2:00

leaders, mastering customer research,

2:02

AI for product managers, and more.

2:04

There's something for everybody. Courses

2:07

started just $299, and you can

2:09

get access for your entire team with

2:11

a discounted team license. Just

2:14

go to gigantic.is for

2:16

more information. Again, gigantic.is.

2:20

And level up your product team

2:22

today, because we believe that your

2:24

potential is gigantic. OK,

2:27

I've got to admit, when a company like Shopify

2:29

reaches out and says, hey, Mike, we were curious

2:31

if you were interested in getting an early look

2:34

on our upcoming product launch and talk to our

2:36

VP of product about it all for rocketship.fm. They're

2:39

going to get a quick yes. I

2:41

mean, as a product person, I'm always

2:43

interested in seeing all their companies, especially

2:45

super successful ones like Shopify, how

2:48

they manage major product launches. So getting

2:50

a peek behind the curtain, honestly, it's

2:52

one of the reasons why I love

2:54

doing what I'm doing with rocket ship.

2:57

And Shopify has a pretty interesting take

2:59

on product launches as it is. And

3:01

if you go to shopify.com/editions, you'll see

3:03

how they introduce their new products and

3:05

features to customers. They do it in

3:07

seasons with a winter release and a

3:10

summer release. And every release

3:12

has its own theme, its own branding.

3:14

It's packaged up almost like a product in itself.

3:17

I mean, it looks like a list of apps

3:19

when you're on the website almost,

3:21

right? I mean, it's like their own

3:23

little app store, except these are their

3:25

product launches. You click and you're taken

3:27

to a pretty rich web experience that

3:29

goes into great detail on what's included

3:31

in that season's edition. In

3:34

some cases, it's feature updates. In

3:36

other cases, there are

3:38

entirely new products that Shopify customers

3:40

are treated to. Now, this season's

3:42

theme, well, probably shouldn't get too

3:44

far ahead of myself quite yet.

3:47

First, I should introduce you to

3:49

Glenn Coates. Glenn is the VP

3:51

of product at Shopify. He

3:54

leads the development of Shopify's

3:56

core commerce platform, which is

3:58

the storefront checkout. going

18:00

on, right? With AI being so

18:02

early for everyone, I asked

18:04

Glenn what it was like to

18:06

incorporate those AI features into Shopify's

18:08

products. Is it straightforward? Is it

18:10

overly complex? Back to Glenn. Everyone

18:13

is going to go through this thing

18:16

over the next year or two of

18:18

building their first AI enabled features if

18:20

and I'm sure most people have already

18:23

taken stabs at this. I think the

18:25

interesting thing is like the best practices on

18:27

how to build with LLMs and how to

18:29

train models are moving so, so, so quickly.

18:34

That one, you

18:37

have this interesting experience of it's

18:40

really easy to get to quite

18:42

compelling prototypes quite quickly. Like you can get

18:45

to the thing that smells like, oh, it

18:47

kind of it's doing the right thing. It's

18:49

smart. It's cool. You can get to

18:51

that quite quickly. But depending on your

18:54

on your risk tolerance for the for the

18:56

model doing the wrong thing, getting

18:58

from something that's kind of 70% prototype quality

19:01

to 90, 95% production quality is surprisingly difficult.

19:07

And it's very unlike the curve

19:09

that you experience in like more

19:11

straightforward traditional software development to like to

19:13

get to 70% and then you get to 90%. Everyone kind

19:15

of knows what that feels like. That

19:17

part of the curve is like way,

19:19

way, way more brutal in

19:22

in this part of development, which I'm

19:24

sure many people in the audience have already experienced.

19:28

But I think the thing that we are, it's

19:32

hard to keep in mind, but we're trying to

19:34

keep in mind here is that because

19:39

the state of the art is moving so

19:41

quickly, right, because the models are improving so

19:43

quickly, because the techniques of working with the

19:45

models are improving so quickly. It's

19:48

actually a little bit like, you

19:51

have to have this mindset of like, you remember like video

19:53

games when we were younger, where it was like a video

19:55

game would come out like like when doom or quake came

19:57

out. And it's like the

19:59

computer The computer you had at the time Quake

20:01

came out couldn't really run Quake.

