Latest / AI for Founders with Ryan Estes / Vibe Coders Rejoice

Transcript

Vibe Coders Rejoice · AI for Founders with Ryan Estes

0:00
Welcome to AI for Founders. I'm Ryan Estes.
0:02
Today's episode starts with the simple but brutal truth vibe.
0:05
Coding breaks the moment your idea gets bigger than make the
0:08
button blue. Maybe that's a bit far, but
0:11
every founder knows this moment when you've stretched lovable
0:14
cursor or Claude code as far as your vibes can take you.
0:17
And suddenly the whole thing collapses under the unclear
0:19
prompts, messy code, and your total inability to explain what
0:23
you want in the first place. Well, today's guest built that
0:26
fix. I'm talking with Tyler Wells, Co
0:29
founder of Brain Grid dot AI, the platform powering over 2000
0:34
AI builders and 10,000 plus ship features by turning messy human
0:38
ideas into precise AI ready specs, tasks and prompts.
0:43
If you've ever wondered how non-technical founders are
0:46
suddenly shipping multi tenant SAS apps, or how senior
0:49
engineers are spinning up six agents at once without imploding
0:52
A repo, this is your episode. We get into how Brain Grid
0:56
builds the missing planning layer of AI development, why
0:59
your prompts suck and how to fix them, how founders with 0
1:03
engineering background are shipping real products in the
1:05
future of anyone with an idea can actually build it software.
1:09
This episode is for vibe coders, indie hackers, AI native SAS
1:13
founders, and anyone who has ever thought, wait, can I
1:16
seriously build this myself? Spoiler yes, and Tyler shows you
1:20
how. And if you like learning AI
1:22
through founders who are actually building, not just
1:24
posting screenshots, go sign up for the newsletter at AI for
1:28
founders.co. My newsletter is where real
1:31
experiments, failures, frameworks, and AI tools get
1:34
shared first. If this episode fires you up,
1:37
the newsletter will absolutely ruin your free time in the best
1:40
possible way. And if this episode made your
1:42
dopamine spike at least once, leave a review.
1:45
It's the podcast equivalent of giving me +5 energy and a rare
1:49
loot drop. This episode is brought to you
1:52
by Kitcaster, the podcast booking agency trusted by over
1:55
800 founders and now part of MO Burst, a global marketing
1:58
powerhouse. Here's the truth.
2:00
Podcasts are the highest leverage channel for founders in
2:03
2026. Your customers want to hear your
2:06
voice, your investors want to know your story.
2:08
And the market wants authenticity, not another
2:11
LinkedIn carousel Kitcaster books found on the perfect
2:14
podcast so they can tell the right story to the right
2:18
audience at the right moment. No more cold pitching.
2:21
No more dead end in boxes. No more who will I talk to?
2:25
Just high quality interviews that build trust, authority, and
2:28
inbound demand. If you're a founder trying to
2:30
grow, raise, or simply stop being invisible online, go to
2:34
kickcaster.com. It's time to get booked, get
2:38
heard, and get your message everywhere.
2:40
That's Kick caster by MO Burst. Tyler, welcome to the show.
2:58
Thank you so much for joining me.
3:00
Right. Appreciate you having me on.
3:02
Absolutely. It's my pleasure and I can't
3:03
wait to talk about your company Brain Grid AI, which is we're
3:08
going to have a deep conversation about vibe coding.
3:11
Before we get into it, give us a quick introduction of the LLMS
3:14
you're using and kind of what capacity right now.
3:17
I mean, we use typically I would say the, the, the two main
3:21
foundational models we're using right now are Anthropic and
3:25
Gemini, those ones that have provided the, I would say the
3:29
best output as well as I would say the best bang for the buck.
3:35
We obviously just started switching over to Gemini 3 for a
3:38
number of our agents and I've been very happy with what we're
3:41
seeing there. But I would say pound for pound
3:45
that had dropped with the, the sonnet 4-5 is, is, is still the
3:48
winner right now. But you know, it remains to be
3:51
seen if, if Gemini 3 is going to going to edge it out.
3:54
Yeah. And then what kind of
3:56
applications are you using it for?
3:57
Is this for generative like writing code?
4:01
Well, how are you using them? So I mean, at the heart of what
4:04
Brainbrit does, you know, it's a series of of agents that we're
4:10
using to predominantly the, the 4-5 model for those agents in,
4:14
in Brainbrit that you're seeing. And they are writing very
4:17
detailed product specifications. They're writing epics.
4:21
They're taking those requirements and your the
4:24
analysis of your code and the understanding your code.
4:27
And they're, they're breaking that down into actionable tasks
4:29
or prompts that can be fed into things like Claude code.
4:33
And all of that is is driven by those foundational models where
4:38
we use Gemini for is for all the code analysis.
4:41
So part of what Bregburn also does in order to provide rich
4:45
context and sort of ground the agent is we connect to your
4:49
GitHub repository and we take that code base, feed that
4:53
through through Gemini and allow it to produce 5 different types
4:57
of documents, like architecture, you know, key, key workflows,
5:03
sort of data structures, what the directory structure looks
5:05
like. Because obviously the
5:07
foundational models know nothing about your code unless it's open
5:09
source and they've happened to scan it.
5:11
But the vast majority of code is probably not, and your private
5:14
repos are not going to be seen by them.
5:16
So you've got to provide context to those models in terms of what
5:20
it is your code actually does. Percent and to to kind of tie a
5:25
bow on that, from what I understand from brain grid and
5:28
correct me if I'm wrong is it's kind of a graduation process of
5:31
people maybe that got into some vibe coding, had some success
5:35
with lovable, had some success with Bolt replit and they're
5:38
ready to kind of graduate to the next levelled Claude code and
5:42
tools like that. And why that kind of is helpful
5:47
for me is I'm that guy. You know what I mean?
5:49
I'm that guy. I'm unlovable, making awesome
5:52
stuff. But my prompts look like, yeah,
5:54
but make the button bigger. You know, I've dabbled with
5:58
cursor, I've dabbled with Claude code, but really is kind of
6:01
intimidating. And all of this is kind of
6:03
rooted around the quality and integrity of those prompts.
6:06
So really what brain grid is here to do is to to get a level
6:11
up on be basically using prompting as the engineering and
6:16
coding language. Is that correct?
6:18
Absolutely. I mean, I think that's a great
6:19
way to think about it is, you know, we want to, we want to
6:21
ensure that you're prompting like a pro.
6:24
We want to give those vibe coders the planning, the
6:26
structure, the building blocks that they need to build, you
6:29
know, real software with the, the tools that you just
6:31
mentioned. You know, how do you take
6:33
somebody that does not have an engineering background and get
6:38
them to explain in the proper terminology and and nomenclature
6:42
for what it is that is in their head that they've been
6:46
ruminating their head for years that they've never been able to
6:49
build because, you know, they couldn't afford software
6:51
engineers or they have a day job and that's not something they'd
6:53
be able to do. But then give them those super
6:55
powers. And in order to get those agents
6:58
to build what it is that they want to build, you've got to
7:02
write the prompts that the prompts are are key in order for
7:05
this to get that idea out of your head.
