Ep#6: Figma Config, Source of Truth, and Craft in AI Adoption

Tom and Jonny share updates from Tom’s Mostly Working meetup in London with Kate Pincot (Multiverse), focused on scaling design impact through tooling and agents, and preview upcoming podcast guests in July.

They reflect on how the meetup format has evolved as more people have hands-on AI experience, shifting from first experiments to practical, scaled adoption inside teams, with more focus on token costs and measurable value.

They discuss Figma Config releases (Motion, Figma Weave, Sites/Make, and upcoming code components), debating canvas-first vs code-first workflows and the “source of truth” question for enterprise teams.

Jonny summarizes a Near Future Works post on AI transformation: set a coherent strategy, secure org-wide buy-in, and maintain craft/quality so AI speed doesn’t produce worse outcomes. They also touch on music ownership, a Claude Code World Cup calendar prompt, Anthropic “Fable” returning and Sonnet 5, Z.ai’s open-weights GLM model, a keyboard-shortcuts tool, and an open online workshop (Aug 13) to build design tools.

00:00 Welcome and agenda
00:12 Mostly Working recap
00:52 Podcast updates and banter
03:01 Meetup lessons so far
07:13 Figma Config reactions
12:39 Source of truth debate
22:51 Jagged frontier and craft
28:11 Quality Bar Cannot Drop
28:47 Barbell Strategy for AI
30:18 Craft Beyond Output
32:37 Atrophy and Writing Muscle
35:47 Craft Across Disciplines
36:40 Music Ownership Stories
41:22 Claude Calendar World Cup
42:57 AI News Frontier Models
45:27 Beautiful Shortcuts Tool
46:57 Open Workshop Invite
49:48 Signup Feedback and Wrap

Links:
Mostly Working: https://luma.com/mostly-working
Near Future blog post: https://nearfuture.works/updates/a-new-jagged-frontier-why-craft-still-matters
Figma's config keynote: https://config.figma.com/san-francisco/session/829e6ced-3257-4f5c-b675-aa72f4d1f98f/
GLM5.2 by z.ai: https://z.ai/
https://shrtcts.click/
Our next workshop: https://nearfuture.works/build

As always, feedback welcome at [e-mail]

Find us at nearfuture.works

Creators and Guests

Host

Jonny Burch

Co-founder of Near Future

Host

Tom Harman

Co-founder of Near Future

What is Near Future Podcast?

A regular podcast covering design and AI from the founders of Near Future, a boutique AI consultancy focused on teams that care about craft. We cover both what we're seeing on the ground and industry trends, ways of working and occasional guests from the design world.

Jonny: Hi, welcome to another
episode of the Near Future

Podcast with Tom and Jonny.

We've got a lot on the docket this week,
so we're gonna get straight into it.

first of all, Tom, you ran
another Mostly Working event

with Kate Pincot from Multiverse.

Wonder if you wanna tell us a
little bit about that upfront?

Tom: yeah.

It was, super fun.

Ran that last week, from the
Gradient Labs office in London,

and we had a great crowd.

Thanks to anyone who's listening
to this who came along.

was a great speaker.

gave some really interesting insights
into how Multiverse are thinking

about the opportunity for, design
to scale their approach and impact

through tooling and agents to
scale the, their brand is applied.

It was very cool.

Jonny: Nice

Tom: also just got a call out at the
start, Jonny, that you, the last time

we recorded there was a reversing truck,
and this time it sounds like there's

Jonny: There's a power saw?

Yeah, I'm gonna turn-

Tom: saw, angle grinder.

I think it really enhances
the ambiance though,

Jonny: Makes it feel real,
We're, th-this is the real world.

We're not, we don't exist in a vacuum.

We're not AI

Tom: this

Jonny: from space.

Tom: Yes.

Not, at least not yet.

Jonny: We're not Synthesia
models and ElevenLabs voices.

Tom: It

although I wonder if there's
an Eleven Lan- Labs reversing

truck that we can add on to our

Jonny: Probably.

Tom: in the future.

Jonny: Yeah.

Tom: anyway, Mostly Working
last week was great with Kate.

it was, really fascinating to hear
the approach there, and, just great

to host it in a new space and,
be surrounded by great people.

the other thing to just share up
front is that we're gearing up to

welcome our first podcast guest next
week, or we will be recording with

our first guest next week, and we
have guests lined up throughout July.

really excited for this next chapter
of exploring with this podcast.

So if you're listening to this and
have any feedback that would help us

not make a fool of ourselves when we
invite real important people on to

speak with us, please let us know.

Jonny: Awesome.

I also have a bit of news up front,
Tom, which is that I'm standing up.

Tom: yeah

Jonny: I ran one podcast in a prior
life, in a prior company, and podcasts

then you could just do audio, and you
could be lounging around or staring

into space or picking your nose,
and it would be absolutely fine.

And now you have to do video, and I
feel like I'm getting found out here.

I watched-- or I edited the last
episode, and I was like, "What

are you doing with yourself?"

