviously potential security

concerns and things like that to to

doing that. But uh I I suspect that I

mean there's nothing technically that's

preventing uh one that from existing and

two you from potentially even just vibe

coding that thing and uh and and then

it's easy to integrate it to mine studio

like that part's easy. Uh, Yakov.

>> Yeah. I wanted to know,

so you showed us how you there's the

prompt generator to kind of help build,

you know, very much more detailed,

elegant, and thought out prompts. Does

that would that work, you know, kind of

to go step by step through these

multi-step

um kind of chains that you showed us and

properly format the syntax where it's

pulling the, you know, kind of the

bullets from one and the the post from

the next and to generate the final

process or you still need to learn the

syntax of how to send you know collect

and send information from agent to agent

in those flows.

Yeah, we uh uh encourage people uh

strongly encourage people to learn to

build agents manually and so that you

understand how all of it works. But once

you've done that and again you can do

that and you know I would say half a day

to a day of real focus or if you come to

our boot camp you know in two days we'll

take you deep deep deep into it. So like

it's very learnable. You can learn to do

all of this stuff

later today by going to this learn

section, watching some of these

tutorials or going to documentation and

following some of these basically guides

of doing this stuff. Really easy to do.

You know, thousands of people have

learned to do it on their own without

any boot camps or whatever. Uh but once

you've learned to build agents on your

own, you don't actually necessarily have

to build them manually. You guys are

seeing my screen. Yeah.

I can say build new agent. And right

here

on this main automation screen of a

workflow, there's a button called

generate agent. You remember we had a

button here

when we did full screen generate. This

was generate the prompt for this

model call. But here what we're going to

do is we're actually going to generate

the agent, the entire agent. And so I

clicked that. It opened up a new tab.

And as I mentioned earlier,

this is just the Mind Studio agent was

built with Mind Studio and now it's

integrated back into Mind Studio is this

button opens up in a new tab. It says,

"What do you want your agent to do

and you can describe to it what it is

that you want it to do and push this

button and it might ask you some

questions and then it'll build it for

I'll just use this like recommended

thing. Track changes to a website and

email me updates, right?" And so it's

going to evaluate this which again is

actually quite a um ambiguous right uh

thing. And so it wants to ask us some

questions. Great use case. Let me ask

you a few questions. What specific

websites do you want to track? TechMe.

We can always change it later.com.

All right. How often would you like to

check? Uh daily at 9:00 a.m. Pacific.

Are you interested in tracking the

entire website? Specific sections or

element entire page. This is going to be

tracking a web page. Uh what kind of

changes are most important to me? Any

changes? Uh do you want simple

notification? Something's changed. Would

you prefer see exactly what content was

added? Exactly. Okay. And then I can say

cool. It says okay, I got you. Thanks,

lazy human. And then it's going to go

off and it's going to do a bunch of

work. It's going to create a plan

and then it's going to by the way as I

mentioned this thing itself is just an

agent and as you can see it's like it's

running itself and it's creating a spec

here a draft spec then it's going to

take us to this checkpoint block which

Luis mentioned that's going to allow us

to modify that spec in some way if we

want to and then we're going to say

that's great see this is the checkpoint

block it says here's what I'm going to

build I'm going to call it website

tracker daily monitoring ing checks

check me 9:00 a.m. Pacific.

Cool. This all looks good to me. If I

want to change anything, I can talk back

and forth with it. Say change this, move

that, whatever. I can just say build.

And then this thing is going to go off.

And again, it knows about mine studio

blocks. It knows about how mine studio

architecture should work in in agents.

And it will build for us the agent. Now

sometimes the agent is exactly what you

wanted and you can just run it and

congratulations. Sometime the agent is

not exactly what you want and that's why

we say if you know how to build then you

can just modify it quickly. If you don't

know how to build you might get stuck

and then get you know etc. But but

that's it. So this is yeah and again

sometimes this takes even a couple of

minutes to run but still faster than you

would do it you know and so so there's a

bunch of these types of things that are

inside of mind studio that are helpers.

And these are today we've learned you

know

5% of that 2% of mind studio there much

more to learn but it's all kind of like

that meaning it's not it's not harder

than that

and certainly if you come to like a boot

camp says great I've built you your

agent congrats here's the agent there it

is you know and and that's it and yeah I

can then like modify it and do whatever

okay so uh a bunch of things to explore

again I highly encourage you to not stop

to go to this learn section, spend some

time on these. These are awesome. This

is Louise mostly that's done these.

These are great. And you can just follow

them in order or not in order. It's up

to you, but an order works. And you will

learn a bunch of things. Give yourself a

few hours. You're going to be feeling

like, oh, I can build a bunch of agents.

You're going to be building agents.

You're going to be quite proficient in a

few hours. If you give yourself 10 hours

over time, you're going to be much more

proficient. You want to come to a boot

camp, come to a boot camp. We're going

to take you through all the paces.

You're going to be ultimately proficient

and certified level one. There's also a

30-day boot camp that you can attend.

You can get level three certification. A

bunch of people here have been through

that. Just learn much more. By the way,

there's a number of universities that

are uh going to start offering the Mind

Studio 30-day boot camp as their

continuing education curriculum.

Yeah. The first one uh up is San Diego

State University and then looks like

UCSD and University of Southern

Mississippi, you know, other folks. So,

like you can take it at your alma mater

perhaps as well if you want a

certificate from them.

Cool. Okay.

Uh all right guys well look again most

folks here already have heard all the

answers so it's a small number typically

we have many more so there's many more

questions so if we don't have them we

can give folks you know 35 minutes back

otherwise if you got questions happy to

answer them

>> hey Dmitri this is Solomon here I just

had a quick question for you

>> um right now Um, I'm working at at an

internship for a company and I want to

link my SharePoint into Mind Studio

using the API. I'm not sure too much how

that works. I've given them a link or

something like that and I've turned on

certain permissions from uh Microsoft's

developer developer page because I think

I need to do that to get it running, but

I haven't been able to actually pull any

information yet. I'm just wondering, is

there a common bug I'm running into?

>> Uh, not that I know of. probably

misisconfiguration. The best way to get

these things solved again for everyone,

join our Slack community,

>> okay?

>> All of these folks that you see here

that are certified are active there.

>> You can ask questions there. They might

be able to help you. Our team is there.

We might be able to help you. We can

help you figure out where you're, you

know, it's misconfiguration, I'm sure.

Yeah.

>> All right. Thank you. I appreciate that.

>> Yeah, it's a really friendly community,

so take advantage of it. For sure. Yeah.

And is there a link? I see someone asked

in the in the chat there.

>> Uh I'm sure there is a link. Let's see.

Can Did anybody find it yet? Or I can do

it too?

Here we go. Invite people. Copy. Invite

link.

There you go.

Thank you. Mhm.

All right, folks. Yeah, good to see you.

Uh, thanks for coming. We'll see you in

Slack. Yeah, and then if anybody shows

up out of this group to our next boot

camp, then we'll really see you, a lot

of you. All right, guys. Be well. Take

care.

>> Thank you very much, Me.

>> My pleasure.

>> Thanks. Byebye.