7:08
And what we do is we allow them to take that messy idea that's
7:11
in their head. Tell brain grid, Hey, here's
7:13
what I want to go build. And then we do a lot of that
7:16
deep thinking, OK, and, and through questions and answers
7:19
and, and either looking at your code or, or starting from from
7:22
scratch, we're going to entice what it is you're trying to
7:25
build. Think about the edge cases,
7:27
think about the best technology and give you those sets of
7:30
prompts that you can then that are fed into the agents for them
7:33
to go and implement. Amazing, I'm going to be real
7:36
selfish here and and create a case study based on my problem.
7:40
Let's do it. OK, so I've got a database of
7:43
contacts and the contacts have different attributes in a
7:46
spreadsheet. And what I want is be able the
7:49
user to walk through a survey and then the responses to the
7:52
survey will actually filter that database.
7:56
And I've got it to a position in lovable where it it fakes it,
8:00
you know? And I really can't.
8:03
I can't get it to actually use the attributes from the from the
8:06
survey to accurately filter and give me the the data on the
8:10
backside. So walking into brain grid, how
8:12
would I begin to use the tool to maybe solve my?
8:15
Problem first and foremost, the the thing that's going to help
8:19
Brain Grid be the best or perform the best is we need
8:23
access to that source code. Unfortunately, if that's just
8:26
sitting at Lovable, we're not going to be able to see that.
8:30
So hopefully you've connect that up to your get up account and
8:34
you've synced that code that was written in Lovable through
8:37
Lovables agents and you've gotten that into into GitHub.
8:40
If you've got that into GitHub, then you can install the brain
8:43
grid get up application. So that way we can generate that
8:47
rich set set of documents and contacts that I talked about
8:49
earlier. And then once you've done that,
8:52
just express your intent. So when you want to build
8:55
something, you express that intent at natural language in
8:59
English. And, and we're going to help
9:02
tease out that idea of, of what it is that you're trying to
9:05
build. And we're going to give you the
9:06
best technical approach that we can so that you can get to the
9:09
the task and actually get that thing building.
9:14
This episode is brought to you by MO Burst, the AI first growth
9:18
engine trusted by iconic startups like Calm, Robin Hood,
9:22
Reddit, Uber and Google. Moburst is a digital, full
9:26
service, mobile first marketing agency that helps companies to
9:29
scale and become category leaders.
9:32
Whether you're finding product market fit or ready to
9:35
accelerate growth, Moburst delivers creative, data-driven
9:39
solutions that help with scale and efficiency and impact.
9:43
At Moburst, we're not just guessing what works.
9:45
We're predicting it. As an AI first agency, Moburst
9:49
uses cutting edge tools to analyze user behavior, adapt
9:52
creative in real time, and cover hidden growth opportunities.
9:56
The result? Marketing that's faster,
9:59
smarter, and built to convert from early stage startups to
10:02
global brands. We've helped hundreds of
10:04
companies overcome complex challenges, unlock new
10:08
audiences, and optimize every metric that matters.
10:11
At Moberst, we believe that creative thinking, advanced
10:14
technology, and data are the core drivers of success at
10:18
scale. If you're serious about scaling
10:20
your business with precision and purpose, it's time to talk to
10:23
Moberst. Visit moberst.com and let's
10:26
build something category defining together.
10:30
OK, I've got a confession. As a podcaster and founder, I'm
10:33
constantly meeting incredible people, guests, listeners, or
10:36
other entrepreneurs. But here's my little secret.
10:39
I'm terrible at following up. My inbox is basically a
10:42
graveyard of missed opportunities.
10:44
Does that sound like you? You meet someone amazing,
10:47
exchange emails, have great intentions, and then life
10:50
happens. Six months later, you're kicking
10:52
yourself thinking I should have stayed in touch with that
10:54
person. This is exactly why I started
10:56
using Warm Star dot AI and honestly, it's transformed how I
11:00
think about my network. Instead of letting relationships
11:03
go cold, Warm Start analyzes my existing contacts and helps me
11:07
identify who I should be reaching out to and why.
11:10
It even writes the outreach messages for me.
11:12
But here's the key. I review everything before it
11:15
goes out. I'm not just talking about this
11:17
because they're a sponsor. I'm talking about it because
11:20
watching a healthcare executive use this to build his entire
11:23
Advisory Board in one weekend with four CEO signups made me
11:26
realize I was leaving money on the table every single day.
11:30
Your network is your net worth, but only if you actually nurture
11:34
it. Check out Warm start. dot AI.
11:36
Use code AI IFF to get 50 bucks off your first month.
11:41
Stop letting your relationships turn into miss revenue.
11:45
Amazing. So I'm I'm connected to GET with
11:49
my lovable account. I'm connected to brain grid
11:52
through git and now then once I'm actually operating it, am I
11:57
still prompting in lovable or am I now prompting in brain grid
12:00
and it's updating the lovable code source?
12:02
How does that work? On Brain Grid, we don't actually
12:04
write the code. So we're, we are not a coding
12:07
agent. We do everything before the
12:09
coding agent is actually going to start doing the
12:11
implementation. So once you've gone to Brain
12:14
Grid to create that account, you've connected your, your get
12:17
up through installing that application.
12:19
You will be, you know, in our, in our web app and you're
12:23
literally, like I said, you're, you're, you're typing out that
12:25
intent and then you're answering the questions, you're iterating
12:29
back and forth with our agent. And then you're winding up at a,
12:32
at a, at a requirement or you're winding up at an epic, depending
12:35
on the, the size, what it is you're trying to do.
12:39
It's like if you go back and I'll get to the rest of it, but
12:41
if you go back to sort of like the traditional SDLC or software
12:44
development life cycle, right? You've got at the top of that is
12:47
like if it's a big project, you're going to, you're going to
12:49
write an epic. And the epic sort of defines
12:51
like some very large body of work that you're trying to do.
12:55
And you're going to break that down into requirements.
12:57
And then each requirement is going to be a feature that sort
13:00
of satisfies the that epic. And you, you kind of go through
13:02
all that until you're done. Then you end up with a, you
13:04
know, a full product or, or a very large feature that can't be
13:08
done in a single, single shot. And so when you've got to the
13:11
point of you express your intent in, in, in English or natural
13:16
language, you've got that requirement, you've got those
13:18
tasks. The best way to do it is then to
13:21
hop over to something like cursor Claude code or Windsor,
13:24
whichever 1 is your favorite. And then of course, my
13:26
suggestion would be to either use the brain Grid MCP or the
13:30
brain Grid CLI and then literally give it the command of
13:33
like, do you tell the model or you tell Claude, you say, hey,
13:36
build this requirement and it's going to, it's going to go out
13:39
to brain Bread via the API. It's going to pull down the
13:42
requirement. It's going to pull down all of
13:44
those tasks. It's going to adjust all of that
13:46
as context. And then it's going to say, OK,
13:48
you've got 5 tasks that have to be implemented.
13:51
Great. Let's start with task one.
13:52
It's going to go through the entire implementation.
13:55
It's going to do your testing. It's going to make sure you're,
13:57
you're linting, you know, any of the sort of depending on what
14:00
language you're writing on, make sure everything that has been
14:02
written works and it's going to move on to the next task all the
14:05
way until it's done. And then when those tasks are
14:08
complete, you know your your requirement is complete and your
14:11
feature should be built and built properly.
