Design Meets Business

Christian: Avi, it's a
pleasure to have you on.

You bring a wealth of experience and a
portfolio that most designers and design

leaders would be jealous of having
worked for Samsung and Iwoca and Shopify.

And you are now at Deal, a
company that I love very much.

I have so many questions,
probably not enough time.

So let's dive headfirst into this.

Tell us a little bit about you, give
us a high level intro into your career.

Avi: I'm Avi, as you mentioned.

I'll take it back to about.

15 years ago when I moved to
London to do a master's degree

at Goldsmiths University.

at that point of time, halfway
through, I found myself working

in design agency called strite.

We did really interesting things.

we made a festival called Music Tech Fest,
where people connect music hackers, music

startups, record labels, all in one place.

We did a bunch of installations.

this was when I was still practically
designing, in the world of interaction

design, doing code and design and
the intersection between them.

From there, slightly after I moved to
Samsung, working on televisions audio

systems, I made the Lumi, the design
systems that kind of started with

televisions, then moved into fridges.

Their ties and watches as
well, which was really fun.

I think this was the first time Samsung
has ever heard about the design system.

and kind of like about the time where
Google came up with material design.

From there I moved to do a startup, in the
hospitality industry called Tupac, which

essentially lets you do Tet with people.

because if you think about it
let's say you booked as a four to a

restaurant, and now the restaurant is
preparing for you for eight o'clock.

We've all been to the situation
when we came to a restaurant, we're

like, Hey, can I have a table?

And they're like, no, it's fully booked.

And you look inside and you
see four tables that are

available and you're like, why?

Why can't I come?

And so Tupac looked at, at this, from the
different scenarios of like, you booked as

a four, but you met a friend and now you
arrive as a five, or like people cascade

or like suddenly people are sick and you
come as a three and like, how do you.

Do all that stuff.

And essentially like the receptionist,
usually 20-year-old boy or girl, would

do the math and would decide which
tables to plug in with each other.

However, they're not a mathematician and
they're not great at algorithms, right?

So it happens to be that
it's very similar to Tetris.

When you have a hole at the beginning,
you basically screwed the game.

And so, what we've done was.

Being able to make those optimizations
and make restaurants make 20% more money.

And so we sold it after two and
a half years, which was great

because COVID was about to come.

And if we were there in business, when
COVID came, boom, it would all collapse.

I moved back then to ICA and because
of my startup experience had.

For the first time, this opportunity
to work in this hybrid between,

head of design and head of product,
was managing product managers, but

also designers, started the team
from scratch, built it to 30 people.

it was a really, really lovely, lovely
time, over there at Taka when we were

going from like series C, to series D and,
did a full rebrand, hired researchers.

Product designers, brand designers,
and kind of covered all of that.

And then expanded, as
I mentioned to product.

From there, I moved to Shopify where I
worked primarily on internationalization,

especially in the first couple of
years, helping merchants go from local

to global across all the different
things that they need to deal with.

So if you think about it, when
you sell locally, you need to ship

packages, you need to advertise
on Google, Instagram, et cetera.

You need to do taxes.

Now, multiply that with the world
and suddenly you need to do warehouse

fulfillment and languages and
different themes because in one

place it's winter, so you wanna show
winter stuff in other place it's

summer, you wanna show summer stuff.

you have like the different
domains, different taxes you need

to deal with currency exchange and
variety of other complications.

So to make that simple was
a really big challenge.

And we essentially moved as a team.

From a setting inside settings into
the main navigation of Shopify, which

was an enormous, effort from everyone.

And yeah, it was a great time in the
company and then I moved to work on.

Two startups that Shopify acquired and
helped merge them into the ecosystem.

One of them is called Collective,
and it allows you, as a merchant

to sell through other merchants.

If you think of a scenario, if you go
to the to the street, any store that

you're physically seeing is selling.

A bunch of products that they
themselves don't manufacture.

It's all coming from different places.

Whereas online, it's a little bit more
difficult to make those connections

because if I have like a baby store,
I need diapers from one place and

drawers and chest from a different place
and clothes from a different place,

And it all needs to be synced.

So every time that the, prom company
makes new version, it needs to

sync automatically to my website.

Be able to handle things like mixed cart,
like bringing stuff from different places,

but makes it feel for you as the customer,
as if it's coming from one place.

And the other one was called collab,
which allows merchants to connect with

content creators slash influencers.

So if I have like products that I like
and I see someone on Instagram who's

really successful in that category, I
can contact them, maybe send them a gift.

And they can essentially sell
my stuff for a commission.

So every, like, that's why you hear
all the time on YouTube, Hey, use, my

discount link , et cetera, et cetera.

And, that also included an app for the,
creators themselves where they can look

for interesting products that they want
to sell, brands that they believe in.

And then only recently I moved
to deal, and I think that.

For a deal, especially when I
was interviewing, I had probably

like the perfect experience, for
that because deal is all about

working remote across the world.

It's all about being a strategic
partner for companies across the work

os So anything that is related to
work from hiring to immigration, to

paying people in different locations
and doing the taxes on behalf of the

company, essentially like being your
offshore company for anything you want.

Also growing talent and management, as
well as looking at the org chart and

do all sorts of projecting of what my
workforce is going to look like, which

country should I hire for, et cetera

. Christian: Thank you.

A couple of threads I
want to pull on there.

we'll go a bit back, to Woca.

You mentioned about this hybrid role
where you were sort of head of design,

but also hiring product managers and, um.

That's interesting because
usually it's the other way around.

It is very rare that you see product
report into design, let's say.

Most of the time it's, design
reporting into product.

So it would be cool to talk about that.

How did you find that
experience, of managing?

Obviously designers will get
to talk about that a bit later,

about managing product people.

Avi: I remember when I was interviewed
slightly after, and I was told,

how, how did you solve these?

Disagreement between product and design.

And to me it felt very natural
because I was kind of the decision

maker of the both of those teams.

So like there was never, I, I
made sure that the designers are

always aligned with the product
managers throughout the process.

And I think that was one of the key
things that were different when we

were writing, goals, when we were
trying to scope specific products.

We use the method called ShapeUp, which
is, created by Ryan Singer from PACE Camp.