So my focus is now on standing
still, looking at the camera.

And so if anyone's listening to this and
not watching it, it was still not gonna

be great, so I wouldn't tune in quite yet.

But maybe in a year's
time I'll be good at this.

but yeah, the, sunglasses last
week was an interesting option.

I feel like, this week may be a
little bit better, but we'll see.

I, just wanted to ask you before we go
into the, main bulk of the, podcast,

you just mentioned Mostly Working,
but you've had some reflections

on Mostly Working as a format.

Running a meetup is hard, working out
how to make it work in person and remote.

Just, for those people that may have
been to Mostly Working, which may be

some of our listeners at the moment,
just, yeah, like what you're learning

so far I think is interesting for
those people thinking about, "I'd

love to, to run a meetup or an event."

Tom: Yeah, of course.

So we've done five now, and they've
been roughly every month since January.

I think the meta reflection is that
when we started in January, the, state

of the world was that people were
aware of AI but hadn't really spent a

ton of time building with these tools.

And six months later, is very different.

Most people have been trying, exploring,
playing with these things a little bit.

there's still a gap between
playing with something and actually

getting real value from it.

But the need to have a dedicated space
to get hands-on and build things,

to have evolved from just trying
something through to actually getting--

doing something useful with it.

which I think is, as an
interesting reflection of

where we are with AI right now.

And I think

Jonny: Oh, seamless transition.

Tom: yeah,

Jonny: that was so seamless

Tom: it, would be if I carried on,
but there's another thing I wanted to

mention, which is that, I think that
this, the, the-- So when we started

running these events, it was a really
wide spread of what people were interested

in hearing about from opening the
terminal for the first time and starting

to use Claude Code for the first time.

And we had various designers coming
through who did open Claude Code for

the first time or Cursor or whatever
the tool was, in those sessions.

Jonny: Yeah

Tom: they came back a month later
and said, "Wow, I've got so far.

I built so much stuff.

I have now have product that I've shared
with my internal team and they're all

using it and working off of it now," or,
"I'm starting to speak with VCs about

raising money for this tool that I've l-
built for the first time a month ago."

so it's very inspiring to, to, hear that.

But I think the, level of conversation
has shifted at this point towards

designers generally being a bit more
curious about actual applied ways of

working inside companies at scale.

So less one-off experiments or people
starting from scratch in their own

design system constraint legacy
product free world through to how

do we actually make this work inside
our team of five or 10 or 20 or 100.

And, and I think there's a level
of there that means that some,

some ways companies are
approaching this are very unique

to how that company operates.

Jonny: Yeah

Tom: if your team is five people,
it might be a bit different because

of the scale and size and the kind
of challenges you need to tackle.

Jonny: Yeah

Tom: to get a, glimpse into like how
real people are working with these

tools in real settings, not just the
sort of sandboxed, I'm gonna prototype

something on my own version of that.

Jonny: I think that we are past the,
cool demo in my bedroom, local host,

isn't this, an interesting technology
and let's see how far it can go stage.

And companies are looking at their token
costs, but also teams are saying: all

of this promise, what's it turning into?

Is it turning into actual value?

And, so now this is good though,
because we've moved out of this

existential fear and, oh my God,
maybe it can do everything, maybe

we're cooked, to this is a tool.

We can't work out how to use it
well together yet, so let's do that.

And I think, I haven't been to-- I
wasn't at this latest Mostly Working,

but the conversation then matures into,
practically how do we shape this into

something that I can take back to my
team and everyone will be like, "Great,

this is genuinely useful," rather
than, "That's a cool side project.

Let's get back to drawing
rectangles," or whatever it is.

Tom: Exactly.

So with that perfect transition, let's
talk about config, Figma Config, 'cause

that happened at the end of last week,
and, Figma, a tool historically known

for being great at drawing rectangles,
now exploring a bunch of new things.

they released a bunch of new features
from motion to the code layer to, various

other, smorgasbord of functionality
that they're, hinting at and releasing.

an interesting set of things
for us to be exploring.

Jonny and I have Weren't there in
person, so we can't speak to the

sort of what the vibe was like in
the room, but we've had some thoughts

and reflections following it.

Jonny, do you want to k- kick
us off with a conversation

around config and what's next

Jonny: Sure

Tom: in the tooling space?

Jonny: Yeah, hot takes from across
the Atlantic coming your way.

yeah, what's-- So people who've read my
blog in the past and seen what I've been

writing could take me as a Figma hater.

That is not true, although I've always
felt that Figma and the tools that

predate Figma, Sketch and, I, drew
my first UI in Photoshop probably.

But they don't-- They're, a, bad
facsimile of the real thing, and at

best, they can achieve a percentage,
not much more than 50% of the fidelity

of something that is actually built in
the technology that it will end up being

in, whether that's in Swift or whether
that's in the web or whatever it is.

And I have also personally found as a
designer that has coded for a long time,

that the sort of rote work involved in
drawing rectangles and, Figma have now

released slots, which for people that are
in this world will know what that means.