14:13
OK, amazing. So so basically the the step
14:19
from lovable to clawed code seems significant to me when I
14:23
go into clawed code. Little bit intimidating, you
14:25
know, is there, is there a good tutorial of something that's
14:28
going to be able to allow you to get your bearings a little bit
14:30
in there? Kind of what do you recommend to
14:32
kind of soothe the and pulling my hair out, looking at Claude
14:36
Co trying to figure it out. Yeah, I mean it.
14:38
It's definitely a big leap, especially for people that don't
14:42
have a traditional software engineering background.
14:44
And and we've got a customer that was exactly that or exactly
14:48
is this she, she runs Pilates studios, right?
14:52
She is exactly like you. She wasn't lovable.
14:55
She had an idea, started building it there, got
14:57
frustrated. I don't know by the fact that
15:00
lovable couldn't write what she wanted to, what she wanted to or
15:03
couldn't create what she, what she wanted to have done.
15:06
Started looking around, found us, you know, had our first
15:10
conversations and, and realized, like, look, I, yes, I can take
15:13
these tasks or these prompts that they're creating brain grid
15:16
and feed them into lovable. But I, I want a, a, a deeper
15:20
level of control. I, I want to, you know, get a
15:23
little bit deeper into the, into the into the weeds, so to speak.
15:26
And you know, she's jumped into to cursor.
15:29
I think cursor, you know, my standpoint, I'd probably go plot
15:34
code first because it's hard to say.
15:38
I mean, maybe, maybe not because you, you kind of want that ID at
15:41
the same time to, to be able to visualize the code so you can
15:46
actually see what it's doing. But I think like from the vibe
15:50
coder standpoint, like really what you want to be able to do
15:52
is just is, is open up your project.
15:55
You know, you, you've cloned it down and get you open up your
15:57
project and the IDE, you can see the code.
16:00
And then, you know, either connected through the MCP or if
16:03
you want to do it the manual way, you start feeding at the
16:05
implementation tasks and then just start looking at what it
16:09
does. Like you really don't have, you
16:11
want to make it so you don't have to do much more than that.
16:14
And so by the time you've fed it all those tasks, you're not
16:17
really writing anything, you know, you're allowing the agent
16:21
to go write it for you. And then depending on where
16:24
you're at, And then I would, I would say sort of the, you know,
16:28
how much you've done from a testing standpoint, how much
16:30
you've done from a, a linting standpoint, You may not have
16:33
done any of that stuff, but these are all maybe foreign
16:35
concepts that you're going to have to learn some of that
16:37
stuff. You can again, turn around and
16:39
ask brain read like, how do I make this?
16:41
How do I make this safe? How do I make this testable?
16:43
How do I ensure that what's being produced is going to
16:46
actually work? But we want to get to the point
16:48
where like a lot of that stuff is just comes automatically
16:51
through the prompting that we provide to you, you know, where
16:55
the IDE maybe becomes second nature or maybe at some point
16:58
you don't even have to open up the IDE.
17:00
You know, you, you just essentially let brain grit go
17:03
and next thing you know you've got APR that's been completely
17:05
tested and, and, and it's good to go and ready to merge.
17:08
Writable, yeah, cool. So pull pull code base from
17:13
Loveable if you want to keep it there.
17:15
You want to sync it? Yeah.
17:16
So sync it. Yeah, sync it.
17:17
Sync it from Loveable to GitHub, clone that down, clone it down
17:21
from GitHub, Get that onto your machine locally.
17:23
Yeah, get that machine locally. Open up clod code or open up
17:27
cursor, you know, when you're when you're ready.
17:31
But first thing is, you know, once you've cloned that down,
17:33
once you've got that, that reap that, that code base inside of a
17:37
repository, GitHub, go to Brain Grid, create that account,
17:41
connect it, you know, install the GitHub app, create your
17:44
project and start writing your first requirement.
17:48
And then what brain Grid is going to do is fill in the the
17:50
blank spot. So my problem is I don't know
17:53
what I don't know, right? If I'm coming to brain Grid, I'm
17:56
like make Darrell strawberries rookie card.
17:58
It's going to be like cool, no problem, But then it creates a
18:01
prompt that's going to cover all the bases.
18:02
Be like, you know, look at Don Russ and tops and and upper deck
18:07
and research all the the rookie cards for Darryl strawberry.
18:10
And then it it's going to fill in kind of the gaps that I don't
18:13
know I'm missing. Is that accurate?
18:15
That's, that's very accurate. Like if you sit down and you're
18:17
like, hey, look, I've I've got this web app.
18:20
I could login with a username and password, but I want to be
18:22
able to login with a Google account.
18:25
I want to have social login support.
18:27
You could sit there and and type that into brain grid and say,
18:31
Hey, I need to be able to login. I want my users to login with
18:34
Google. How do I do this?
18:36
And it's going to come back with recommendations.
18:38
It's going to come back with asking you some questions.
18:40
It's going to say things like, you know, what do you want to
18:43
use? And, and you know, here's some,
18:45
here's some alternatives. Looking at your code base,
18:47
here's what I think is the best. You know, Are you ready for me
18:50
to write the requirement And you kind of go from there.
18:55
Blocking with Google is a much better use case.
18:58
I don't know where Daryl Strawberry's rookie card came.
19:00
From who's going with it? I mean, I'm, I'm fine.
19:03
I'm like, hey, you know what? Maybe, you know, maybe, maybe it
19:06
can, you know, maybe the models are that good now that they can
19:09
somehow conjure up a Daryl Strawberry rookie card.
19:11
I mean, that'd be great, yeah. You know, he had a, he had a
19:17
beautiful swing. I mean, maybe it's just
19:19
ingrained in my brain somehow. OK, cool, awesome.
19:23
So would you say that the majority of your your clients
19:26
are are like me, which are like, you know, intermediary to low
19:30
intermediary, you know, basically decent can knows where
19:36
to go to find answers, put it that way and is looking for to
19:40
graduate with their project or are you look, is it most of your
19:43
customers a little bit more senior?
19:45
No, it's, it's a 5050 mix right now.
19:48
We've, we've, we've definitely, we're, we're trying to open
19:53
ourselves up more to vibe builders to make it easier for
19:56
them to build and create, but. I would still say.
20:00
We've got quite a few of a very senior, you know, professional
20:05
software engineers that like the structure that understand the
20:09
necessity to get these agents to perform the best.
20:14
You need to provide them a lot of context in the form of
20:17
requirements and well crafted prompts.
20:21
You know, they have come to the same realization that that we
20:24
came to, you know, towards the end of last year that just
20:27
sitting down with an agent and expressing minimal amount of
20:31
intent for what you want to do yields, you know, at times, you
20:37
know, explosive results to to sort of say the least.
20:40
Whereas when you up that level of context and you know that has
20:45
been grounded in your code base and give it really, really good
20:48
plumps, you've now put that thing on Rails and focused it
20:52
and it can create amazing software for you.
20:54
Well, and we have a mutual friend, Matt Bernier.
20:57
Shout out to Matt, friend of the show, great guy, very senior,
21:01
very talented developer. You know, how, how is he?