And the idea of ShapeUp
is very similar to life.

they have like, one of the best
assets that they have is this

visualization of the hill chart.

And the Hillel chart basically means
you're climbing over a hill until

you get to the top of the hill.

You do not see how you're
going to get off that hill.

Also, you're not seeing if you're
on the tallest hill as well.

So you might climb onto the hill and
then realize, oh, actually now from

this new vantage point, I can see
that I have way more to go forward.

And in the world of product
management and even design, this is

the the period of time where you are
dealing with these unknown unknowns.

You're trying to figure out what is this
project, what is the problem, what are

all the facets of the problem so you
can find the best solution I know that

historically, like designers were the
ones that have more time to go and, and

climb and explore those things in advance.

They used to always have that also
exothermic ness, which is basically

like, an internal burning feeling of
wanting to change stuff, of waking up

critical of things and being unsatisfied
with life, not in a way that makes

you depressed, but in a way that
you're like, this could be way better.

I can see where this thing is heading.

And I think there's also like
a little glimmer of that,

you are addicted to change.

And being addicted to change is
essentially being addicted to growth.

Change is growth and change elevates.

And so to do that heel chart together
as just as designers is problematic

because that creates this always
perpetual like design is ahead of

everything and in reality, design
doesn't need to be ahead of everything.

In reality, design needs to be
together with product, it needs

to be together with engineering.

And we all need to climb that
hill and see the vantage point

and see the technical limitations.

We don't need to come up with like
a full design and be like you guys

from engineering, figure it out.

Sometimes it works, but
sometimes it's better to just

go ahead and do that together.

And that's something that I took from
Myka, and we did at Shopify as well.

Engineering and product people and
data people prototyped at the same

time as designers, and we all.

Learned way more and were able
to kind of achieve much faster.

Christian: I think that saying
of design, having to be a bit

ahead of engineering, at least.

Usually happens because engineers don't
like waiting around and they kind of

always wanna have something to pick up.

If there's a, a day that's a bit
slower, there's something that design

has already done that they can pick up.

How did that work in an environment where
everyone's kind of on the same, not on

the same level, but rather sort of working
on the same problem at the same time.

Did you find that engineering
was always a bit blocked

Avi: It's a matter of the quality
of the engineers that you have.

And usually, of course you have
way more engineers than designers

and you probably have like more
designers than product people.

And I think that's a decent ratio.

But if you have too many engineers or
you are shipping low quality products,

you end up in a situation where
engineers are always firefighting.

they are bug fixing.

They don't have time to
think of what's ahead.

Now, I'm not saying that all the
engineers, by the way, need to do that.

It could be that you are just making it
effect with the, at least at Shopify,

we used to do that with the engineering
leads, the product leads and the design

leads, and they would be the ones going
into the forest trying to understand,

like, and climbing that hill, trying
to understand where we're heading.

Sometimes with one engineer who is
responsible for that area or not.

It wasn't always like everyone,
because different people have different

roles and people have different
vantage points, but it was really,

really useful, to be able to do that
and to socialize those insights.

And then you're aligned from the get-go.

You don't have those issues of like,
oh, you asked me for this and now

I won't be able to do this, or let
me kind of go and explore and come

back to you in a couple of days.

This thing is happening together
and it's shaping each other.

Now sometimes I agree that it's
useful to completely disconnect.

Work completely detached.

And that is when you're doing high
design, that's when you're designing

way further into the future.

But anything that is within the realms
of the next six to six weeks to like

two months should be done altogether.

Christian: Yeah.

What I found useful in the past is,
and this works really well, especially

when you have a specific kind of
engineer, I think these days we

call it product engineers, which is
engineers That care a bit more about

the product, not just engineers.

We're there to necessarily just deliver,
uh, which there's nothing wrong with

, they can be useful as well at times.

But I have found that whenever you
work with product engineers, they also

want to be involved a bit earlier.

And what often happens because you have
this small overlap or you bring them into

a workshop or even just grabbing them to
jam really quickly on a solution, is that.

the whole idea of handover kind of
disappears I remember when Envision,

appeared and became really big it allowed
designers who were designing in Photoshop

to do a bit of a handover to engineers.

Today, I find that that's disappearing
a little bit when you work with product

engineers who do want to get involved
a bit earlier, and oftentimes they can

already start on things before you even
finish the design because they're already

up to speed with what you're doing.

So I guess that's what you mean in the
sense of overlapping a little bit, right?

Avi: Yeah.

that's what I'm saying.

Christian: So usually when, when design.

Reports into product.

What, I have found in the past is
that you lose a little bit of that

sense that craft matters, obviously
this depends massively on who in

product you actually report to, but.

I have had both.

sometimes when you report to a product
person who perhaps doesn't have

a background in design or doesn't
care too much about it, you end up

becoming a little bit of a feature
factory or an experimentation factory.

Just these are the numbers we need to hit.

Like everything is very
surgical in that sense.

Now let's look at the
other situation, which.

Product reports into design.

Did you find that you were trying a little
bit to instill a sense of craft or a sense

of, you know, that the experience matters?

, A a way of thinking about not just
what metric we need to reach, but also

how does a person feel using this and
all of that, um, into the product.

People that reported to
you and how did that go?

Avi: I think that when you look at design
and how design and product are working

together, it's important to mention
that there is an org chart, right?

And a lot of companies end
up shipping their org chart.

For example, let's imagine
that you are structured with

product teams across a funnel.

You are now at the stage where you
have something that was working 70%

good, it was fully built, and now
you want to get to crunch it to 90%.

You are tightening all the
screws, et cetera, et cetera.

So you'd have, let's say.

An onboarding team.

And then after the onboarding team,
we would have like whatever specific

product team and then, whatever end
case, scenario that you were gonna have.

And then each one of those teams
would try to squeeze things and

without knowing often that would
make, an impact on the other teams.

What do I mean by that?

Think of, let's say meta.

So now I have like an ads team,
and I have an ads placement team,

and I have an engagement team.

Now, mark comes to me and says, Hey,
meta doesn't make, enough money.

We're not making enough money out of ads.

So the ads team is like, okay,
I'm just gonna show more ads.

Awesome.

So now they're showing more ads.

they're making more revenue.

However, after two months, users
are tired of seeing more ads.