But, a big time saver, but it's amazing
that it took them so long to do that.

That feels like the actual biggest
release of the year, in many ways.

but it's like really time-consuming
to get to something that in code you

could get to very quickly, even before
AI, because I could have a blob of data

and just repeat that data to create
a table and et cetera, et cetera.

Anyway, so I'm not a Figma hater.

It's come from a place that
made complete sense, I think,

up until coding models appeared.

And they are very bound to their
architecture and their way of

working and the workflows that they
have been selling for, many years.

And I think that what in the features
they're releasing now is effectively

trying to protect the golden goose,
which is what they currently have,

while also reaching into some of the
other aspects of the design process

that they have no-- haven't been
able to service, up until this point.

So they've gone in both directions.

On one side, which is the more
creative, like image making and

what could this be and, th-that like
earliest stage of the design process.

They've built Motion, which is
their sort of After Effects,

Lottie type tool for motion stuff.

and they bought a company, Weavy,
which they've turned into Figma Weave,

which is like asset generation with AI.

Both interesting things and feel like a
good fit probably if you think of Figma

as a tool where you start doing your
work and imagine what it could look like.

and then they've built in the other
direction fairly consistently.

They've built Figma Sites and,
and, Figma Make, which was

their sort of vibe coding thing.

That was last year, and now they
have built these code components

within Figma, where you can drop
code blocks into your Figma canvas.

And haven't tried it because it's
not available yet So we can only

go by what they've announced.

But it sort of-- it does slightly feel
like forcing code into the canvas, which

still feels like the wrong way around.

if the source of truth is code, why
am I putting the source of truth into

something that isn't the source of
truth just to push it back again?

It feels of course, Figma-- that's
how Figma should do it because they

need to keep people in their canvas.

But it doesn't feel like if you were
to design something from scratch,

that's how you would do it today.

So that's a long way of saying it feels
like a bit of a mess and a bit of like

grabbing all the sort of bits of the, the
design process that they haven't touched

yet and trying to do a bit of an Adobe
on it and scoop up all of those existing

enterprise contracts and make sure that
they don't wander off because they have

something good enough in the existing
r-relationship they have with Figma.

I don't know if that sounds
really negative or…

But I think, what-- the one thing I
did come out with was none of this

feels like the thing that solidifies
Figma's place in the design universe.

it's good for people that
want to stick to Figma.

I think a lot of these features
will be underbaked for a while yet.

So if you want best in class at,
for example, designing in code, you

probably wouldn't use Figma for that.

If you want to design in code a bit
while you mainly stay in your canvas,

it makes sense to have-- to, use Figma.

But as they start to move into these
things that they're historically less

good at and less geared up to be good
at, I think it weakens their position

generally in the market, would be my, bet.

and I'm talking my book because I
wrote an article in February about

this, and I still believe it to
be true, that the s- the source

of truth is the ultimate question.

Like, where does this all
ladder up to or down to?

Where are the,

the core components, quite literally
or figuratively, of the design process,

where do they live and where's the
truth that sits underneath all of those,

whether that's a design system or the
actual code base or things like that.

and that probably points the direction
to where the future is, I posit.

and what that means, I think, for a
lot of other tools that have sprung

up, and I've built a website to capture
a, big list of them, designtools.fyi.

I'll plug it again.

but, I think a lot of those tools
will look at this and say, "We

probably expected Figma to, do this."

And some of those have made a bet,
which is one part of what Figma's now

shipped, and probably they'd be worried.

But those tools that are building on code
as source of truth, I think can feel like

this doesn't invalidate what we're doing.

This is what Figma should be doing
for them, but we can do something

different, and there'll be enough
people in the world that will want

that, that we can feel confident in it.

so mainly choice is still alive.

You can still use Figma, and that's
great, and for a lot of teams,

that's the right move right now.

but it doesn't remove any of the other
exciting, often open source or really

interesting projects that are happening
to explore new ways of designing and more

sort of modern ways of designing as well.

So I talked for a little
while, but there we go.

yeah, what, do you think, Tom?

Tom: I th-- Y- yeah, I th-- just to
play back some of the sort of themes I'm

hearing from you, The design process is,
I think for a while, Figma has owned the

entire design process as a function, as a
discipline, and that process is starting

to be pulled in two different directions.

There is the upstream exploratory,
creative asset generation, things that

would inform a source of truth, but
aren't necessarily the source of truth.

They're more the for and exploration.

And then there is this other
end of the process, which is

implementation, execution.

Like how do we get this
into people's real hands?

And yeah, Figma are pushing
both directions simultaneously.

But that second one, I think the
first one there, they are well-suited

to, but I think that's a smaller
piece of the overall design puzzle.

that handoff question, I think, is really
the really juicy, interesting one, and

I think they're coming at it from this
sort of canvas first into code versus

Jonny: Yeah

Tom: other tools that are
code first into the canvas.

