HomeVideos

Testing AI Web Design on a Real Business: The Results (w Genspark)

Now Playing

Testing AI Web Design on a Real Business: The Results (w Genspark)

Transcript

279 segments

0:00

Every make money with AI video sells you

0:02

the same fantasy. Type one prompt, wake

0:04

up rich, blah blah blah. I'm going to do

0:05

the [music] opposite. I'm going to take

0:07

one real local business today, build

0:09

them a real website, and try to land

0:10

them as a paying client. And I'm going

0:12

to show you where it actually works, and

0:14

where it quietly builds you. I just want

0:16

to be up front about it. This video is

0:17

sponsored by GenSpark. They paid me to

0:19

test their tool. They did not pay me to

0:20

lie to you. And I kept [music] full

0:21

control of the verdict. So, you're going

0:23

to watch me hit the limits live, not

0:25

just the highlight reel. My name is

0:27

Daniel Jindeel, and I used to run an AI

0:28

info business that questioned my moral

0:30

every single day, and eventually I had

0:32

to quit. Now, I'm just giving away

0:33

everything I know about AI for

0:35

completely free. [music] There's no

0:36

gatekeeping. You can join my free

0:37

community and newsletter to get as much

0:39

value as you can. You can check it all.

0:41

All I'm asking for is maybe hype this

0:42

video and leave a comment [music] down

0:44

below. Now, here's the actual test. And

0:45

by the end, you'll have a repeatable way

0:47

to land your first paying client [music]

0:48

with or without this tool. Most people

0:50

never land that first client. It's

0:52

actually hard. And it's not a talent

0:53

problem. To sell a small business a

0:55

website, [music] you normally need to do

0:57

five separate jobs. You research who to

0:59

pitch, you design something, you build

1:01

it, you host it somewhere, then you

1:02

write the [music] outreach. That is five

1:04

tools, five logins, five things to

1:06

learn, and most people quit at tools

1:08

number two. So, the real bottleneck was

1:09

never the idea. It was the friction.

1:11

[music]

1:12

Too many moving parts, so the project

1:13

dies on your laptop. So, why don't we

1:15

think of it like a hiring a temporary

1:17

agency? You do not do the work yourself.

1:19

You walk in,

1:20

>> [music]

1:20

>> you describe the result you want, and

1:22

the agency staffs the job. The one

1:23

worker does the research, one designs,

1:25

one builds, one writes the pitch. You're

1:27

not the worker on the floor. You're the

1:29

manager. You get the brief, you check

1:31

the work, you send it back [music] when

1:32

it is wrong. That is what GenSpark calls

1:35

super agent. In plain English, it is one

1:37

place where you hand off a whole entire

1:39

project, and a team of these AI workers

1:41

run the steps. [music]

1:42

You stay the managers the whole entire

1:44

time, which I think is really important,

1:46

because the manager is [music] where the

1:47

judgment lives. Hold on to that thought,

1:49

because we're going to come back to

1:50

that, and this is the entire what this

1:52

entire video is about. Now, the actual

1:54

run. Quick note first, because it's this

1:55

team manager idea. A manager does not

1:57

type the work order. They just say what

1:59

[music] they want. And Jen Spark has a

2:00

free voice app for exactly that called

2:02

Speakly. You talk, it just types. Very

2:04

simple. And it has a shortcut that drops

2:06

what it says straight into Super Agent.

2:07

I typed mine so you can read along with

2:09

me. But the voice option is right there

2:11

so the brief, find me five real local

2:13

laundromats with no website. Pick the

2:14

best [music] one, build a site, make a

2:16

logo, and write the outreach, and let's

2:18

go. First job, finding the business. And

2:19

the first honest catch shows up right

2:21

away. Its map search tool actually fails

2:23

first time so it fell back to reading

2:25

online directories, the Yellow [music]

2:26

Pages type listing, and it pulled five

2:28

real laundromats in Burnaby that have no

2:30

real websites or whatsoever. But watch

2:31

this, it gave me two different phone

2:33

numbers for one of them and it flagged

2:35

that itself. It literally told me,

2:36

"Double check this before you reach

2:38

out." So the two finds, you verify and

2:40

it did not magically know these shops.

