Jonathan and Caleb discuss how we can use actual robots (artificial intelligence)
Follow along as we attempt to build and sell a SaaS company. A build in public behind the scenes journey of a small software agency, Two Story Robot, trying something new.
Caleb: I really want to start
with hello for some reason.
Jonathan: you can say, hello, you
do you do whatever you want to do.
Caleb: Okay.
Hello everybody.
And welcome back to The Robot Factory.
I am one of your hosts, Caleb Sharp.
Jonathan: And I'm the other
host, Jonathan Bowers.
Good intro Caleb.
Caleb: thank you.
This is episode 9, 8,
7.
Jonathan: who knows some
of these other episodes?
I might, I might
Caleb: This might be episode three.
Who
knows?
We do have billing setup now.
Jonathan: Yeah.
So
Caleb: there's nothing stopping us
from just make it heaps of cash at this
Jonathan: heaps, oh, money
Caleb: Like Scrooge McDuck levels of cash
Jonathan: just a big
pool of coin that we dive
Caleb: coins specifically, I set it
up so we can only get paid in loonies
Jonathan: Nice.
So yeah, we've, we've got billing
set up, which, which ended up
being, not that hard, right.
It was maybe how much work,
how much effort that take?
Caleb: Well, I think we
actually, we talked about this.
I tried to guess how much
Jonathan: Oh yeah.
Caleb: Like I think, I think I said one
day, and then I said to just in case.
Jonathan: Two, just in case days.
Caleb: I think the guts of it ended
up being that amount of effort.
I think it was a lot more complex than
I thought it was going to be, but I
Jonathan: Yeah.
Cause there's some edge.
There's some funny edge cases, not
edge cases, but like stuff we've never.
Uh, done this before, like we've
worked on lots of products, but we tend
to just hand them off to people and
Caleb: Yeah.
Jonathan: either do invoicing
some other way or whatever.
Um, but it's kind of the first time we've,
we've really integrated with Stripe and
Caleb: We never really had like
revenue directly from consumers.
Uh, product
Jonathan: yeah, I mean, what, what were
some of the, some of the challenges
that you hadn't really anticipate?
Which if we were to like talk to
somebody about like, they would
say, yeah, this is kind of obvious.
Caleb: Yeah.
I mean, in hindsight it's like, well,
obviously, but it's a good thing to know.
It's there figured out, especially
specifically with Stripe, how to do it.
Um, but it's mostly with subscriptions
and how subscriptions work cause.
I don't know, I just assumed subscriptions
could either like exist or not exist,
but there's like a lot more states
that you can have for subscription.
You can have a subscription
that is active and recurring.
That can be a subscription that's active.
Um, but it's going to
expire and not recur.
You're gonna have a subscription
that's expired and you have the
ability to resubscribe can have
a subscription that's canceled.
There's just, yeah,
there's a lot of states.
some of them are harder to
deal with that than others.
for the most part Stripe makes it pretty,
pretty easy to deal with, which is nice.
And it still does all the heavy
lifting of actually like collecting
the money, which is really
Jonathan: Yeah.
Caleb: the hard part, because if something
goes weird with a subscription, I mean,
we can, we can just manually fix it.
So it's not the end of the world.
yeah, it's pretty sweet.
There's like an account page now.
It's.
Thankfully all the billing
management is handled through Stripe.
So I didn't have to implement all of that.
Cause that that's a lot.
Jonathan: There's a lot of
stuff you just don't have to do.
We thought we could get away
with using this feature in
Stripe called payment links.
Um, but that, that ended up being
not quite, quite what we wanted.
So we just sort of integrated a little bit
more tightly with Stripe, but that's fine.
Um, I was amazed at how fast it
was to set up a Stripe account.
Like you were ready to do it.
said, give me some, give me some time.
I'll go and focus on this this morning.
Um, it, it is like the third
time I've done this though.
Cause we've got another product
that we've set up a Stripe account,
but um, I'm still not like super
practiced at it and I'm not sure.
But this is what they do is just
create Stripe accounts all day.
So I don't know how much experience
you can get from it, but it was fast.
Like I, you know, it punch
in the banking details.
I don't know how they verify that
they must have already know it because
it's the same banking details as
our, we have other Stripe accounts.
And so it's the same banking details.
So I didn't have to go
and verify anything.
I'm pretty sure I had to do that
Caleb: wow.
Jonathan: once before.
Um, but yeah, you just punch in,
just punch in some legal information.
And it's just kind of ready to go.
It took maybe half an
hour and yeah, it's just
kind of done.
pretty slick.
I have no idea how complicated it is
to transfer these accounts around,
but it is a separate Stripe account.