20:04

Kind of sucked. It ran at like 10

20:06

frames a second, right? But then you

20:08

would buy a new computer and suddenly Quake's

20:10

amazing. It's running at like 50 frames a

20:12

second, right? And the

20:15

people who made that game, like Carmack and everyone,

20:17

they knew that. They're like, we know that no

20:19

one can run this game, but we know the

20:21

next generation of graphics cards

20:23

and computers is coming, so let's build

20:25

for that. And there's

20:27

a little bit of that going on right now with

20:31

AI, which most web, like web developers

20:33

aren't used to thinking this way. We're

20:35

not like, oh, let's build a

20:38

web app that is terrible

20:40

today, but somehow, web

20:43

browsers will get better in the next six months

20:45

and it will suddenly become really good. That's not

20:47

how it's worked, but we're having to

20:49

do that little, you know,

20:51

skate to where the puck is going thing a

20:53

bit right now, and it's uncomfortable. It

20:55

feels weird. You're like, this

20:58

kind of sucks, but I think it's gonna

21:00

get better soon, you know? Yeah,

21:02

I can 100% relate to that

21:05

feeling, that uncomfortable feeling, where you

21:07

know you're doing the right thing,

21:09

but it just feels hard, and

21:11

you're not quite sure if what you're doing now

21:13

is actually going to make an impact later. That's

21:16

skating to where the puck is going.

21:18

Now, here's the thing. If you've never

21:20

experienced this before, if nothing's felt uncomfortable

21:22

to you, if things always just sort

21:24

of feel easy, you

21:26

might not be doing it right. At

21:28

least, let me put it another way.

21:30

You're probably not making leaps and bounds

21:33

and truly innovating. These early days of

21:35

AI, even at Shopify, it could be

21:37

uncomfortable. That only leads me to

21:39

believe though, that they're taking the big swings, and

21:41

who knows? I mean, they could miss on some

21:43

of these big swings for sure, but if

21:46

they connect, they could be hitting some home

21:48

runs. All right, enough with

21:50

the sports ball references here. I'm

21:52

gonna move on, but I did ask Glenn

21:55

more about that uncomfortable feeling. Get more to

21:57

share on that right here. We're doing this

21:59

for ladies. to

30:00

run their business on

30:02

Shopify and have the vast majority of their

30:04

business data in Shopify. And Shopify is kind

30:06

of the hub that all the

30:08

apps connect to. Again,

30:11

there's this very interesting question of, okay,

30:15

well, where in the stack do

30:17

you integrate the AI? How

30:20

does that AI join the dots between

30:23

the different apps

30:25

and features of the platform? And

30:27

how do you make it in your

30:30

dev platform? How do you make it possible

30:32

for the apps to expose enough data

30:35

and capabilities to each other so that

30:37

that AI that's sort of sitting in

30:39

the center can pick

30:43

the right tool for the job and even join

30:45

multiple tools together to complete a

30:48

multi-step job? And

30:51

I think that's the interesting part that as

30:53

we move down the road with Sidekick and

30:55

as we sort of invest in that part

30:58

of the platform, I think

31:00

there's just a lot of headroom

31:02

there that remains to be

31:04

explored. We're almost out of time here,

31:06

but I did ask Glenn to wrap

31:08

up our conversation with final thoughts for

31:10

product people and product teams. What

31:13

does this all mean for product people? What

31:16

should product people be thinking about as they

31:18

move forward? Here's Glenn's final

31:20

parting thoughts for product people. We

31:23

are really, really,

31:25

really doubling down on the value of

31:28

having a tightly integrated product. And I

31:30

think this is an easy thing for

31:32

product managers to sort

31:35

of forget to dot their I's and cross their T's

31:37

on. It's easy to

31:39

think about shipping the new feature. People

31:42

tend to forget to budget for the,

31:44

how do I make sure this new

31:47

thing works super well with

31:49

all the old things? That's the kind of

31:51

unsexy part of the job, but

31:53

it matters so much to users. And

31:56

so if there's

31:58

one thing for product managers, to take away

32:00

that lesson. And then

32:03

I guess I would also say

32:05

like this is an interesting time for software

32:07

teams to experiment and

32:09

learn with this new tech

32:12

of AI and LLMs. Nobody really has all the

32:16

answers yet. We're still figuring stuff out all

32:18

the time and we think

32:20

we are sometimes we're we can't get

32:22

something to work and we're really frustrated

32:24

and we think oh maybe we're behind

32:26

here and then we go and talk to you

32:29

know our partners at open AI or Google

32:31

or you know anthropic or

32:33

whatever and you know they're like oh no you

32:35

guys are like right

32:37

on the cutting edge like no one's doing the stuff you're doing

32:40

and we're like oh wow okay well I guess

32:42

I guess everyone's figuring this out. So I guess

32:44

if there's another message it's like be

32:47

a part of figuring it out because this is

32:49

a great time to sort

32:51

of you know learn

32:53

the nuances and the weirdnesses

32:56

of these new tools and

32:58

and how to and how to incorporate

33:00

them. All right that's going to wrap

33:02

things up for today. I just want

33:04

to give a big thank you to

33:07

Glen Coates and the team at Shopify

33:09

for pulling the curtain back for us

33:11

and giving a glimpse into Shopify's recent

33:13

product launch. You could check out all

33:15

the details on their launch at shopify.com/additions

33:17

but this will wrap things up here

33:19

for now but for Product

33:21

Collective I'm Mike Bellsido and this

33:23

is RocketChip.fm.