21:05
How is he? And he like his team using using
21:08
the product. Yeah, that's a great story and I
21:10
appreciate that the the reintroduction when he when he
21:13
sent the e-mail and I was very happy when I saw the the last
21:16
name with Matt Bernier of hey, I know that guy.
21:18
We work together at Twilio and yeah, I would love to talk to
21:21
him. So when we got on the the, the
21:23
first call, I mean, he, he and his team are, I mean they're,
21:27
they're by coding extraordinaires and and they're,
21:30
they're deep in the ecosystem. And when he showed me what he
21:34
was running to sort of achieve the same level of what we've got
21:37
at Brain Grid, I was, I was kind of blown away.
21:40
He was, he had taken some open source piece of software that
21:44
was allowing him to, you know, create tasks and and sort of
21:48
pseudo create requirements. I think it was like some sort of
21:51
like Kanban model that he had to run himself.
21:54
He had to run it locally. So that meant no one else in the
21:56
rest of the team can see it or use it.
21:59
He just had like all of this heavy lifting of infrastructure
22:03
and, and pieces that he had to to operate himself just to get
22:07
to the point where he gets sort of manage all of the artifacts
22:10
around what he wanted to use when, when, when coding with the
22:14
agents. And then when we sat down and,
22:16
and I showed him like, Hey, that mean that's amazing.
22:18
I love what you've done, but man, that's, that's a ton of
22:20
work. Let me show you brain grid.
22:23
He dropped, they dropped it all and, and came across and, and
22:26
they've been using us now for I think they just got, they just
22:28
hit their second month. They want to say they've created
22:33
somewhere around like 50 or 60 requirements.
22:35
And I mean, they've just been churning and burning there.
22:38
And basically he sits down and, you know, opens up, you know,
22:42
loads up brain break rights requirement.
22:45
How's that broken down into tasks pops over into Claude code
22:50
uses the MCP literally says build this requirement and
22:54
you'll probably, hopefully he's using things like work trees,
22:58
which is magic because now you can spin up multiple agents at
23:01
the same time and and work on multiple requirements in
23:04
parallel within the same code base.
23:07
You wait till that's done and all your tests are passing and
23:09
everything's linting nicely and now you've just built an entire
23:13
product well. It's so cool.
23:17
Let's talk about the inception of brain grid.
23:19
When did you launch and, and how did how did you come up with
23:23
this idea as a as a outcome? Yeah, so we.
23:26
Started this in February of this year and the way we got here is
23:31
I think like like most seasoned engineers we started just like
23:37
how could we use this AI? And we all started with cursor,
23:42
myself and my Co founder, and we had another company that we'd
23:47
started back in, in 21 that was, I don't know, probably on its
23:52
way out, wasn't sort of hitting the venture scale we needed, but
23:56
we'd had to lay a bunch of people off and we had to
23:58
obviously keep building. And we're like, well, how can we
24:00
scale this now that we've scaled down the rest of the engineering
24:04
team? Well, let's try these AI tools.
24:05
And, and it came through our own sort of trial and error and
24:09
learning of how to be the most efficient and get the most out
24:13
of the coding agents. And interestingly enough, this
24:17
was right around the time that cursor started launching rules.
24:21
And we sort of belonged on the rules.
24:24
But it's like, oh, great, you know, we can actually constrain
24:27
the agent. You know, we can sort of give
24:29
it, put it on rails, give it some guidelines and let's see if
24:33
we can build a platform around that.
24:34
And so our, our actual, our first iteration of what Brainrid
24:38
was, was a, a, a way to sort of manage and share cursor rules
24:43
out there in the world. And very quickly realized that
24:47
that is not going to be anything sustainable or a business that
24:50
we want to run. You write a rule run once and
24:53
you really don't ever visit that thing again.
24:56
And you know, we kind of took a step back and, and this is
24:59
probably like the January, early February time frame.
25:02
And we're like, yeah, that this is not going to work.
25:06
You know, what is, what is the real, the, the real thing here
25:09
that we've learned, you know, through the sort of trial and
25:12
error and, and, and trying to get the most out of the agents.
25:15
And, and that's what kind of brought us to brain.
25:18
But we're like, look, you've you've got to provide
25:20
understanding the agents, you've got to provide really good
25:24
specifications and you've got to break those specifications down
25:28
into small actionable tasks that can be, you know, executed by
25:34
those agents, hopefully in a single turn.
25:37
You want those to be atomic. You want, you know, I want to do
25:39
this task or maybe some tasks I can do in parallel.
25:42
But you want to be able to interact with those agents to
25:44
allow them to help you develop the requirements and the tasks
25:49
in a manner that is going to be, you know, give you the best
25:51
opportunity to sort of single shot those and efficiently build
25:55
up that the software that you want.
25:56
And that led us to Brain Grit. So you, you, you kind of
26:01
establish what brain grit is. You say like, hey, this is
26:04
actually going to serve a significant Tam.
26:06
Did you, did you raise around to finance the project?
26:09
How'd you finance? Yeah, we did.
26:11
So we ended up raising a, a small precede with Menlo
26:16
Ventures. So Menlo Ventures out of or of
26:19
Sandhill somewhere in Northern California, they had a thing
26:23
called Menlo Labs. And so Menlo Labs was our first
26:27
investment. Just recently they did a a small
26:31
follow on and then we've just added a couple of more investors
26:34
to our our precede ticket here. Incredible how much you raised
26:38
in total. I'll be raised 700.
26:42
So man, yeah, cool. Now fairly worried about that,
26:46
like in the old days in, in 2021, you know, when we started
26:49
our, our company after meeting Twilio, I made right up the
26:52
gate, we raised like 4 million. You know, now that the ticket
26:55
sizes are much smaller because you just don't need the big
26:58
teams and you don't need the, the spin like you used to.
27:01
And you know, the fact that we're, you know, we're sitting
27:03
at 700 right now. I'm very happy about that.
27:06
That gives us a very long, you know, a nice decent runway to to
27:10
really insectualize, you know, really what we want bread to be
27:12
common and what it is today. So cool it seed strapping is
27:17
what I think they're calling it. Yeah, but it, it's, it's great
27:20
argument because a lot of people will talk about AI in general
27:23
about the massive expense, whether that's energy or like
27:27
just the, the expense of AI. But like really the the outcome
27:31
of that in the short term is like it's actually coming out
27:34
very inexpensive now. Now the February 2025 is the
27:38
infancy of high encoding kind of did you have to do a little bit
27:42
of explaining to like what you were working on or how'd that
27:45
go? Every time, I mean every, every
27:48
single time. And and I mean you've got to
27:50
explain it over and over and over.
27:52
And it's like, you know, the the the typical push back is like,
27:56
oh, Claude is just going to do that or what are the
27:57
foundational models is just going to do that.
27:59
Will that be the job they have already is hard enough and, you
28:04
know, leveling up to where the bike coders need to be and
28:07
focusing heavily on on the specifications.
28:12
You know, a lot of the folks that we talked to, especially
28:14
from the investor standpoint, have never been operators.
28:16
They've never built. They've not they've not done a
28:18
lot of coding. So from their standpoint, you
28:22
know, you kind of get a lot of blank stares at times of like,
28:24
well, I, I don't really get why this is important or why do you
28:28
want to write these requirements?