That's not, that's not the same
account as, um, when we invoice
other customers for other things
that we have that are completely
unrelated to OpenHouse.social
so we've
got a separate, a
separate account for that.
So fingers crossed.
We get acquired It'll be really
easy to just transfer this thing.
but I was amazed just how fast
it is to start this stuff up.
It's super cool.
I was reading, I was doing my best to
understand if this is unintended use case.
And it seems to be that, like,
you create just separate Stripe
accounts for different product lines.
even if you don't intend on having them
be acquired, it can just simplify a lot
of things, because one, you just sort
of have You know, a little bit more
control to who has access, um, little
bit cleaner on the products we need to
finding products and pricing and stuff.
and it can all go back
to the same bank account.
You don't need like a separate bank
account for each, for each Stripe account.
So that was pretty sweet.
I'm excited to get some revenue in there
because at some point I want to also
plug in those charting libraries that
let you share revenue numbers publicly.
I've been listening to a few podcasts
of, uh, I can't remember the name of it.
They were just acquired and.
Eh, anyways, they said that like, you
can do this with Stripe by itself.
Like it has some metrics or analytics
or something that you can, you can,
I don't know if you can share them,
but you can definitely look at them.
And they said that they aren't very good.
Um, and there's a better, I mean, of
course they're going to say that I
suppose, but I believe it, and like,
like bare metrics is one example.
I don't think they have sold, but
anyways, like bare metrics is, uh, a
service you can plug your Stripe account
into, and then you get this, this
public dashboard that you can share.
I hope that we can also share
the, uh, website analytics too.
We've got that set up.
we're using Fathom, which is cool.
Cause it's pretty privacy focused.
So we don't have to have those silly
warnings asking you to accept things.
Cause we don't, we don't
really track anything.
So I was also talking with a friend who.
Uh, runs a small agency.
And so they do social media
management for a real estate agent.
he had some interesting things to say,
like, he, he said that ultimately like
making these images isn't that hard.
Cause they he's very
comfortable using Canva.
Um, it takes like five minutes,
five minutes of his time.
but the thing he said, oh.
He, I asked him if it was well,
what if, what if you could just
like, make this all go away?
Is that, would that be helpful to you?
He's like, oh yeah.
It's kind of like a pain in the butt.
Like it's just, it's distracting.
It doesn't take very long, but
it's like, Ugh, it's tedious
and I don't want to do it.
And it's only five minutes though, so I
don't know how much that's worth to him.
But the thing that he said was.
Was a real pain was actually getting the
information from the customer, like from
the real estate agent about what to post.
And so he gets the images and,
he said it's hard to like get
them to write, the writing.
so
Caleb: actual content
Jonathan: yeah, like the caption or not
the caption, but yeah, like when you, you
Caleb: I have a description.
Jonathan: Yeah,
the description or the like on
Facebook, I think it's just called the
post, like the body of the content.
it's hard to get that.
And, um, coincidentally, I also got
access to a DALL-E the, this
week and was playing with that
and like really having fun time
generating, generating art.
And then it was poking
around with GPT-3, which is
this
other, so DALL-E is a, um,
AI generated art.
if you don't know what it is, just go look
it up.
D A L L dash E I
think, um, and then one of
their other products, this AI
company is GPT-3.
I don't
know what that stands for, but
it's a, it's a natural language.
Um, to also it can, it can create.
Writing for you, you can just give it
prompts and sort of tell it what to do.
And it generates this stuff.
And so I was playing around with
generating, um, descriptions of
listings by just sort of bullet points.
Like just, I just like type in some
bullet points of what was there like
three bedrooms where it was located,
what it was located near that's it.
And then told, told a
GPT three to generate a.
Uh, property listing description
for it, for Facebook and it
produced some really good results.
So I like, I really want to plug this
in to the app because he said, that's
like, one of the biggest problems is
trying to come up with this content.
And so, I don't know, it feels like
it wouldn't be hard to wire that stuff
Caleb: Yeah, I think it'd be super
cool because I don't think I've seen
like an actual, like kind of business
use case for GPT-3 that actually
like doesn't seem super gimmicky.
Jonathan: Well, I mean, I think I I've,
I've seen lots, like there's, there's
a lot of really cool tools out there
that I think, um, work really well.
Like,
Caleb: but it's like exactly
solving a specific thing
that people are annoyed with.
It's not like a weird way of
solving it or like a roundabout way.
So me it's like it it's literally
exactly what your friend asked.
People don't want to
do it cause it's hard.
They just it's frustrating.
Cause it's, it takes up time and it it's
Jonathan: Yeah.
So I kind of want to.
Explore adding some of those features.
We're not going to do them right away.