So at that point of time, users.

Using the product less
because there's too many ads.

And at that point of time, the
timeline team comes and they're

like, guys, we have too many ads.

We need to prioritize dms and friends
so we can increase engagement.

They're doing that.

Revenue goes down from the other one.

So this is like a metric and counter
metric, however those metric goes.

Um, in a, in a, in a square.

What does that mean?

We need to grow.

Right?

We can't just like keep
this duopoly of like.

ads, more ads.

We're losing money now.

We need to basically make
a completely new area.

So now we're bringing in reels.

it's a new area.

The engagement is going that way.

More people are in
interacting with the platform.

And at that point of time, because
it's new, there are no ads there.

So I found the right balance in the
timeline, and now I'm introducing like

this new area where people interact with.

And then after a period of time.

We wanna monetize that.

So now we're bringing ads into reels, and
then now there's too many ads into reels.

And now we're like, oh shit,
we ran out of advertising space

and we want to keep on growing.

What are we doing?

Ah, we're gonna bring ads to
WhatsApp, et cetera, et cetera.

And that journey kind of continues.

And the way that journey continues is by
all those companies creating all those

new products so they could have more ads,
real estate, so they could advertise more.

So on.

And so the same thing happens,
quite often with other products.

And how do you counter that?

So for example, in the world of
deal, the company is acquiring 2, 3,

4 products a year at that point of
time, you get into scenarios which

each and every one of those product
teams want to make their own era.

They want to have their
own navigational item.

they want to have their own banners
and communications and notifications

now to, you need to unify that.

So there is this pendulum between we
are growing, we're making new things,

and now we, and we are working in silo
of improving each one of those things.

And the other side of the pendulum
is we want to create like this

seamless experience end to end where
people don't feel that I moved from

one product to another product.

We're utilizing the design system.

To stitch all of this
together, we are redoing ux.

We're thinking from a service
design perspective, we're

thinking from a business design
and upselling point of view.

And so when product reports to design,
you get this opportunity to have

this to kind of move this pendulum.

Whereas when design reports to product
quite often you don't have that pendulum.

It's always about crunching and
tightening and improving, which.

Results usually in those
very bloated, bulky products.

Let's say like, you know, Atlassian, JIRA,
and Confluence and Trello, for example,

which has like, you know, think of the
amount of overlap that all of them have.

It's crazy.

Um, and so.

When design is in, is in charge,
or alternatively when there is a

balance, you have this opportunity to
recognize the stage of the company.

Are we now in growth mode and like
making lots of stuff and creating new

products and that's okay to just get
dispersed for a little bit and we just

need to focus on these new initiatives
and growing and tightening, or are

we at the stage where we wanna make.

this whole thing unified, we need like
a layer of AI that sits on top of it.

We need the design system to
speed up the engineering team and

everyone else who is not a designer
who wants to work on these areas.

And you have this ability to kind of
go between those two to help designers.

Christian: This reminds me of, a
couple of years ago when Brian Chess

Airbnb, or maybe, no, maybe this is
many years ago when they did that.

Um, storyboard, I think inspired by
Disney, where they storyboarded the

ideal experience of a guest from the
moment they are researching until

the moment they leave the Airbnb and.

When I was reading that, I thought
this would never happen in a

company where design is not at
the foundation of it somehow.

And obviously Airbnb is co-founded by
a designer, but I have never seen the

approach of, let's look at people's
experience at a high level and then

draw learnings and hypothesis from here.

Most of the time is what
you said, which is because.

Design reports into product or product
has, even if it's not like that product

generally has a bit more weight.

We tend to start from metrics.

We tend to start from what we need.

we need to grow.

every company needs to grow.

Obviously, I, I don't know of any
company that wants to not grow, that

is a given, but perhaps there is
something else which you start from.

and that could be the ideal experience
a customer might have or something else.

I just found it really
interesting that you said,

Product reporting into design.

I don't think that happens very often.

So, yeah.

you also mentioned Shape Up.

I want to talk about Shape Up.

I'm a big fan of, of the work
that, base Camp and, Ryan

Singer and everyone else there.

Um, well, he's not there anymore, but,
uh, everyone else at at Base Camp does.

For someone who doesn't know what
Shape Up is, can you give us just a

one minute and then we can talk about
how that applies to our workplaces?

how have you found using it
and all of that good stuff?

Avi: I would say ShapeUp is a different
kind of agile methodology that allows

you, especially usually in six weeks
cycles to deliver very quick improvements.

It's all about.

Understanding that there are two phases.

The phase where you don't know what
you're doing or what you're building,

and the phase where you do know
and you need to choose a solution.

And it's all about, giving the
power of making those decisions

of where we're heading to each
and every one of those people.

So you have this really good.

Dissection between engineering,
product design, data.

All of them are climbing the hill
at the same time, ideally, and

then seeing the best pathway.

Discuss also which hill
they want to go into next.

And then take the slope.

either it's the shortest slope
that might hurt a little bit.

You might get injured and you might
have like a little bit of tech debt

throughout that period of time,
or it's a slower slope and you're

taking the time, but it's okay.

You already know where
the next hill is at.

Christian: So why is this better
than a typical agile or why was

be better is very relative, right?

Why?

Why was it better for your teams?

Why did you think that it
was worth trying it out?

Avi: I think what made it better for
us because it was every six weeks.

I think the duration really helped
us engage with the founders and

like at, when I moved to Shopify,
we also did it every six weeks.

We used to meet the CEO
every six weeks as well.

It's just a period of time that
allows you to ship bigger things.

Not just focus on plastering at the end as
a product manager, or as a designer, when

you go into a new project, It's almost
like going into like a water container.

And you heard that there is somewhere
leaking, but you're kind of like

in there with your eyes closed,
trying to feel the wall and trying

to see where there is a hole.

And wherever there's a small hole,
you basically can make a decision.

If you wanna put a posture.

Or you, you wanna
completely fix the thing.

the duration of time allows
you to do more big fixes.

and it's not too big of a rock that now
we're going into this like crazy project

of planning for six months and it's
not, I'm just gonna solve this thing

temporarily, but I'm acquiring, tech
debt or design debt, throughout that.

So I think it's just.