28:29
Like, can't you just, you know, you know, my e-mail, but the
28:34
agent and just goes out there and just does all this for you
28:36
magically? Like, I think so.
28:39
Not quite. I mean, you can try that and I'd
28:43
love to see what you get coming out of that.
28:45
But like here, this is a proven way to do this.
28:49
This is grounded in in, you know, software development
28:52
principles. You know, yes, it, it may seem
28:56
antiquated and, and I'm sure at some point down the road that
28:59
this will change, but you know, right now to get the best bang
29:03
for your buck and, and the, the most efficient use, I think out
29:06
of the agents like we're, we're applying the same type of
29:09
principles that we applied to human coders, like, you know,
29:13
writing those requirements. They sit there thinking about
29:15
deeply about what does this look like and what are the edge cases
29:18
look like? And of course, I've never seen a
29:20
product manager write a requirement document like our
29:23
agents can, but they're not for humans necessarily.
29:26
These are to be consumed by, you know, other, other AI agents.
29:30
But you know, we're applying that model and I I think it's
29:33
working really, really well. Yeah, well, and that's the
29:36
boring part. You know, it's like, you know,
29:38
they allman's talking about like this is the age of the idea guy.
29:42
You know, this is the time that people would have an idea.
29:45
They can act. You got a Pilates studio, you
29:47
want to create some software. You can do it now, you know, but
29:50
what you want to do is is keep your head in the clouds and be
29:53
thinking about what ifs. Actually writing the
29:55
requirements is like brutal, you know.
29:57
So having a tool that can do that, especially, you know,
29:59
there a lot of things come into play here, which is like
30:02
imposter syndrome. There's like all these, so I
30:05
can't do it. You know, you want to have, you
30:08
want to be able to use the, the, the momentum of the idea, the
30:12
intoxication of like what can be to propel you.
30:16
And a lot of times I think you get lost in the weeds.
30:18
You know, people are working on a 20 page business plan, they're
30:21
doing the requirements, all this stuff and it's just the thing
30:23
fizzles out. It seems like this allows you to
30:26
keep that spark much longer and actually go much deeper into the
30:29
product much faster. Yeah.
30:31
I mean, the, the dopamine hit that, you know, our customers
30:34
are getting from, you know, describing something that's been
30:38
in their head for possibly years and watching that materialize in
30:43
front of them on the screen, you know, it's, it's intoxicating.
30:47
And, and the feedback that begins, I mean, the Pilates, you
30:50
know, the woman that runs a Pilates studios is like, look,
30:53
I've had this idea forever and I want to build this multi tenant,
30:57
you know, Pilates studio platform that, you know,
31:01
multiple studios can come into and they can have their own sets
31:04
of videos and they can play them and their own subscribers and
31:08
their own payment systems and blah, blah, blah, blah, blah.
31:10
Like she's like, I would have had to hire a huge team to do
31:13
that. Now she's pulling it off on her
31:14
own, you know, using brain grid, you know, and it's, and it's the
31:18
same, you know, it's the same story.
31:20
We've got a mortgage broker, you know, out of Michigan that had
31:23
an idea for building essentially a, a SAS for his mortgage
31:28
business. And he's like, look, I've had
31:31
this floating around in my head and, and I, you know, I started
31:34
off with, you know, probably cursor or something like that.
31:37
And I've discovered you guys and he's like, I can just, I can
31:39
take this thing anywhere. I mean, it's a very, it's a very
31:43
brave new world when you see a, a mortgage broker creating a
31:47
requirement for semantic analysis of telephone
31:50
conversations. Like what, how is that?
31:54
How is that in his job description on his on his, you
31:57
know, for his day-to-day work, yet he feels empowered enough to
32:01
take that leap of like, hey, I've read about this or I've
32:05
learned about this thing. Now I want to go see it actually
32:07
in action. And all of a sudden you see
32:10
those requirements, you know, coming across being built inside
32:12
a brain grid. And then, you know, the task
32:15
created and him implementing it. It's, it's incredible it.
32:19
Really is, you know, it's like dopamine hits.
32:22
Brave New World Aldous Huxley When did you know that Brain
32:26
Grid was going to be kind of like a drug dealer?
32:29
I mean, when I started using it myself, but I mean, it's yeah,
32:33
it's I not only my building brain grid.
32:36
I mean, I'm using it on a daily basis.
32:39
I was, I was up this morning, I guess getting by head at, you
32:41
know, 5:00 AM, you know, there's there's, you know, some, some
32:47
probably, you know, some services that needed to be
32:48
built. And then here I am, 5:00 AM, you
32:51
know, telling it like, hey, look, we've got to do this, this
32:53
I, you know, I, I need to do secure, you know, storage of
32:57
environment variables. It needs to be on a
33:00
organizational level for each project.
33:04
You know, let's come up with a way to do this.
33:06
Write the service, write the test, you know, and it's just,
33:08
I'm sitting back there just interacting with the agent, you
33:11
know, over my morning cup of coffee and let's get this thing
33:15
built. And by 6:00 AM, the thing was
33:16
built and shipped. Incredible, incredible.
33:20
You got some impressive stats on the on the website, 2000 plus AI
33:23
builders, 10,000 plus features shipped.
33:26
Now we're going back to February.
33:28
You raise some cash, you wrap a brand around it, you go to
33:31
market. How do you go to market and how
33:33
do you get your first couple of customers?
33:35
And did you use design partners in the early stages?
33:38
Yeah. I mean, we definitely have a few
33:39
design partners that are in there early giving us feedback.
33:43
And it was typically we were using, you know, people from our
33:47
own network in the, in the very beginning, you know, getting
33:51
their feedback. And then we started opening it
33:53
up. We, we had a wait list going,
33:55
obviously. And, you know, we would, we
33:57
would sort of like pick off the marquee emails that we saw there
34:01
and like we would target people that had to clog Mac's accounts
34:03
because we're like, OK, those are serious people.
34:05
Like let's get them in. And, you know, we would bring
34:09
them into our Slack. We'd start a Slack channel with
34:11
them and, and then stay very tight with them just to get
34:13
feedback. But you know, a lot of our
34:17
customers have come in through, you know, a single ad that we
34:20
had out there. We, we, we advertised with
34:23
cursor dot directory that gets a lot of traffic and it gets the
34:26
right type of traffic for the people we're looking for.
34:29
So if you're going there, you're using cursor or you're using
34:31
some sort of one of the AI tools and you're trying to find ways
34:35
to, to make it better. And that's, that's why you're
34:37
there. And then they find us and they
34:38
come in, they they create an account, they start creating
34:40
requirements and they're sort of off to the races.
34:44
And you've got great pricing, you know, kind of entry is kind
34:47
of 10 per month build yearly. You got a pro at 23 per month,
34:51
75 per month for team. How did you land on pricing?
34:56
It's always. AI don't know that's it's, it's
35:00
part of it is obviously, yes, I, I can look at the token usage.
35:03
I I understand my infrastructure costs.
35:06
I can sort of, you know, shake my finger in my mouth and, you
35:11
know, wag it in the air and say, OK, like, let's come up with
35:13
something. You know, lucky for me, my, my,
35:16
my Co founder, Nico, you know, he did, he did a lot of that.