I think we need to get some, get some
people, like at least trying the thing
out and giving us some feedback and
then tease, tease out some of that,
some of that idea of adding AI to it.
Um, but it looks phenomenally easy.
Like it's, it's almost
offensively easy to add this.
Um, based on, based on, uh, just
a little bit of reading I was
doing around how the API works.
so I'm keen, I'm keen to
explore that a little bit.
I mean, one of the reasons I
was using DALL-E was to generate,
I was playing around with generating
different cover art for our podcasts.
So I'm going to, I'm going to share those,
um, cause they're, they're super fun.
They're very impressive.
Um, I'll share those in
the, in the show notes.
Um, but so since talking with, uh,
my buddy who runs the agency, he
sent me a few, um, uh, for real
estate agents that he, uh, manages.
And so one of the things
that I've noticed.
And then I've started poking around
in a few different, real estate
agents, social media presence is
it's not always pictures of listings.
Like sometimes it's pictures of listings.
Sometimes it's also like little
tips as like big sort of, I
don't know how to describe it.
Like, if I look at someone's
Instagram page, it's like
pictures of interesting things.
And then occasionally like a quote or a,
uh, a little bit of advice or something.
And so I thought, I thought
that was interesting.
I don't know if we want
to do anything about that.
But the other thing that I noticed is
they were rarely posting single images.
It's almost always.
Many many photos, right?
Like it does.
I agree with that.
It doesn't make sense to just post a
picture of the outside of the house.
Like they're posting a picture of the
outside of the house at the, you know, the
bedroom, the dining room, the bathrooms.
Um, maybe not all of the photos,
but at least at least more than one.
And I think, again, I don't think
we should act on this just yet.
Cause it's just a suspicion or a hunch.
I think that we'd want to be able to.
Cater to that.
So if a, if a real estate agent is
uploading, say 5, 6, 7 images, um, how
do we return those images to them, or
even just automate the posting directly?
So they just upload some images and
we like do the cropping, do the, um,
do the template to get that image
with their face and their logo and
their phone number and stuff on it.
Maybe on the first one, or
have you just on the last one.
Um, match it
up with GPT-3 and get the
nice description.
It just posts straight to Instagram.
To LinkedIn, to Facebook, whatever.
Um, but also catering to that idea of
having more than one image and making
it like, I dunno, like, as I'm looking
through, as I'm looking through somebody's
images as you scroll through, the
edge, the edge is strange sometimes.
Like sometimes it's like,
am I actually looking at.
A divider in the middle of the
room, or just because of the way
the pictures kind of line up.
I wonder if, if it's like, I wonder
what it is that I'm actually looking at.
And it makes me wonder if, if
there's some interesting thing where
we could like, sort of tile them.
Across and have like a little bit of
overlap between images to one, like
a, it's a technique that we often
use in when we're building apps to
signal that you can, you can swipe
through something is, will sort of
tease out the fat or hint at the fact
that there's something else over here.
And so like show you a little piece
of it, like the left edge of it on the
right side of your screen or something
so that, you know, like, oh, there's
something over there I'm gonna slide over.
Could we do something like
that on Instagram posts.
And so you could have the picture of
like the main listing, but just slightly
to the right of that is the hint that
there's something else to swipe through.
And I know there's like indications in
the Instagram UI that shows you that like,
yeah, there's like more pictures here.
You can see the little dots down there,
so you can, you can roll through,
but could we make that even better?
Yeah.
Caleb: Interesting idea.
Yeah.
I know what you're talking about though.
There's no gap between pictures on
Instagram and you swipe through.
I think sometimes people use that
the post, the thing, like a Panorama
or something and then cut it up
and then it looks kind of cool.
Cause
Jonathan: I like that.
I think that's
Caleb: scrolling through one image.
Jonathan: I think that's kind of
cool, but I think you could, I
think you could do that in a way.
That's a little bit, more deliberate
about the fact that these are actually
separate images, not, not a single
image that is cut up into smaller
pieces that you're scrolling through
Caleb: Maybe, I don't think Instagram
users have trouble knowing when
there are multiple images or not.
Because like you said, there is UI and
Jonathan: no, I suppose not.
but even still like, like I found myself
as I'm scrolling through that, that post
sort of wondering, am I looking at the
edge of a photo or am I looking at a weird
wall divider in the middle of the room?
Because they were sh they were,
they were pictures of rooms.
And often it was like on the right
side of the one picture was a wall.
And in the left
right next to it was a wall.
Caleb: It was like some non-Euclidean
geometry of a house.
Jonathan: Yeah, my shooting, like
directly at the wall, the corner
of a room, like, what is this?
And it was it just like, it just
jarred me momentarily at that.