The period of time that really makes it
powerful, that allows you to ship big

things whilst fixing small things with
the aim of knowing what the strategy is at

and, and getting to the really big things.

It just, it speeds up your process and
it allows everyone who's not in your

department, who's not necessarily r and
d to be fully involved in that as well,

Christian: so you're talking
about six weeks, you're talking

about the period and how important
the, this specific period is.

How is it different?

Let's say someone from product
is listening and they're, they're

sold on this idea, or from design.

How is that different than, than
having three month cadence, right?

The three month, OKRs, which
are very typical these days.

How is it different from
saying we, we still do agile.

We still do what we've done until
now, but we reduce from three months

to, to six weeks, like you said.

Is it just a matter of the period or
there's something else inside of ShapeUp

that makes it such a good methodology?

Avi: I think what makes it a really good
methodology is , the acknowledgement

of these unknown unknowns, right?

Like there is this period of
exploration and there is this period

of execution You have enough time
to do both of them within the course?

I can tell you different companies
work differently at deal.

We have like one week sprint,
literally one week we kind of like

every week we just come on a Monday,
we're like, these are the amount

of stuff that we're going to ship.

Boom, let's go.

we are quite speed junkies.

At deal, there's a lot
of dopamine involved.

We quite often would have scenarios where
we're gonna have like a client call with

CEO or the COO and they would come and
say, okay guys, everyone, we just talked

to each client, I need your opinion.

And everyone would open their account, try
to utilize, log all their issues, and then

we'll spin up a channel, bring all the
right people into a place and say, okay,

within a week this thing is resolved.

Christian: How is that working for design?

Because I assume in a one week cadence,
it just doesn't give you a lot of time

to go deep, gives you a lot of time
to solve the problem, but can you go

deep enough to solve the problem Well.

Avi: Sometimes you can,
sometimes you cannot.

I think what we're trying to do
in those scenarios we try to make

sure that the things that are like
easily resolved, like within a week.

Will be resolved, within a week, and
we are trying to take out all the

things that we believe that are bigger,
lift into our planning, et cetera.

Christian: So you would take
things that are bigger lifts and

perhaps split them in one week.

Sprints.

Right?

it is not like you only solve problems
that can be solved within a week.

You might still solve the bigger ones.

You would just have to split
them, in different sprints.

Avi: exactly.

Christian: Okay.

And that also gives design time.

So perhaps this week for design, they
might just do research or they might

just dive deeper into the problem.

Maybe by the end of the
week there's nothing.

to deliver while engineering works
on some other things, but then there

might be more to deliver next week.

So it's, it's not always as, as
structured as every week something

needs to be delivered or is it,

Avi: it is.

I don't think that you,
you have a scenario where.

Not at the, like, honestly
not at Shopify either, where a

designer is like dedicating their
time only to one task, right?

Because yes, even if you have like
lots of research, like at the end,

there's a lot of things that weren't
possible in the past to do with

research that are possible today, right

? you just need to have
the conversation, right?

Like you can, you can use stuff like gen
AI and not even have the conversation.

But let's say you are
doing the conversation You

don't need to write notes.

You just throw everything into
notebook, lamb, notebook lamb,

kind of like summarize all that
stuff and you're ready to go.

And that thing gets
amalgamated over time as well.

I think that every designer gets the
chance to kind of like walk across the

whole spectrum of their T shape or W
shape, whatever you want to call it.

I love w because.

feel it's spiky in both directions, right?

It's quite stable.

It's probably more stable than a t.

and you have the skills that you're expert
of and the other skills that you're not,

but that's okay because you have a double
sided, base, just metaphorically saying.

Christian: Yeah, I think T-shaped
skills are, have been a concept that's

pretty well known from the past.

It's been discussed for a long time.

If someone doesn't know what
a W shaped skill designer is,

give us a 32nd page for it.

Avi: Yeah, I would just say that
generally what we want people

to be is spiky object, right?

If you're in a river and the
water is kind of like coming and

the, the river is life, right?

then you're going to be very smooth
and therefore not opinionated, not good

at any one thing, and pretty mediocre.

And I know another thing
that is pretty mediocre.

It's called ai.

And so like we don't need people who are
mediocre from that perspective because

AI basically facilitates for that.

And because of that you
need To be a spiky object.

What, where I guess like, you can
be an X, you can be a W, it doesn't

really matter, what shape it is.

You just need to have
like a very strong base.

You need to be stable and
you need to be really, really

good at some specific things.

And I think, you know, metaphorically,
w or an X, or a Zed, reflect that.

Christian: Fair.

what I'm picking from that is have
an opinion, have a good, strong

foundation so that water doesn't
flow around you that easily.

I like that.

you mentioned AI mediocrity.

Maybe this is a good segue
into talking about ai.

let's keep it a bit practical.

Have a think about how we're using it
today or how you're using it today.

Maybe at work, maybe not at work.

give us a little bit of a high
level overview of how it's

helping you in your day-to-day.

Avi: So AI is great at helping
you streamline the task

that you don't like doing.

Generally, AI cannot
replace the whole spectrum.

There is no tool that can take the
whole product creation from one side

to the other, but you do have a bunch
of tools and you can do tool stitching

across the bar from all the surfaces and
all the areas that you're working on.

So let's start from research
when you want to discover.

So from desk research, we all know we can
ask perplexity, whoever understand stuff,

when we actually want to conduct research.

We can utilize Janeway AI to interview
on our behalf with a prototype

that we've created and the list of
questions that we've created and

basically freestyle with the people.

And of course, you can take it all
the way to the other side where

you can utilize synthetic personas.

And synthetic personas essentially
are clusters that are made of,

research that you already have done.

Different pieces of feedback
that you got, through support

calls, et cetera, et cetera.

And basically they simulate the
customer so you can share a design

or ask a question to your customer,
and then based on all the feedback

that the system knows, until that
point of time you'd get a response.

And that is something that
we've utilized at Shopify.

There is something that we are
also utilizing at Deal as well.

There's a lot of data that
comes from everywhere.

From sales calls, from support calls,
from emails, from Slack channels

with partners, and you have startups
also in the world that do these type

of things themselves, like clean.

It's a really powerful, interesting
startup that basically connects all the

support system into something that you
can ask that can even go and create PRDs.