35:20
He put that, that MBA to work and, and, you know, thought long
35:24
and hard about what these things should look like.
35:25
But, but what we wanted, what we had to do was is, you know, make
35:29
it enticing enough that you know, as, as somebody's looking
35:33
at that price and looking what they can do is, is like, OK, you
35:36
know, from what I get out of this, it's, it's a couple cups
35:40
of coffee. You know, from what I get out of
35:43
this is, is, is not even a meal at a nice restaurant, but I can
35:46
build the anything that comes to mind.
35:49
And so, you know, we, we want to entice people to come in and,
35:52
and start building, you know, obviously the, the, the, the
35:55
tokens don't come for free. Some people are building a ton
35:59
of stuff and and some people are dabbling one or two requirements
36:02
a month. But hey, the big deal, the
36:04
pricing with this stuff is always tricky, right?
36:06
It's all usage based. You're gonna get some folks that
36:09
are gonna use it a ton and other folks that are gonna just be
36:11
dabblers and kind of poke around so.
36:15
Yeah. And so you're charging usage
36:17
based correctly or what they? The users don't see it today.
36:20
So the users today aren't seeing, they're not seeing the
36:24
actual amount of tokens that are being utilized in order to
36:27
generate the requirements, breakdown their tasks, read
36:29
their GitHub files, summarize those GitHub files, analyze
36:33
their code base. They're not seeing any of that
36:35
stuff. We're just giving them a blanket
36:36
price today. Yeah.
36:38
I mean, obviously that will have dangerous.
36:40
See, it's very dangerous. It'll have to.
36:41
Change in the future, right? Pretty with Matt out there, you
36:43
know? What I mean, yeah, with the with
36:46
the Unicorn folks and Matt, yes, it's it's extremely dangerous
36:49
and and obviously we don't have the deep pockets like open AI.
36:51
I think what was their loss last quarter like 18 billion.
36:55
Yeah. So we're not losing anything
36:57
like that. But obviously at some point,
36:59
yes, we'll have to put some some limits in there and kind of
37:03
ratchet it down a little bit so it's not just the Wild West and
37:06
a free for all. Yeah, and you mentioned lots of
37:09
runway, you know, I'm assuming you're in the black.
37:12
If could I just go straight as like 10 bucks per month per user
37:16
times 2000 to give a a close approximation of MRR?
37:22
You know, not everybody's paying yet, so that would be nice to be
37:25
there. No, I, I can't, you know, I'd be
37:27
lying if I said we're, we were there right now, but you know,
37:31
we're, we're, we're dabbling. We're, we're adding more users
37:34
every day. But yeah, we're not.
37:37
The nice thing is not the bulk of the users are not at that
37:40
lower tier builder. They're coming in the middle.
37:42
That one seems to be making the, the most sense for them because
37:44
now they've got multiple projects.
37:46
Everybody starts with one project.
37:48
They do the trial for seven days and then they realize like,
37:51
shit, I can build a lot with this.
37:52
Let me add some more, you know, a few more repos and creates a
37:55
few more projects and they they kind of go from there.
37:58
Cool. And what's the end game when
37:59
you're looking, you know down the road?
38:01
Is it is this like a logical acquisition for a cursor or
38:05
Claude or something like that or do you guys want to stay
38:07
independent, maybe acquire some new companies, new capabilities?
38:10
What are you thinking is kind of the the next the next couple of
38:13
steps I. Think the most important thing
38:15
for, you know, at least from my standpoint, is I want to be able
38:19
to turn anyone with an idea into a builder.
38:22
And, you know, before I start thinking about or we start
38:25
thinking about, you know, acquisitions or, you know,
38:28
anything else like that is, you know, what is the, the next, the
38:32
next thing we can add to, you know, the, the vision for the
38:35
platform and, and where do we go?
38:37
So today it's, it's, you know, you come with a code base, you
38:40
connect it, you know, you build requirements, you, you create
38:47
those into tasks and you, and you let you know, you let the
38:49
agents go and code. The thing that we've found is
38:53
everybody wants to build AI native SAS products like that is
38:59
that is what we see being built, I would say 60 to 70% of the
39:03
time. And, and if you look at sort of
39:06
the core tenets of a, of a SAS product or an AI native SAS
39:09
product, these things all need the same thing, right?
39:12
They need multi tenancy, they're going to need a database,
39:14
they're going to need some level of authentication, they're going
39:16
to need a way to easily create agents.
39:20
You know, they're obviously a whole bunch of other things, but
39:23
we want to get it to the point where you come in with an idea
39:28
and from that idea, you have that base already ready to go.
39:33
And so you come in and say, like we always like to use I, I, I
39:36
want to build a workout generator.
39:38
For whatever reason, a lot of people gravitate towards that
39:41
when they're talking about creating an agent or something
39:45
else like that. So like magic, you want to
39:46
create an AI powered workout generator.
39:50
We want you to be able to come to brain grid type out describe
39:54
that AI powered workout generator.
39:56
We ask you a few questions and without you ever writing a
39:59
single line of code, you've got a fully functional multi tenant
40:05
AI powered workout generator from day one that's customized
40:10
based upon how you've answered those questions and and what it
40:14
is we teased out of you for what's in your head and.
40:17
We've got that intent that's been floating around there for
40:19
the last 10 years and pulled it out of you, turned it into the
40:22
requirements and the tasks and without you ever even going into
40:25
claw code or anything else like that, you've got it there and
40:29
working. First thing it says is burpees,
40:32
and then you just burn it. You're like, no, man, I'm not
40:33
doing burpees, man, no. Freaking burpees, man.
40:36
Those are miserable. Yeah.
40:39
It's like, no, but it's like it's like it's like because we
40:43
like when we see people like there's no differentiation in,
40:47
in creating a multi tenant product.
40:49
There's no differentiation in the auth.
40:50
There's no the differentiation is what comes from your head and
40:53
your ability to execute that idea and get it out to the
40:56
market and and let people use it.
40:59
Man, I'm getting so excited. I'm getting fired up.
41:03
That's kind of that's the horizon, that's sort of the, you
41:05
know, that's our, that's our vision.
41:06
That's where we want to take this thing.
41:08
So. Love it, love it.
41:11
Warning, we're entering the the lightning round now.
41:13
Just to let you know you're obviously very bullish for vibe
41:18
coding. What's the bear argument against
41:21
a vibe coding? This whole thing's just going to
41:24
collapse and fall down and no one's going to be able to ever
41:26
run any of this stuff. And it's going to be a, a
41:29
spaghetti, you know, a spaghetti shit show of, of code that can
41:33
never be maintained or iterated on or or built on top of and it
41:37
will just all fizzle out. And the vibe code haters are
41:41
very loud and snobby. I would say of course, of
41:46
course, because I'm there is a contingent of people out there
41:50
that are like, look, I went to school for four years or six
41:55
years. I've got 15 years of experience.
41:58
There's no way somebody that runs a flower shop should ever
42:02
be creating a multi tenant AI powered SAS app because they
42:06
don't have the, they don't have the credentials.
42:09
I think AI is, is, is changing all of that and making it a
42:12
reality. There are plenty of people that
42:16
don't want to see to succeed. All, all the AI, all the code
42:19
that's produced is total trash. I, I don't believe that.