I feel like some version of a
canvas will need to exist, but

I think the sort of ownership of that
is less clear than it historically was.

for me, I think there's also these extra
aspects of as we enter this era of,

codifying strong design judgment into
design system or whatever the code base

is for the source of truth, I think that's
moving away from just and UI elements

into how do we make judgment decisions
about what our bar for quality is.

Jonny: 嗯

Tom: more about design.md

or other of, and defining our
bar for what craft is that into

the way we ship products or the
way we make design decisions.

And so curious what form that's gonna
take because it doesn't feel like it

naturally lives on a canvas in the same
way that, sort of level design has.

But there is some amount of
workflow logic in what is our bar,

how do we hold everyone to it?

And not just for designers.

Maybe designers are the people who
are best placed to help and set that

bar, but could imagine that being
used by engineers or PMs who are

making changes to the design of the
experience they're responsible for

without directly with the designer.

But, there needs to be some sort of
opinionated position on where is our bar

for quality and our bar for craft, and
how do we, scale that more broadly across

the organization without being, without
individual designers being a bottleneck.

Unless those individual designers
are-- are really important features for

an individual designer to be pushing
against and really handcrafting.

That's a different

Jonny: Yeah.

Tom: But for at an organization level,
that probably isn't gonna be true

100% of the time, always forever.

Jonny: Yeah

Tom: anyway, that's perspective
and, some of that is what you've

already said, but just it, back.

But I think it does mean, if we're
thinking about our design tools

as, is this about execution and
implementation, or is this about ideation?

I think that really what the purpose
of that tool is and how we, use it.

And I think for anyone who's brand new to
design, perhaps who's listening to this

podcast for the first time or as a PM
listening to this, I think throughout my

career I have found that not always super
obvious that there are these two versions

of design, and think AI has collapsed
those now, where you can prompt an AI

tool like a Lovable or a Claude.design

into creating UI that looks
convincingly polished.

there's a separate conversation
about whether that's good or not.

But I think it does, think further
exacerbate the challenge of having these

two very different forms of design output
converging into, to make a, an assumption

of how good or bad this design is I'm five
minutes into the design process versus

Jonny: Right

Tom: which I think, I think is an
interesting challenge to overcome.

And design history, there's been
of the artifact determines the kind

of feedback you're gonna get on
it and how to shape and evolve it.

don't need to go down that path right
now, but I think it is this sort of

interesting, the illusion of getting
to a final artifact then having to

un- back out of it to like really
interrogate those decisions and work

forward is this interesting, feels
like a shortcut, but actually is

Jonny: Yeah

Tom: a much longer process that just
seems to start from this weird opposition.

Which is fine if you're using
it for ideation and testing

prototypes and ideas really quickly.

But if you're trying to get something
into production sooner, it's a

different way to, to think about that.

Anyway, those are my thoughts.

Jonny: M-

Tom: else you wanna mention?

Jonny: yeah, I think what I suppose
there's a bit of a call back to

a conversation we had a couple of
episodes ago about sovereign AI also,

which is, Figma is a US-based company
that you can imagine a world in

which there's some sort of hostile or
otherwise takeover by, a trillionaire,

unnamed trillionaire or someone.

and then sure, the design
process is maybe less, of a state

issue than access to models.

But you st- just having choice,
into how you do things and

having a solid alternative.

I think Affinity going free and really
shaking up Adobe was just fascinating.

And, Canva, Australian
founders and, less US beholden.

and that's, very much in the design
world as well, just seeing, different

models, different places you can go for
your design work and things like that.

It's just exciting.

So let's move on from config.

we've done enough Figma bashing.

and, it's not bashing, it's, a
healthy critique, I would say.

Tom: Can I actually add one more thing?

I don't know if I was clear when I was
talking about this, but the source of

truth aspect of this, and I've spoken with
a few startups recently who talk about

using AI to prototype really quickly.

then once they have some clarity on the
direction they're going in and got some

feedback from customers, then they will,
design it up in Figma from that prototype.

And so Figma has remained
that source of truth.

So I think the source of truth question
will be ultimately the, decider of these

big enterprises that have invested lots
of time and money and workflows into

making S- Figma the source of truth.

Jonny: Yeah

Tom: big thing to change into code, and
if it does change into code, I think it's

gonna be really cha- hard to change back.

So for me, that feels like the biggest,
factor in, what's gonna, influence where

the, where that source of truth lives.

And

Jonny: There's definitely a
design equivalent to, unscrew

up my V- vibe coded app, which
is, unscrew up my sloppy design.

And there is, yeah, for
a while Figma have…

You've been able to import, a snapshot
of HTML in, and it converts it to Figma

layers, and then you can fool around
with it and, make it look better.

as you say, how long does that last?

that could be Figma or it
could be somewhere else.

It could be an open s- like Pencil or
some open source tool that just builds

a good enough canvas that actually
makes sense to do it there, and it stays

in HTML, so it's more you're actually
changing the code rather than changing

a, an object that looks like the code.

yeah, r- really interesting.