You have bagel AI as well.

They also do these type of things.

Then we're getting into
the creation process, so.

The actual product design and brand
design in the world of brand, you can

utilize Vy and Whisk as an example.

The reason why I like those tools
is because they allow you to combine

multiple inputs into one input, and
they also give you this ability to

practically edit, not by just like.

Smearing over something or circling
something and get it removed or changed

to something else like you have in
Photoshop Google photos or apple photos.

This is more sophisticated.

I wanna take a bird and I'm gonna
take a specific building and I want

to take a specific scene and plugging
them all together and I can move them

around to generate the best ai, image
that I want to get to the brand.

Now, there's a lot of things that we're
talking about at DEAL nowadays, like

for example, being an AI first brand,
and that basically means the historical

type of illustrations that we had.

If you look at it way back,
it was trying to be classical.

But when you look at it, it is
something that the only things that

are similar to it are relatively cheap
illustrations that you have, like in

iStock photos or things along those lines.

And then when you go to AI and you want to
generate that, that's what you're getting.

And in reality, you can get way more
if you think of what AI has been

trained on, video games, be hands.

Deviant art.

If you go into those realms, you have
way more interesting surfaces and the

ability to generate and create things
that are much more quality is much higher.

So that's a transition that we're
kind of like going through at the

moment in the world of product design.

If you are working in a world like
Shopify, we have lots of product images

and you want to enable, you know,
dress ups and like, how would this

thing look like in my home with ar.

Or how would this outfit look like me?

You have a lot of startups also that
interacts with that, but Shopify

also dabbles that, especially
in the shop app and a deal.

there's a lot of things that just
like helps you speed through things.

Like I make a table with a bunch of,
data that I need to prefill and I can

plug it into real data and, and import
that, or alternatively, generate a semi.

Non PI data based on
real data that I have.

So all those things are things that
the design system are also looking at.

The design system is not just the
component and puzzle pieces that you

construct with, but also like how you
utilize those things for generation.

At the end, the design system
would be the thing that would

enable not designers really.

It would be the thing that would enable.

Product people and engineers to
make designs, to prototype stuff

and to just work more efficiently.

And what is required from
designers would be to break it.

And that's why you need designers.

You hear more and more nowadays of
this renaissance of taste and craft,

and that is because that is the only
difference between you today and ai.

And so if you think of the essence in the
beginning of design, this is it like art,

design, visual fidelity being different.

And over time I think that like
universities and even courses started

doing these, like methods, double time
and, all these, all these type of like.

Design thinking, follow this
process and you will get to

an amazing result at the end.

This is all now can be done by ai.

What we're looking for now is the
differentiator and since AI can do all

these processes, because it learned,
because it's all over the internet and

everyone did it again and again and again.

Probably not very good by the way.

Now we're looking for the,
the one differentiator.

That is the thing that actually.

Makes a difference, and that
is the essence of design.

And so design becomes harder,

Christian: so if you're saying AI can
take all over all of these repetitive

tasks that have been done a million times
and really they're more of a process.

So, um, because it's repetitive,
AI can do it probably as well

as us and definitely faster.

You say taste is now what's gonna
differentiate designers from ai?

How do you build that taste?

Because taste is so subjective.

How do you know if you have taste?

What is craft?

'cause to me, something might be
ugly, that to you is beautiful

or the other way around.

Or we might look at the same app and
I might think it's a terrible app and

you might think it's a really good app.

So how do we marry that idea of
taste matters now when it's not

really an objective thing that
you can learn or can you learn it?

Avi: I think you can learn it to
some extent, but if you look down the

street and, you know, you don't see
as many people like, unless you're

in Paris, you not see as many people
getting dressed like extremely well.

Many people just like throw stuff,
whatever comes into the cupboard

on them and then just walk out.

They don't do their hair,
they don't do whatever.

To learn taste, you need to be curious
very much like, you know, to learn

design, like you needed to be curious.

you needed to read all the books,
you needed to know the artist you

needed to practice and try the
different things and not necessarily

just and play design system stuff.

People that are designing design systems,
the second you give them something that

is out of the design system, they would
try to make it inside the design system

and then they would fail and again, and
again and again, and they won't be able

to basically break through that paradigm.

that's a way of getting stuck that
it took us a little bit longer to

differentiate and identify because
it's harder to get stuck because

you have all those components.

hypothetically.

The number of options that you can
export out of, all the components

that you have is much bigger.

Potentially you can get to like any
design, but in reality you cannot.

There's always this limit that at
that point of time you need to make

something new, and that's the necessity.

Now, the ability of doing something
new and the only way to do

something new is to have a bigger
repository of existing things.

Not only the things that are
belong to your company, but

things that belong to other.

That's why like websites, let's say
like mobbing, that lets you see like

all the spectrum of, of patterns
across, all the apps are very useful

and that's why a lot of designers are
utilizing that because suddenly you're

not limited to just your design system.

Now you're limited to all design
systems and then vice further.

You have people that are more taste makers
who are able to make those things from

scratch, who are able to think outside.

Basically create those things.

Historically, wherever I worked, I really
enjoyed, the most working on operating

systems deal, the operating system for the
workforce, Shopify, the operating system

for commerce, Samsung, the operating
system for phones and televisions.

so when you work in that level,
you essentially work ahead of

the actual software companies.

Figma needed to do liquid glass.

Apple did liquid glass, not with
Figma or Sketch or anything like that.

Someone created it physically, then
someone mimicked it digitally with

code, and now everyone needs to fall
in line for that new innovation,

for this new taste that is coming.

Airbnb is doing that.

Apple are doing that.

Samsung did that.

Shopify is trying to do that.

And that is kind of, again, not all
designers need to do that, but the

better AI gets, the more that will be
the requirement of designers to do.

You need to be able to
have a really rich world.

You need to be super curious and you
need to utilize that world in order

to make new things because usable
things would be way easier to do.

That's why we have the systems.

if you look at like why people are
buying Tesla, why people love linear,

why people love Faye Finance, it's
because the design of it, the taste

of it just makes you feel good.

You feel like you're
using a premium product.

You feel like you're using
something that people care about.

And you have a lot of
people that would love that.