42:23
And you know, it's, it's, you can never run that.
42:25
It's never going to be safe. It's, it's never going to be
42:28
testable. I mean, what whatever argument
42:30
you want to put in place, take a step back before AI.
42:34
Think about all the work that went into, you know, enterprise
42:38
software company or startup software company.
42:41
Think of all the time spent and effort around testing procedures
42:45
on code styles on, I mean, you name it on, on, on all of the
42:52
safeguards that we had to put in there to protect our users from
42:55
our software engineers. Well, it's it's the same thing
42:58
in this world too. Like you, you can't just unleash
43:02
vibe coding software up, you know, out in the wild and expect
43:05
it to behave perfectly and, and amazing.
43:08
You still need to do all of the all of the proper software
43:11
engineering techniques and procedures, you know, to ensure
43:15
that what's going out there is actually going to work.
43:18
Maybe that'll change someday and and it'll become so good that
43:22
it's self correcting and, and rewrites itself and does all
43:24
these things. But today's world with AI, like
43:27
you still have to do all of that same stuff.
43:30
Yeah, it's just instead of the humans writing it, now the
43:33
agents are. And instead of it being one
43:35
human writing it or one agent, I've got 6 agents writing this
43:38
stuff at the same time. And I'm sure somebody's going to
43:40
listen to this. Like, do you have 6 agents
43:42
writing slop? Well, maybe, maybe not.
43:46
I've seen plenty of human slop come, you know, come out of, you
43:49
know, very good software engineering houses as well, so.
43:53
It's it's kind of like when your kids teacher is writing you an
43:57
e-mail about they don't want you, that you, they don't want
44:01
the kids using LLMS to help with their writing and their e-mail
44:04
is so flawed that you're like, dude, you should actually be
44:07
using LLMS. Have you heard of Grabberly?
44:10
Yeah, exactly. Yeah, I'd appreciate it, you
44:13
know. Was it next time you send me an
44:14
e-mail, right? But they're threatened and
44:17
they're scared and they should be for sure.
44:19
What what would you say to, to soothe their fears?
44:22
What should they be thinking about to to maybe protect
44:25
himself a little bit of the incoming florists?
44:28
Fucking embrace it, become a master be let it become your
44:33
disciple. Like, treat it as if you are
44:37
like controlling a horde of senior software engineers that
44:43
you want to do your bidding and you want them to behave in the
44:47
manner in which you would like them to write code.
44:51
Give them all the style guides, tweak your Biome and your ES
44:55
lent to whatever degree you want to ensure that that TypeScript
45:00
code is, is beautiful and, and, and, and, and meet your
45:05
standards and then turn them loose because it's, it's, they
45:10
were sort of like unburdening yourself with the syntactical,
45:16
you know, not necessarily nightmare, but the syntactical
45:18
burden of implementation. Now you can take a step up and
45:22
really think deeply about what it is that you want to build and
45:26
how it needs to be built. You don't have to just go hands
45:30
off and, and, you know, hands off the wheel and just let it
45:33
go. I mean, that's one of the
45:35
reasons I, I, I love working with my own product is because
45:39
the ideas and the intent and how I want it done can flow out, you
45:44
know, into brain grid and I can read its output and I can sit
45:49
there and do the exact same thing as if a, if a human had
45:51
done it, you know, two years ago.
45:53
And I can be like, no, I don't like that.
45:54
The decision right there is, is a bad decision that's never
45:58
going to scale. Fix it, change it, you know, and
46:01
of course it's going to be it models can be very polite and
46:04
it's going to be like, oh, you're right, you're so
46:05
brilliant. Let me fix that for you and
46:08
thank you for pointing that out. But you know, sometimes I don't
46:10
know why it does that. But anyways, like you can have
46:13
this discussion, you know, it had a completely different level
46:16
now. And then once you've gotten it
46:19
to the point, it's like you're still putting in all the same
46:22
things as if you're writing it yourself, except for you don't
46:24
have to burden yourself with that syntax anymore.
46:27
And now you can spend up five or six of those things at the same
46:30
time and, and let that thing churn and burn.
46:33
And then you take the step back and it's like, let me go review
46:35
that code. OK, that PR is done.
46:37
Let me review that. I don't like it.
46:39
Well, how often do you do as when, when humans were reviewing
46:43
other humans code? Do they look at that be like,
46:44
man, that's garbage. I don't like that.
46:46
Fix it all the time. It's constant, you know, and
46:51
then it's like you get into these, then the humans get into
46:53
these massive political discussions over variable naming
46:56
and all this other crap. Well, guess what?
46:58
That doesn't happen anymore because the AI agent doesn't
47:01
give a shit. You want it to be camel case,
47:03
You want it to be Pascal case. Just tell it and it's going to
47:06
look exactly how you want it, but embrace it like if this is
47:10
this is this is the world that we were going to be in.
47:13
So I think embrace it fully and become a Jedi Master at this
47:17
stuff. Yeah, yeah, You know it.
47:19
I wonder if the tension is is like similar to maybe like a
47:23
musical tension. So if you want to go study
47:25
music, there's kind of two paths.
47:27
You can go to the orchestra and they put the charts in front of
47:30
you and you play how it's written.
47:31
You can go study jazz and you learn how to improvise.
47:35
Now as founders, I feel like we're all kind of jazz
47:37
musicians. We want to improvise.
47:38
We have an idea, we want to express ourselves perhaps.
47:41
But perhaps it's the orchestra that's like, hey, you know,
47:43
actually I like the comfort of the structure.
47:46
And it's, and that's what is feeling threatening to them
47:48
because now they're going to have to open up.
47:49
They're going to have to use, you know, the other side of
47:52
their brain, you know, in order to get done what they need to
47:54
get done. I mean, absolutely.
47:57
I mean, I, I, I like the analogy of, of being jazz musicians.
48:01
It's like, yeah, you, you've got to improvise all the time,
48:03
especially as a founder. Because you know what if, if
48:06
there is a road map to being a successful founder, trust me, we
48:09
would all be doing it right? We'd all be following that path.
48:12
But that's, that's just not the case.
48:13
There's a lot of improvisation and a lot of learning and a lot
48:16
of failure, actually, unfortunately a ton of failure.
48:18
And you know, you've got to learn from it, pick yourself
48:20
back up and go battle again. Totally.
48:24
OK, so as founders, we're jazz musicians and dopamine addicts
48:28
naturally. So maybe you can help us with
48:32
shiny object syndrome. So which is which is to say that
48:36
how do you stay focused working on Brain Grid and not just using
48:41
Brain Grid to make 1000 different projects at the same
48:44
time? It was like, well, I, I love my
48:48
own dopamine hitch, not only obviously using brain grid,
48:51
implementing brain grid, watching what my customers are
48:53
doing, watching my customers succeed with Braingrid.
48:57
And that's what keeps me locked in.
48:59
I mean, every, you know, every time I, I see some crazy idea
49:04
come across that somebody is building and implementing and,
49:07
and then maybe giving us feedback about like, holy shit,
49:09
I couldn't believe I built this. Like I had no idea I was going
49:11
to do this. And like I was able to do
49:13
something I never thought was possible.
49:14
Like that's, that's why I get up in the morning.
49:17
That's what keeps them going with, with Braingrid.