2:42

It read the listing and told me what I

2:43

was unsure about. That is you being the

2:46

manager again. Before you ever send an

2:47

email, you pick up the phone and

2:49

confirm. Okay, so it picked one shop, a

2:51

family-run place, husband and wife

2:53

owners, free coffee, fragrance-free

2:55

soap, and none of that shows up online

2:57

anywhere. It's [music] a good pick

2:58

because all of that is website gold that

3:00

is currently just invisible. Then in the

3:02

same run, it made a logo, made a hero

3:05

image, and built a full one-page site,

3:07

hours, services, a map, a contact form,

3:09

a few minutes of me doing nothing but

3:11

just watching [music] or working on

3:12

something else. And this is the exact

3:14

moment every other video stops. Shiny

3:16

results, sign up with my link, done. But

3:18

I'm not going to stop there because when

3:19

I look closely, the first pass was not

3:22

good enough. I wasn't really satisfied

3:23

with it. It was bad bad bad. Look, three

3:25

real problems. One, it looked like AI

3:27

generated [music]

3:28

because AI made it. Generic font, a

3:30

boring navy and gold combo you've seen

3:32

like a thousand different [music] times.

3:34

And secondly, the hero photo is melted

3:36

text on the washing machine. That

3:38

[music] word, dreamy AI look. You cannot

3:40

send that to a real business owner even

3:41

for a demo, right? So lastly, and this

3:43

is the one nobody really talks about

3:44

right now, the deployment part. It tried

3:46

to put the website on a real independent

3:48

host and it could not. It timed out so

3:50

it parked the site on a link owned by

3:52

Jimspark instead of one owned by myself.

3:54

So, the catch here is that the easy

3:56

version keeps you on the rails and there

3:57

is a cost catch, too. That first bill

3:59

cost me about 1,000 credits on the

4:01

cheapest paid plan that is 10,000

4:03

credits a month. So, you get roughly

4:04

nine or 10 of these bills before you are

4:06

buying more. So, I did what a manager

4:08

does. I sent the work right back. I gave

4:10

it a clear design direction. Basically,

4:12

a picture of what good looks like and I

4:14

pushed it through two refinement passes.

4:16

New fonts with actual character, one

4:18

confident color use [music] boldly

4:20

instead of timid, real trust signals a

4:22

laundromat customer cares about, card

4:24

[music] and coin big capacity machine,

4:26

free parking, a call button stuck to the

4:28

bottom of the phone screen so you can

4:30

tap it from anywhere, hours you cannot

4:32

miss. And then it tried to add amenities

4:34

it could not actually confirm. So, I

4:35

made a flag those with a verify tag

4:37

because putting fake details on a real

4:38

business site is bad bad [music] bad.

4:40

And a photo, I regenerated it and it got

4:42

really better. But my take on the photo

4:44

is even the good AI [music] version is

4:46

not good enough. I wasn't really

4:47

satisfied. For a real client, you want a

4:49

real photo. Even a photo snapped from

4:51

the owner because AI pictures of real

4:53

places always have a tell if you look

4:55

close enough. And after my passes, this

4:56

looked like a real [music] business

4:58

hired a designer. The jump from generic

5:00

to good did not come from the tool. It

5:01

came from me directly. If the tool is

5:03

fast, the judgment is yours. Now, the

5:04

question my people always ask, "If I

5:06

build it here, am I trapped here?" So, I

5:08

tested it and I exported the whole

5:09

thing. And the part genuinely surprised

5:11

me, you get [music] clean normal website

5:14

file. There's no weird format, no export

5:16

fee. I checked and nothing in those

5:18

files needs Jimspark to keep running.

5:20

It's yours. You can take them and host

5:21

them yourself for free on your own

5:23

domain anywhere you want. Going back to

5:25

temporary agency, the work the crew

5:27

built is yours. You walk out with it and

5:29

set up your own shop on your own lot.

5:31

The only thing tied to Jimspark was that

5:33

easy preview [music] link. Once you

5:35

export that hosted yourself, the cord is

5:37

cut. Now, the question my people always

5:39

ask, "If I build it here, am I trapped

5:41

here?" So, I tested it. I exported the

5:43

whole thing, entire thing, and this part

5:44

generally kind of surprised me. You get

5:46

clean normal, Webflow files. There's no

5:48

weird format, no extra fees, I checked.