Oh, I wonder, I wonder what this would
look like if it was a little bit more
deliberate and like clearly defined.
And then I think it also gives us some
opportunity to add, add some more.
Um, I don't know if some more, some more
words or some more decorative content
or decorative imagery or something.
To just give it a little bit of pizzaz.
Caleb: yeah, I think
multiple
images would be, yes.
Jonathan: yeah, and, and giving,
giving the user some choice about how
those will look when the generation,
cause I don't think it makes sense
to generate them all using the, using
the template that we have currently.
They shouldn't all have the real
estate agents face and logo and price
and that kind of stuff like that.
That be on one of the.
Caleb: I think having different,
like maybe pages would be a good one.
But different
Jonathan: Yeah.
Pages.
Caleb: of pages, I guess.
So that, like, I don't know, it adds
a little bit of kind of authorship and
creativity for like a real estate agent
or, or whoever's doing the social media to
like pick and choose what they want to do.
So it doesn't feel so robotic.
Jonathan: And I also think like the other,
a couple of other things that I noticed,
what, well, one thing that he shared with
me was, um, his client shared with him
that they get compliments all the time
from their real estate agent friends about
how great social media page looks like.
And so he, like, he's trying to
turn that into referrals, but,
but people really compliment this,
this, uh, this real estate agent on
the way their social media looks.
And so I think that's
like, I think that's an.
I don't know, that's an
interesting point of pride, right?
Like to be like, have people
say, Hey, this looks really good.
Right.
I think that's, I think that's a nice
feeling to try and create for people.
So the other thing that I noticed
is, maybe this is just the
example that he showed me because
he's using a Canva template.
It's not
actually changing that much
between posts and you really start
to notice the lack of variety.
and so one of my fears is that the
template that we have right now is
going to be tiresome very quickly.
Like after three or four posts, it's
going to be like, okay, look, I don't,
I'm, I'm sort of sick of seeing this.
like we've never actually taken
the images that we've, created and
mock them up as what it would look
like on someone's Instagram page.
If you go and look at it, I
bet it doesn't look very good.
Even if they're, even if they're
sort of breaking it up with
other pictures and other posts.
I bet it looks repetitive and
unpleasant, so I think some, variety
and maybe a little bit more simplicity.
would be helpful.
Caleb: Yeah.
Jonathan: What is.
The next thing.
we have a hard deadline of,
we want to launch this thing.
Absolutely no later than August the
11th, because that's when we're going to
announce it to a local group in Kamloops.
I think we should try
for sooner than that.
How much more do we have to do?
Like what do we got to do?
Like we're kind of ready, right?
Caleb: uh, huh.
I,
I think, yeah, there's lots of like a.
Uh, ideas of things that we could
still do, but I, I think it would
be really easy for us to just keep
adding features and have no users and
Jonathan: Nope.
Let's not
do launch it.
Let's launch it.
And then start showing
it to real estate agents
Caleb: Yeah.
I
Jonathan: and getting some feedback.
Caleb: I did a deploy, some changes
and then I think we're good.
Jonathan: Sweet.
Do we have a way of actually
letting them sign into.
Because I noticed on the landing
page, there's no login button.
There's no,
Caleb: Oh,
Jonathan: there's no like
Caleb: you have to,
Jonathan: You have to guess.
Caleb: yeah, you have to guess the URL.
Okay.
Maybe a, maybe a sign up button
is the one, one thing left to do.
Jonathan: Um, we don't
necessarily have to do that yet.
We don't have to have that to launch
because what I want to do is have
people, um, get into the newsletter too.
We've added, uh, uh, a
MailChimp newsletter.
I'm going to try and.
Try and craft some content that will
entice people into the newsletter
that will create value for just
being part of the newsletter.
And then as part of that,
we can show how, um,
how using OpenHouse.Social would
be beneficial to them.
But ultimately, like I want the
newsletter to be valuable on its own.
I haven't totally figured out what that's
going to be, but I've got some ideas.
Caleb: Yeah, I guess we have the
newsletter and then also the, the free
version of the form and the landing page.
And then at some point, cause,
cause you can sign up for
account if you can find that URL.
So it's just actually adding a
link to, it would be the next step,
Jonathan: Well that's The
Robot Factory.
Thanks for listening.
See you next week.
Caleb: Bye bye.
Jonathan: Our target demographic is moms.
Hey moms, do you have a son in tech and
kind of want to know what they're doing?
Well, if they're not going
to share it to you, we will.
We'll tell you what we're doing
and we will be your adoptive son.
Caleb: it solves the issue when people
are like, what do you do for work?
And you're like, ah, I don't
really want to explain it.
Jonathan: Just subscribe podcast
Caleb: Uh, huh?