I think these canvases and
the rectangles plays a role.

it really does.

it doesn't go away, and I think
you've been saying that as well.

it's just whether it is an ancillary
tool versus whether it is at the

center of the workflow, and I think
that is the more subtle change.

Nothing dies overnight, but that is the
more subtle change that may come about.

let's move on.

Tom: So Jo- uh, Jo- Jonny, you
wrote, I'm gonna interrupt you

there, but let's move on obviously.

you wrote some thoughts down, about
jagged frontier and, how craft sits

within that and some, lessons and
takeaways, reflections, over our

work in the space over the last year.

would love to hear any thoughts
you wanna share on that, maybe for

Jonny: Yeah, I f-

Tom: read your piece.

Jonny: Yeah, I finally broke
ground on our Near Future.Works

blog.

So far the blog or the, update
section was just the podcast, and

so now there is an actual written
article on the, blog as well.

so I mainly wanted to do that, but I
also wanted to encapsulate something

that we've been thinking about a
lot as we've been putting together

creds decks and positioning ourselves
against, various other o-outfits that

do outwardly similar things to us.

and I think one thing that we've
really learned over the last year is

that good transformation of a process,
whether it's a whole organization or

whether it's within one team, from
non-AI driven to AI driven, is not

as easy as you think it's gonna be.

It's not a case of someone sitting
in the corner and like bashing out

a new way of doing it, and then
everyone's "Cool, that works."

it takes-- It's just change management.

And anyone-- At my last
company was an HR tech.

I saw how complex it was to change process
from the inside, and we ended up having

to spend a lot of time with customers
in three-month rollout plans, g-doing

tours of thousands of employees to make
sure everyone understood what this tool

did, even the simplest version of it.

So change is always hard when
you think about a lot of people

changing at the same time.

and so I wanted to try and get it
into three points, 'cause three

is the magic number, and that's,
that's what trips off the tongue.

So three things to really think about
when you're like, "I want to change

my organization to be AI first, and I
want this transition to be successful

and not just something that we're tick
boxing for the board or some other reason

that isn't the best reason to do it.

I want in a year's time for us
to be industry leading in our

category, and I think AI is the
way that, that, we can do it.

And, but I want it to go well, and I
don't want it to break our culture, and

I don't want people to leave because
I'm pushing AI down their throats.

and I want it to result in actually
materially good outcomes that

I can point at and I can graph
rather than like good vibes and

everyone spending loads of tokens."

I tried to capture that.

I captured it quite quickly, and I
think that this is a first version.

I'm sure we will iterate over this,
and I think it captures both of our

thoughts, but I wrote it, and we haven't
had a lot of chance to talk about it.

But, so the three things are, a
coherent strategy I've written, which

is basically like we get together and
decide why we're doing this, which

is what I've j-just been describing.

what is the impact we wanna have?

Why, how much money do
we wanna spend on it?

How much do we wanna break
the culture to do it?

AKA how fast do we wanna move?

all of those things should
be made centrally and then

comms-ed out in the right way.

Maybe this is a little bit more
like CEO stuff rather than,

in the weeds for a designer.

and then the second one is hearts and
minds, what we, call hearts and minds,

which is- Is really just, it can't just be
one person vibe coding and pushing it out.

It has to be buy-in from everyone from
like IT and procurement who need to buy

a tool or give you access to a model, and
decide where the data lives and all of

that kind of boring but important stuff.

it needs to be the people that are
deciding the work you're doing so

that they can say, "Don't worry,
do a little bit less of that

work so you can spend time here."

So just like getting buy-in and
support from your boss and the exec.

and then everyone who's actually gonna
work on it, really just making sure that

they understand and are excited about the
outcome that they're trying to get to.

and it is a team sport.

This is a team sport.

You can't exact change on your own.

That's the second one.

The third one is the one that I think is

where we have the strongest opinions,
and I also think where people miss out.

Like the first two, fairly
obvious, have a strategy and then

make sure everyone's bought in.

duh.

But the third one, is Not assuming
that the things you build are

just automatically gonna be better
just because it's AI, and AI is

magic, and it seems to happen very
fast, and you built a cool demo.

I think where we've seen issues
happen is everyone's very

excited by the first version.

You show the stakeholders, they're like,
"Wow, c- can I have this yesterday?

Let's just turn it on."

And everyone gets excited about
it, and it-- what a win to get

something done so quickly as well.

you've done this in 24 hours,"
all of that kind of stuff.

and you get so swept up in it that you
d- forget to check whether the output is

any good, and this comes back to craft.

the craft is ultimately still
the human's responsibility.

I can't blame my tools
if something goes wrong.

And so even though it's been way more
efficient than if a human did it, if it's

even 5% worse, then that is a bad outcome.

It needs to…

The, bar cannot drop, and especially
if s- what you're turning on does 10

times the amount of work a- as a human,
things go wrong 10 times faster, and

that sort of, descent into slop, if you
like, starts to happen really quickly.