Christian: Something I've been thinking
about when it comes to training ai,

obviously the a all LLMs are being
trained on everything that's public,

on the internet, and I've been thinking
about the counter argument to this,

which is a counter argument to, to
what a lot of people are saying, which

is taste will matter more, you know?

You could probably make the argument that
you can train AI on Dribble and mobbing

and AI will have maybe not necessarily
able to create new things, but will

have a full understanding of a lot of
different products or potential products.

And Dribble is not products, a lot
of just, there's a lot of creative

ideas and then you might go and
say, I need to make a screen.

I need this screen that we have.

Help me, because now I wanna
capture some other type of data.

And then the ai, the LLM or whatever
tool it may be, will go and look through

Dribble and MO in, and all these other
places where it's being trained on.

And we'll be able to give you a
pattern that is aligned to your

design system, but you don't have it
in the design system, but it's, you

know, it's aligned with the right
colors and right tokens and all that.

And suddenly you've created a new
pattern in the design system by simply

looking at what others are doing.

But there's still no human
involved in that sense.

Or obviously the human is prompting
and perhaps can do a bit of

QA at the end to see does this
really fit what we really need?

But I've been thinking a lot about
what the future of this is, and I

don't see why a world in which AI
can also do visual cannot exist with

a world in which AI can also do, the
research and the brand and all of that.

So.

My question off the back of that is, if
that were to be true that AI can also

be trained to do visual work, where
does craft and taste come in then?

In that case?

Avi: Curation at the end, the AI
would sometimes cross wires and

would make something unintentionally
new and exciting, but that's just

a thread that, you know, you as
a designer would want to pull on.

At the end, you are the decision
maker, at least for now, as long

as you're the decision maker,
it needs to make sense for you.

That's why, you see, the Johnny
Ives, the Brian Chesky, those are

people who take a lot of input from
everyone, like CEOs generally, right?

they take a lot of input from
everyone, and they make a decision.

That's kind of like the companies that
have chief design officers as well.

These are the places where they basically
said, we trust this one person's taste,

and this one person is going to weave
all the ideas that are coming from

across the different teams and create
a cohesive story that tells the same.

Narrative across multiple surfaces.

All the kind of like, you know, it's,
it's kind of like all the Apple products.

The curvature and the radiuses
are all gonna be the same,

like the AirPod and the thing.

You can put them on top of each
other and it's gonna look the same.

There is a decision, it's a limitation.

And that limitation is caused
by a piece of opinion that

that is the thing that matters.

Setting up those rules.

And so for ai.

We need to be able to set those rules
and pull those threads because AI

won't be able to, replicate something
necessarily just because you said it.

You need to be the one who are
prompting and evolving that.

Christian: I got one more question
on taste, and then I'd love to

talk about the design tasks.

I know it's something that you have,
some strong opinion on, so it would be

good to, draw those out a little bit.

something I've been thinking about
and I've been finding myself in that

position sometimes as well, and I'm
sure there are a lot of other designers

who do so as well, where you look at
a design, you came up with something.

Again, from that perspective of
craft and taste, you look at it and

you think, this is just chef's kiss.

This is great.

This will kill, this is just
the way it's meant to be.

Then you put it live in an experiment
and it bombs , and there's this

conflict, internal conflict between
I feel that this is the right thing

to do but the numbers are showing
something different and that tension.

You need to deal with
that tension somehow.

And what's the right answer.

Just go with the numbers,
because numbers don't lie.

Although they do sometimes.

There's a great book, uh, called
how to Lie with Statistics . so

yes, numbers can lie.

But anyway, let's say in this
context, numbers don't lie.

And then the other one is, I just
feel inside that this is the right

thing to do I just know it's the
right thing to do for the user.

how do you resolve that tension?

Avi: So story time.

in the past we had a design
system change at Shopify.

We moved from, I don't
remember, like Polaris 1.0

to 2.0,

or from 2.0

to 3.0.

It was quite radical.

Everything was much smaller.

The whole thesis was Shopify
is a work surface and people

needed to be more of a workforce.

It needs to be less spacious, right?

Like a Bloomberg terminal.

it's not useful for you to be spacious.

You need to see everything.

If you work in Palantir and need to manage
a battlefield, people will get used to

however, you're gonna map this thing out,
and if you change it, an airplane will

get bumped because you changed the thing.

So.

The, that was the opinion.

Shopify needs to be of a more of a working
tool and considering the fact that 80

something percent, 83, let's say percent
of the surfaces were on the design system.

The idea and the thought was, we are
just gonna come and we're gonna flip this

switch and everything's gonna be fine.

Because 83 is, kind of like
merged with design system.

There's like a deep link.

It's gonna be, everything's
gonna work great.

It turned out that all the new
surfaces, new things, new areas,

when we put the beta flag, it all
pumped and either, didn't work or

like completely ruined the design.

And in addition to that, to
your point about numbers,

there was a lot of onboarding.

Things that we've done.

we used to have, like in settings,
we used to have two columns.

One column is more kind of like
explanations and tips and whatever.

And the, and the main column,
we moved into one column.

What do you do with all of that?

Suddenly it's really hard to set it up.

Suddenly they're at the bottom below
the fault people don't even see them.

we were approaching additions, which
is essentially every six months.

Key date where there's like a big
advertising and we tell about all

the new features and we had a choice.

Either we stop building everything and
actually go and fix the design system,

make sure that when we flip the switch
it's okay, or we push it to the next one.

And the decision was we're
going ahead with it, right?

Because we truly believed.

This is the future of the tool that
it was okay for stats to tank for a

short period of time, and then we'll
fix it later with the new system.

But we didn't have enough time
to do this whole thing with

the new system from the get go.

We just needed to make sure the
new system is implemented well

visually and doesn't look broken.

And so that was the decision Toby, the CEO
was like, we're pushing ahead with this.

All the front end developers just
stopped everything that they've

done and everything that they worked
on for the next like month and a

half was like implementing this.

All the designers redesigned all the
surfaces to fit for that and then we

shipped properly and metrics tanked,
then we made adjustments based on

that, and now it's better than it was.

So sometimes you need to take a
short term hit to get a long term

gain if you really believe in it.

Christian: And I think the key
there is that Toby pushed for it.