5:50

And nothing in those files needs

5:52

GenSpark to keep running. [music] You

5:53

can take them and host them yourself for

5:56

free on your own domain. Probably not

5:57

free, but yeah, you know what I mean.

5:58

Anywhere you want. Let's go back to the

6:00

temporary agency. [music] The work the

6:01

crew built is yours. You walk out with

6:03

it and set up shop on your own lot. The

6:05

only thing tied to GenSpark was that

6:07

easy preview link. Once you export that

6:09

and host it yourself, that cord is

6:11

pretty much cut. So, my take on the

6:12

boundary is this [music] one. Editing

6:14

the site again means another credit

6:16

burning run, which means you rent the

6:18

builder, not the result. GenSpark is a

6:20

fast way to produce the real thing you

6:21

[music] fully own. It is not where you

6:23

run 50 client sites long-term unless you

6:26

[music] are happy paying for the

6:27

credits. Build it, ship it, then own it

6:29

what you made. Let me split the cost

6:31

honestly because these are two different

6:33

things. There are cost to build. Tool is

6:35

about $20 a month plus those credits per

6:37

site. It's real, but very tiny. Cost to

6:39

run the Webflow site you deliver,

6:41

basically zero. Once you export the

6:43

files and drop them on a free host like

6:45

Vercel or Netlify, your client pays

6:47

nothing ongoing and neither do you

6:49

because you [music] own the files. And

6:51

one more thing, GenSpark says they are

6:52

at a $250 million annual run rate.

6:55

[music] What that means in plain English

6:56

on run rate is a pace. You take their

6:58

recent sales and stretch them across the

7:00

year. So, it is not money already

7:02

banked. It's cool. That's their number.

7:04

It's not proof it works for you. The

7:05

proof is what you just watched on my

7:08

screen. So, even if you never touch this

7:09

tool, take the play because the play

7:11

[music] is the asset. One, find local

7:14

business with no website or a bad one.

7:16

They're everywhere if you look hard

7:17

enough, okay? Two, build them the thing

7:19

before [music] you pitch. A real demo,

7:21

not a promise. Three, reach out with it

7:24

already done. [music] So, you're

7:25

showing, not selling directly. Four,

7:27

charge a first-rate to finish it and

7:29

host it on their own domain. That works

7:31

with any tool. [music] GenSpark just

7:32

makes our steps fast. I put the exact

7:35

prompts I used, the outreach email, and

7:36

a single pricing guide [music] in a PDF

7:39

so you can just take it. So, why don't

7:40

you grab the prompts? It's free. It's in

7:41

the link below, if you want the prompts

7:43

and the outreach template [music] to run

7:44

this yourself. If you do want to try

7:46

Jasper, my link below gives you free

7:48

credits to test it, same way I did. Use

7:50

it on a real project, not a toy. That is

7:52

the only way you learn if a tool [music]

7:54

is worth it, you know? And remember that

7:56

one line that matters here, the tool is

7:57

fast, but you are the manager still, and

7:59

the judgment is what gets paid. [music]

8:01

Anyway, that is for today. If you want

8:03

to see more, why don't you comment down

8:04

below on what you want to see next? For

8:05

the next video, I'm going to cover

8:07

hidden open-source repos every local AI

8:09

enthusiast should know right [music]

8:11

now, because that is how our channel

8:12

runs. You ask, I build, we all learn.

8:14

See you in the next video.

Interactive Summary

This video provides a realistic assessment of using AI tools to build websites for local businesses as a way to land paying clients. The narrator demonstrates the 'Super Agent' feature of the GenSpark tool by building a website for a laundromat, emphasizing that while AI can automate labor-intensive tasks like research, design, and coding, human judgment is essential for quality control, verification, and final touches. The video concludes with a repeatable, four-step strategy for landing clients that works whether or not this specific tool is used, highlighting the importance of delivering a tangible demo instead of a promise.

Suggested questions

3 ready-made prompts