So that was what I was trying to
capture in the blog post, which I

will put in the show notes for sure.

but I think it's just, a vote for
craft and a vote for slowing down

a little bit at the right time.

People talk about this barbell
strategy around, AI generally, which

is where you decide what you need up
front, and you spend more time there.

And I think that's basically what
you were just talking about, Tom,

is like you, you take the time to,
to decide what, good looks like.

Maybe it's an art direction or maybe
it's just, a company direction, and

then you spend more time at the end.

And the middle is the we- the bulk of
the execution that happens quite quickly.

But then at the end, it's like more
QA, more time spent looking at it,

evals, all of that kind of stuff.

it's very tempting to just turn
on 50 agents and, fire your team

or whatever, the, a certain group
of leaders would want to do.

and w- I think we've seen the negative
side of that, and so I wanted to write

that as just as a p- as a piece to be
able to share with maybe potential clients

and just to establish our opinion against
the rest of the world a little bit.

That said, I do think we're now at the
phase in the AI adoption cycle where

a lot of people are starting to think,
actually what we built doesn't work very

well," or, "It wasn't the, the silver
bullet that we hoped it would be."

So I don't think it would be difficult to
persuade people that this is important.

But as designers, I think
it's where we can really help.

So that was,

Tom: Yeah

Jonny: that was…

It was very quickly written though,
it's probably got typos and all sorts.

I wrote it all by hand,
didn't use AI at all, so

Tom: Nice.

yeah, I think that-- it's, yeah, it's--
an interesting element to what you're

describing of the output of the work,
and the sort of human versus AI aspect

of that, but also the, process that
you go through to get to that output.

And I do think something we believe is in
the importance of craft being central to

how AI integrates into ways of working,
because I think there is-- there are ways

that craft not necessarily be associated
with the final details and polish of that

final output, but, is a mindset and a
way of iterating, refining, honing what

your internal ways of working are to
get to a consistently reliable output.

And so I think there's sort of-- want
to get craft just associated with…

And I think in the current conversation
about AI quality, I think craft

has been heavily anchored in that
final, what, customers see, what

that-- those final details are.

And I think that's part of it.

But I think there is something deeper
there as well about you can't just

have people focusing only on the final
output and expect it to be high craft,

the, output is there without having,
incorporating craft into the practice of

building products for your organization.

yeah, think the barbell is also
interesting with the sort of setting

direction at the start and then
coming in and, evaluating quality,

which I think are obviously the, most
high-- the highest leverage human

aspects of the process right now.

But I think there is a potentially a
risk of giving away too much of the

middle that the craft at both ends of
the spectrum atrophies a little bit

because less of a connection between
what are the materials we're actually

working with and how do we get the most
out of them in this specific format?

So yeah, that's maybe a separate,
slightly separate piece to your, article.

But, yeah, the

Jonny: No, it's

Tom: you're working with is interesting.

Jonny: Yeah, it's r- really interesting.

And I think your point around the atrophy,
I think it's probably fairly pernicious

and s- slow to spot at first, and hard
in any given moment to justify the cost

of doing it when AI could just do this.

AI could just do this really
quickly, so why am I doing it?

and I think, said I wrote this by hand,
but writing is thinking for me, and I

find it really hard to let-- to give
AI any part of writing really beyond

a copy edit or does this make sense?

Does it fit together?

Are the, paragraphs in the right order?

that's the sort of level that I'm
comfortable with AI writing for me.

writing, there's craft in
writing, of course, as well.

And I, would never claim
to be a great writer.

In fact, I would probably say
I write worse than the AI could

write what I've written, but it's
mine, so I want it for myself.

And I think what with when, when people
overuse this stuff is there's too much

of it for, a start, but there's also
a just sort of genericism about it.

And you can see it when you ask Claude
to write an article for you with--

around these three bullet points.

It's a- anyone could have written that.

There's no opinion there, and
it's the same with design.

It's, good and convincing until
you realize that it looks the same

as everyone else's, and it doesn't
say really anything about you.

so making it worse is, a weird thing to
advocate for, but craft can sometimes

feel like worse but different and
getting worse and then coming back in

a different way that's more powerful
or more unique or says more about you.

I think it's, that's the
work really, isn't it?

That's the-- There's no
shortcutting that, so

Tom: Yeah.

Yeah.

And I think there's a, subtle distinction
there between is the thing you care about

the one-off output or is the thing you
care about building the muscle of getting

better at that for the 10th version
of that thing that you're producing?

And in your case with writing, yes, AI
might be able to help you get a more

coherent V1, but if the thing you really
care about is strengthening your reasoning

of that idea or the thinking in that
space, then by article 10, your opinions

are gonna be so much stronger from having
gone through the manual labor of thinking

through it and writing things out.

yeah, it's a, balancing act, and
I guess that varies also by the

individual and what their strengths
are and where they want to focus.

And then the- there's a whole category
of things that are completely valid to

outsource to AI or get AI to support with.

But I think being really clear on
where, how you want to use it and

where it fits into your world, and I
think a craft mindset is a good one

to channel that energy to the right
place, I think can, really help you

get the most out of the value there.