Toby was okay with it on a daily basis.

A designer doesn't have Toby's
power in a company to say, yep, I'm

okay with, with metrics tanking.

We're a public company.

It's okay.

It's on me.

we'll solve it.

A designer just oftentimes sits in
this, triangle, with product and

engineering and has to, for lack of
a better word, convince his or her.

Partners that this is the right
thing to do long term at the

expense of the metrics tanking.

Short term, I think that's a much more
difficult position to be in when you

don't have the support from higher up.

So I dunno, maybe there's
no solution to that.

Maybe there is some sort of stakeholder
management to be done there, but

I do think it's a bit different
than when you have the CEO of the

company telling you, this is okay.

Go

Avi: Well, the only way, the only way
to convince for the long term is to show

what the long term is going to look like.

The designer does not have time
to design for the long term if you

don't do low design and high design.

low design is like I'm designing
for the next six weeks, shape

up, et cetera, et cetera.

and even in low, lower design, actually,
it's a little bit more than that, right?

Because like to really solve
a problem, you need to look

at the problem holistically.

It's not about like, oh,
this button is incorrect.

I'm fixing it.

That's not low design.

That's literally being a technician and
like problem solving and bug fixing.

From a design perspective, when I
say there's a problem with this flow.

I redesigned the whole flow.

It takes six weeks, or it
takes like, three months.

Both of them are okay.

It is still short term.

A quarter is short term.

High design is designing really, really
far into the future, like three years

into the future, really understanding
where we want to head, and that is

also a thing that keeps on changing.

It's not something that like stays still.

Because every change that you're making
today would make an impact on where

you're gonna get to in the future.

And both of them should be maintained,
probably not at a 50 50 ratio,

but 80 20 ratio I think should be
fine to be able to maintain it.

when I say 80 20, it's on an individual
designer, and some designers would be

the ones that would spend more time
on high design versus low design.

Christian: I think that's the
answer I was trying to get out.

Just show the vision, show where you want
to go with it and show why this is a, Look

towards what you're trying to build later,
rather than this is gonna work because

I think it's the right thing to do.

So that's what I was trying to draw out.

Let's talk about design tasks.

we don't have a lot of time left,
so I'd like to, understand you've

made a strong case before for design
tasks during hiring high level.

Why do you think there's such a good
idea to use during your, hiring process?

Avi: I think the main issue that people
take with design tasks is that why do

I need to work for that company Why
do I need to in advance, do some work?

moreover, if I'm interviewing
with multiple companies,

which usually is the case,

It just becomes unsustainable, because
you need to spend like seven hours

on each task that's like takes over
your whole week and makes your life

at your current work even harder.

people have kids, et cetera, et cetera.

And I would say in, in addition to that,
you don't, at least people don't feel, I

feel you do have a partner to work with.

So I think, you know, historically
when we used to give tasks like this.

I would actually rank higher people
who would throughout doing the task,

email me and be like, Hey, I have this
two, three questions about the task.

I have started doing this, but I want
to understand more what you guys want.

But most people are kind
of like shy about it.

They either do it in the last
minute or they just let it go.

And come with whatever
they did themselves.

And I think, that was kind of like
the, the prehistoric task, like

the task that you need to come in
advance, send it, and maybe you get an

interview to even come and present it.

We've moved away and I have also moved
away from that type of task because

I understand it is not fair and in
reality I want to have the session with.

The people, I want to
actually like, see them live.

But the, the other problem that happened
was at, at that point of time, people

were like, pay people to do that.

Pay people to do that
is a little bit hard.

Like it's, it's just hard
technically as well, by the way.

It's not just the company doesn't
want to spend money, it's just hard.

Technically they're not your employee.

It's out outside of the, it's harder
also for the person who needs to get paid

to some extent, the middle ground that
happened was you give the task in advance.

You send the task like a week
in advance to the person.

You're like, Hey, you're
coming to a session.

This is what we're going to talk about.

That is to prevent surprising the person
and having them to kind of just like

on the spot make up stuff, which is not
fair as well because some people just

don't function well on the spot and in
reality at work, you're not gonna have

to function on the spot, like someone's
gonna give the thing, you're gonna have

some time to think about it, and then
a week after or two weeks after, you're

actually going to start working on that.

This is what we're trying to mimic
right now, where we're giving

the test like a week in advance.

You don't need to physically do anything.

You don't need to come with
anything to the actual interview,

but you know what it's about.

So you can simmer it in your head, you
can think about it, and then by the time

you come, it's not a total surprise.

And then you work with the people.

Like, I wouldn't mind someone,
when I work with people would

be like, Avi, you go do this.

can you get me this?

Can you.

Activate The team.

That's fine.

Like the people are here
in the interview with you.

They're not just like here to kind
of cross arms and talk to you and

grill you and just watch you do stuff.

This is a working session.

How would you work in a working
session with other designers?

Now, there is one constraint over that,
which is you need to get to some sort of

output within the hour, and I would argue.

That people should do that in
every meeting that they have.

You should not end the
meeting without key decisions.

You should not add a meeting about design
without an actual improvement . If I

presented a prototype, for example,
and someone gave me some comments

as they gimme the comments, I'm
prototyping it into the prototype.

I don't need to wait for the end of the
session and then do this because I want

to show it to them in case it's not right.

They can tell you, ah, yeah,
actually, you know what?

I was wrong, or it's not right.

Let's do something else.

the faster we can go, the faster we
can get to the end of the result.

So that's essentially the only
constraint that you do have.

You have a time and you need
to produce an artifact and

the end of this piece of time.

But in reality with AI and with,
you know, Figma and Figma make.

You can easily, produce that.

Christian: So there are two things here.

One of them is the older
types of tasks that you talk.

About where it's like, here's your
task, we send it to you, you have

to work on it in your own time.

You maybe come back and, interview
for two, three different companies.

I think that is the type of design
task that's maybe a bit older

is still very prevalent today.

A lot of companies are still doing it.

what you are talking about.

I wouldn't call that a design task To me
that's more like a whiteboarding exercise.

Avi: call it coworking session.

Christian: I think that
is a very different.