49:20
Yeah. I mean, sure, yeah.
49:21
But I want to, are there other things that I want to build?
49:23
Like, yeah, when I was learning more about coding with AI, built
49:27
my kids a math tutor for as they were learning something called
49:31
Singapore Math and I didn't know what the hell that stuff was.
49:33
And very different than the way I learned math.
49:36
So I'm like, well, guess what? AI is pretty good at this.
49:39
So let it let it teach them. And I, I did at one point build
49:43
a workout generator here as I, you know, as I was learning my,
49:47
my own path of figuring out how to be efficient with, with AI,
49:50
there's, you know, a number of other little things that I've
49:52
done here and there. This, you know, has 1000% my
49:56
attention, attention. And I I want to see people be
50:00
successful with their ideas. See my daughter's a freshman in
50:05
college. Shout out she listens to the
50:07
podcast. Big fan of the podcast.
50:11
What should she be studying? How to think critically, Think
50:16
AI is not going to take that away.
50:19
It's not going to ever think for it.
50:20
Well, I shouldn't say ever. It's not going to think for us
50:22
today. You need to question.
50:25
You need to be able to think. You need to be able to, you
50:28
know, understand the things in your world that you know when AI
50:33
is presenting to you, you have, you know, you have two choices.
50:37
You can just be like, OK, fine, that's the Internet setting.
50:40
It must be true. Or you can look at it and be
50:42
like, I don't know about that, you know, I'm not so sure.
50:47
You also be able to think because again, when something is
50:51
in your head, it's not like AI can reach in there and pull that
50:55
out and say like, oh, you're, you're thinking that over in
50:59
this part of the brain. I know exactly what you want.
51:00
Let me go do that for you. You have to be able to describe
51:03
it still, you know, maybe you can use less words to describe
51:08
it than you would need to in the past, but didn't or describing
51:11
it to a human, but you still need to be able to articulate
51:15
it, what it is that you want these models to do for you.
51:19
And so I think that, you know, learn to think, be, be a great
51:23
thinker if you're, if you're going the college route,
51:26
otherwise trades are a great place too.
51:29
AI is not going to replace plumbers, electricians,
51:31
carpenters or any of that other stuff.
51:32
And, and, you know, there's a, there's a lot of stuff that can
51:35
be done over there too, But I, I think just be a great thinker
51:39
and, and, and understand what those tools are like.
51:41
If you just use the tools to do your work for you, you're not
51:45
going to get any benefit. But if you use those tools to
51:49
help you research better, to research, research more
51:52
efficiently, to learn better, I, I I think you're doing it the
51:56
right way. My daughter's very
51:59
entrepreneurial, so if you're listening, sweet baby, it sounds
52:02
like workout generator. No burpees.
52:06
Sell it to your college friends, It's on fire.
52:09
We're good. Well, the, the thing that we
52:11
found that everybody wants to do when they're doing the workout
52:13
generator is describe what tell us all about the equipment you
52:16
have in your garage so we can tailor the workout to what it is
52:20
you have in your garage. Because I guess after the COVID
52:24
times, right, or during the COVID times when we everybody
52:27
was at home locked up and and couldn't go to the gym, A lot of
52:30
companies supplemented that gym membership by giving them cash
52:34
to fill out their home, you know, to fill up their home gym.
52:37
So a lot of home gym people now so like, you know, like bro went
52:39
crazy and sold tons of squat racks and whatever else.
52:42
So. Oh I know, I got some kettle
52:45
bells gathering lots of dust out there.
52:49
Right next to the rowing machine, right?
52:51
Yeah. Exactly.
52:53
Amazing. And Speaking of college, I we
52:55
want to give a shout out to the Guardian Scholars.
52:58
You mind give us a quick introduction to the Guardian
53:00
Scholars? Yeah, the Guardian Scholars was
53:03
started by Ron Davis, and Ron Davis was the, well, I don't
53:09
screw this up, the former CEO of Perry Group, which is bought by
53:15
Nestle. He was very philanthropic and
53:18
done very well. He got to a point where he was
53:21
tired of just writing a check and and really wanted to start
53:24
something that he was deeply involved in and deeply cared
53:29
about. So he put those efforts into the
53:30
Guardian Scholarship. The Guardian Scholars, sorry,
53:33
the Guardian Scholars and the Guardian Scholars focus on youth
53:38
that have been foster care. Guardian Scholars provide
53:41
scholarships. It was started first at Cal
53:44
State Fullerton. My wife actually was the very
53:48
first Guardian Scholar. It is.
53:51
You can look online. It has gone on, there have been
53:53
multiple chapters started all over the country.
53:55
I don't remember the numbers, but it's it's, it's pretty big.
53:58
I know Ron has most recently started another one at Colorado
54:03
Mesa where he spent a lot of his time Business School that he
54:08
named is, is up there and he is very deeply ingrained in that
54:11
community helping foster youth and and helping them realize
54:16
that that dream of, of getting a college education.
54:22
Amazing. And if people want to check it
54:23
out, we'll put a link in the liner notes for Colorado Mesa
54:26
University. Shout out to them.
54:28
Just down I-70, Best Peaches in the world down there.
54:31
It's supporting cmu.org where you can learn information.
54:34
Tyler, this has been absolutely incredible.
54:37
Your website is Brain Grid dot AI.
54:40
Those will be in the show notes. It'll also be in the resource
54:42
first page under Vibe Coding at AI for founders.co.
54:45
What's the best place for people to connect with you online?
54:48
I mean, I'm on LinkedIn, Tyler Wells brain grid I'm on I always
54:54
want to say Twitter, but X now or that's Tyler S Wells.
54:58
I know Tyler Wells on on Twitter.
55:01
Yeah, I don't I don't do a ton of social stuff.
55:03
It's I'm pretty busy building, but I can always be found on
55:06
LinkedIn. I can always be I'll pop into X
55:10
here and there so them around those things, but find us, find
55:14
us a brink grid and that's the place to go.
55:17
We do have a public slack channel.
55:18
So, you know, you sign up, we can share that with people and
55:20
they can come in and, and we can chat channel about how to get
55:24
the most out of your, your coding agents.
55:28
Amazing. Tyler.
55:28
Thanks for joining me. My pleasure.
55:30
It's great being here. Thank you, Ryan.
55:34
Hey founders, let me ask you this.
55:36
Are you building with AI or getting left behind by it?
55:40
If you're serious about scaling smarter, faster and bolder, you
55:44
need to check out AI for founders.co.
55:48
It's your go to resource for cutting through the noise and
55:51
actually using AI to grow your startup.
55:54
Each week, AI for Founders drops a sharp, no fluff newsletter,
55:59
dives deep with expert guests on the podcast, and hosts monthly
56:03
workshops where you can get your hands dirty with the real tools
56:06
and strategies. Whether you're precede or Series
56:10
B and beyond, launching something new or just staying
56:13
sharp, AI for Founders gives you the edge you need to win in 2025
56:18
and beyond. Hit.
56:19
AI for founders.co to subscribe to the newsletter, tune into the
56:24
podcast and grab your spot in the next workshop.
56:27
And it's all free. Build smarter, move faster,
56:30
think bolder with AI for Founders.
56:33
Now back to the show.