Jonny: Yeah.

This is t-- I feel like this has
turned into like the broadest

possible conversation about AI
and design, but that's okay.

M-maybe they all just, it's, there's
an inevitability to it that, it's

such a big conversation to have, and
it really is all the, conversation,

and it's what this podcast is about.

So we're always gonna end
up, descending into broadness

Tom: Yeah.

Yeah.

We shou- we should, definitely move on.

But I think the last thing I
want to say on this is, we've

used the word design a lot.

we have consciously tried to use the
word design a little bit less in the sort

of broader work we've been doing, and
instead focusing more on the idea of craft

because, in many organizations, design
is closely connected with the idea of

craft and making sure the quality of the
work that ships is to a high standard.

But I think there's definitely--
You mentioned writing craft.

There's the sort of craft
of people management.

There's the craft of, being a great
PM or a great user researcher.

And so if you're listening to this,
don't-- Sometimes we'll say design,

but we're generally focusing on that
sort of craft outcome, through any

discipline that touches a product
which reaches a customer at some point.

Jonny: Yeah.

Tom: Anyway, that's my
final thought on there.

let's switch gears and
lighten the mood a bit.

I wanted to talk about, this last
weekend, I went on a camping trip, with

some friends, the weekend where it was
unbearably hot, but fortunately, wasn't

actually as, as hot on the campsite.

but there, there's, this talk of craft
and AI got me thinking about, ownership,

in this era, and is that even relevant?

What does it mean?

but my friend went on this camping
trip, and he a- an MP3 player, the only

MP3 player that Argos in the UK sells.

It was 20 pounds.

And he, Sam, I loved him for his
dedication to, owning music and owning

MP3s, but just hearing his stories of how
hard it was to get MP3s onto this iPod.

Well, not-- It wasn't an iPod.

It was a MP3 player.

Jonny: Yeah

Tom: the experience was made me--
took me out of my bubble a bit and

made me realize how, imagining a world
without Spotify or Apple Music or

these streaming music services is just
increasingly difficult to imagine.

Jonny: Yeah

Tom: ownership is, a sort
of strange concept in this

current moment, current era.

But it was also really inspiring to
see my friend wrestle with this MP3

player to try and make a playlist
and make it, and he had to bring his

laptop and sync it with the MP3 player.

and, Anyway, it was a, funny thing
to think about music ownership right

now and the idea of an MP3 being old
school technology at this point, even

Jonny: Yeah

Tom: it's, probably still the standard
of how music would be shared if

it wasn't for streaming services.

Jonny: Right

Tom: services are so ubiquitous now.

so

Jonny: So

Tom: yeah.

Jonny: he paying 99p for a song on iTunes?

what's the, practically

Tom: I don't know that whether he's using
iTunes or not, but he did specifically say

he was buying music, supporting artists.

he's a musician himself,

Jonny: Right

Tom: lots of, respect for
artists and buying their, MP3s,

but just it's hard, hard going.

another friend of mine, hosted a,
record party on Friday, which I couldn't

go 'cause I was at the camping trip.

But, sort of concept of this party
was to bring your own record and

tell-- so everyone at the party
could play their record and tell

the story of what it meant to them.

which I, love that as a concept.

Jonny: Love it.

Tom: extremely

Jonny: class millennial?

Tom: yeah.

Yeah.

yeah.

I, imagine that was a, core demographic.

but, but the, yeah, the sort
of the, narrative around, music

ownership and the sort of stories
associated with that, I think is

Jonny: I love it.

Tom: very, interesting to

Jonny: I just struggle to choose one.

That I wouldn't know.

How do you choose one?

Impossible

Tom: yeah.

do you own many records, Jonny?

Do you know

Jonny: yeah, about three or four
yards as, I think record people say.

so not a big collection, but I, try
and I pick things that, it's like a

subset of what I listen to, but it's
things that I feel like I would like

to put on and sit and listen to.

And I buy it if it's a good album.

I'll buy it if it's a good
shape of a sort of album.

If I think that the album tells a
story, 'cause when you put a-- as you'll

know, when you put a record on, you
have to actively listen to it because

you have to flip it halfway and the
number of times I put a record on and

then an hour, come back an hour later
and it's just like ch, on the end bit.

And I'm like, "Oh, I
forgot I put that on."

but yeah, I'm a part-time collector.

how about you?

Tom: Amazing.

I own one yard of records,
physical records, so not,

the same four yards as you.

think it'd probably be an easier choice
for me to decide what record to bring.

But, but yeah, there is something
very, rich and interesting about the

sort of physicality the, material.

the idea that you're holding the
music that you can then play.

anyway, we don't need to go on a

Jonny: Yeah.

Tom: on this, but, I think as, we get
deeper into the themes of craft and AI,

we haven't yet talked about impact to
musical artists or other sort of art

forms in general, and I-I'm curious if
that will come up naturally in time.

But this is a small sort of glimpse
of one small aspect of, this corner

of the creation of art and what
that could look like in these forms.