Way of interviewing and I find
that to be a much better gauge.

of the candidate . If we're talking
about the coworking session, like you

called it, What are the main benefits
of doing it like this, and is this the

only thing that you prefer to do or are
there other sessions as well that kind

of fill some of the gaps that you don't
get out of these, coworking sessions?

Avi: What am I getting out of the
coworking session is I can see

how someone manage their time.

what can they produce Within 60 minutes,
I can see their technical skills and

how they work and how comfortable they
feel working with the others as well.

So you can see communication, you
can see like AI fluency, you can see

technical capabilities Figma wise.

You can see, their ability to guesstimate
and utilize AI to kind of validate

their idea or invalidate their idea.

I can see how much they're
looking at competitors and all

of that within a time constraint.

and that's essentially, you know,
a great reflection of what they're

going to do on a day to day.

I think that companies that do,
take home tasks have a lot of

candidates and it's like a mechanism
for them to filter the ones that.

Want to really work for them
or not really work for them.

But I believe that the coworking
session also does that because

you have a week in advance, right?

let's imagine kind of like the task is
like designing some something physical

with like, you know, number of buttons.

You, you can research, you can like,
do a bunch of stuff in advance.

You can even, you know, take an hour
and try to do the exercise as if.

You were like in the session
and trained for it, right?

you can do that to see
how you structure it.

Or you can just write for yourself,
oh, I'm just gonna validate for

five minutes and then I'm gonna
prototype for this amount of minutes

and then I'm gonna do that and that.

And so you can sense that,
but there is less emphasis

of how much do you want this?

And more emphasis of how.

You work.

And I think that the design task
of the past were more anchored.

It was a different market.

The market was very much
like an employer market.

I would argue that today it's an employer
market, but it's like a more woke

employer market, than it was before.

Right?

So like, people understood, and I
think like Gen Z also, they, they know

that and no workplace is promised.

They know that people get
fired, right, left and center.

They don't let people don't stay.

Very long in places.

So why should they commit to something?

And I, you know, I respect that.

Well, like, why should they commit
to someone who's not committing

to them in the past, you know,
my dad like still works in the

same company for 36 years, right.

It's different, it's different like
the types of treats that you need.

Like, you know, being able to be
in harmony, like knowing people for

so for 30 years, like being able
to deal with them like, you need

to forgive so much stuff today.

It's very abrupt.

It's very quick and companies
don't invest enough, in people.

You know, there's not enough training,
which I hope will change over time,

but the tasks and the application
should reflect that as well and

should reflect the state of mind.

And I think coworking session does.

Christian: So practical tips.

Stay on time, learn from.

Manage your time if you
do one of these tasks.

When you say technical skills,
so you said communication, AI

fluency, your capabilities in Figma.

I mean, obviously if your
capabilities in Figma are not good,

there's very little you can do
on the spot to fix that I guess.

But, some of these other things,
communications, so you said probably.

Bring people in, use it as a coworking
session rather than you being on

stage and everyone watching you.

what other small things would you give as
practical tips to do in these sessions,

other than what we've mentioned already?

Avi: I would say you can hack
your way through stuff, right?

Like there are generic, you can prepare
for yourself, like generic design systems,

so you can plug in and play, like, you
don't need to create like a dropdown

now dropdown menu from scratch, right?

There's a bunch of stuff that
you can prepare on the side that

you can like pull from, or like,

you know, have your toolbox
set up for you on the side.

And that's another thing that
is being measured, right?

Like, all these things are tools.

how much do you use the hammer?

How much do you use the screwdriver?

How much do you use the drill?

All those things are the
things that are being measured.

Because historically I would say
education moved into this world of doing

everything in a specific structured way.

It doesn't work like that in reality.

Not every time you have time to research,
not every time you do double diamond

or empathy map or whatnot, or personas.

Sometimes you do, sometimes you don't.

Sometimes you do it like once
a year and then you use it

for throughout the whole year.

You need to be able to be
flexible and versatile.

Christian: Cool.

We're getting to the end.

We don't have a lot of time
left and um, I ask him.

Everyone the same two
questions at the end.

one of them is where do you look
for inspiration in your day to day?

Avi: I love podcasts.

I love books.

I love X and Threads.

I think specifically in X and Threads like
I under, before I joined Shopify, I under.

Valued those platforms for design
inspiration and for great creators

and builders that just like ship
on a daily basis and show you

cool stuff that they're doing.

And, yeah, I completely am now in these
platforms to, to understand that I take

a lot of my knowledge from economics.

So like, I love Professor Scott Galloway.

I love re um, like business,
uh, business analysis.

I love Econ Talk, which is a
brilliant guy, who talks to

people about books that they read.

there's loads, in the world of books,
I would say, we have multipliers.

It's a great book for anyone
who's a design leader.

It basically.

Teaches you to do more with less.

I would say it's like pretty, prevalent
for what we're experiencing today and

kind of this age of ai, how gradually over
time you can actually do more with less

and for the fun and to just fill me up.

Music has always been the thing.

Christian: Lovely.

And, um, we have touched a little bit
upon this, so maybe the answer has been

given already, but I'll ask it again.

What's something that you believe that
AI will not be good at and therefore

that designers should double down on?

Avi: I think at the end AI will be good
at everything, but I think at the end

it's about us to be opinionated and we
are the ones who can weave story, like

we all see like amazing videos from VO
three and things that can create, Streams

that are extremely funny and makes
billions of watches on TikTok, and some

people make stuff that get zero views
and get nothing, and it's the same tool.

So it's all about us.

It's all about our opinion, our
ability to connect the narrative,

and our ability also to market.

It's a, it's a very important thing.

You don't just do design, you
do design, and then you sell it.

And that's 50% of the job.

Christian: I've heard this thing once
that, regardless of what job you're

in, your main job is sell sales.

So, uh, sort of the same.

Avi, this has been really great.

Is there anything else you want to add
at the end before we bring this one home?

Avi: Thank you so much for listening.

I hope it was useful.

Feel free to contact me on
LinkedIn and we are hiring so.

Christian: Yeah.

we'll put all of this in the show
notes so people can easily find you

as well as some of the tools that
you've mentioned and everything else.

Uh, from this episode.

Avi has been a massive pleasure.

Thank you very much for
being on this amis business.

Avi: Thank you.