Jonny: Yeah

Tom: anyway, that's way deeper
than I was expecting that to go.

final, quick, thought I wanna
share before we jump into some

news is, watching the World Cup.

It's been super fun.

used this, Code, Claude CoWork.

This isn't really a calendar hack, it's
just one prompt, but I asked Claude Code--

again, I asked Claude CoWork to, add all
of the games to my calendar, obviously

in the correct time zone, and then write
in the TV station that is streaming them.

And it has been incredibly useful as
a way of just looking at my calendar

and seeing, oh, this game is being
streamed on this channel at this time.

I also, had it add the flags to the
teams, so I could just quickly scan my

calendar and see what games have been…

I haven't been watching many of
them actually because of the time

zone difference, and most of them
are in the middle of the night.

But, it's been a very,
surprisingly useful single prompt.

Have you used any, any

Jonny: No,

Tom: the World Cup?

Jonny: that is so clever
and I've not done anything.

I feel like a dinosaur.

I go to bed and then in the
morning I wake up and I watch

the eight-minute BBC highlights.

And then when England are on,
I Google to see what channel I

should turn on and that's about it.

That's, I'm, I am as basic
as they come probably.

wow, you- you've, brought AI to work.

This morning, I was like, "I need a
haircut," and I just genuinely got

annoyed that I couldn't ask Claude
to give it -- give me a haircut.

So I'm, I don't know.

I've…

Have a strange relationship
maybe with Claude at the moment.

let's move on to some news.

I've got a couple of
things that I just spotted.

One interesting and one silly,
or not silly, but cool and just

like a beautiful piece of work.

so Fable, Anthropic
announced that Fable's back.

Fable's coming back tomorrow.

Sonnet 5 is out today, I think,
so it'll be interesting to

see the reception that gets.

So these state-of-the-art frontier
models are, still advancing and

it's good to see Fable coming back.

There was a sort of vague, obscure
post about good conversations

with government and a few more
guards being put in place around,

cybersecurity and things like that.

I don't think it undoes any of the damage
that was done when Fable was taken away.

I think all of the alarm bells rung
around sovereign AI and not-- and being

beholden to the US are still ringing.

I think there's a lot of work going on
in the background, especially with non-US

domiciled companies and people to try and
work out how to mitigate for the risk.

One of which I think is really
interesting and seems to have blown

up over the last week, which is, Z.ai,

which is a, Hong Kong, Sh-
or maybe Shanghai based, lab.

They, built this model GLM and GLM 5.2,

which is an open weights model, is
out and everyone's raving about it and

how good it is at coding and some of
the things that, people have so far

relied on, these frontier models to do.

And so there is like a healthy narrative
now, which is like, "Oh, maybe this,

maybe the open source models can keep
up and maybe they can even close the

gap on this, on the state of the art.

And maybe a lot of the work we
need to do doesn't need to be

done by a state-of-the-art model."

So we can use, the, Fables to do the
really hard stuff and then these open

source models where we can host them
anywhere and at a 60th of the cost,

the token cost can do the, grunt
work around writing, writing code.

so interesting.

I won't go any more into it right now,
but definitely tracking it and I think

that there is-- it's good news for
The rest of the world, the Fables back

and the, open source models are good.

It's still a Chinese model, I think
Europe and the middle powers need

to work out how to have their own
frontier or near frontier, models.

But it's good in the meantime.

The other one that I found, and I
will share my screen for this 'cause

I think it's cool and I think people
need to see it, and hopefully I'll

remember to actually switch to the
view which shows the video this

time, which I didn't do last week.

but let me just share my
screen for Short- Shortcuts.

It's shortcuts, but spelt with no vowels.

so here you will see it.

It's, I think, just an example of a tool.

It's built by a design studio,
bra- branding studio, actually.

and it-- all it does, if you're
listening, is it's got a list down the

side of all of the different, design
tools basically, and coding tools.

And you can click one, and then it goes
into-- it drills down into it, and you

hover over the different shortcuts.

So for example, I'm in Figma, and
I've got pen tool, which is P, and

you hover over it, and then this
beautiful keyboard just like animates.

If I just refresh this, it like
animates in, and then you can

click Photoshop and scroll up and
down, and it highlights the keys.

Really simple.

this is just one of those examples of the
kind of tool that you can just build now.

obviously this took taste, and it
took time, and it was work for sure.

But, this may have been someone who
previously wouldn't have been able to have

bro- brought any of this stuff to life.

Feels genuinely useful, but it's
also just like a nice piece of art.

so put-- I'll put that in the show notes
as well in, beautiful design-y things

that now exist that didn't previously.

And this is quite a good
transition into, our workshop.

So

Tom: Yes.

I tell, the listeners about our

Jonny: Yeah, let's, yeah, you do it, Tom

Tom: So I think we
hinted at this last week.

We are launching our first open workshop,
and by open that means it's open to any

individual who would like to join us to
build a design tool together, and we're

gonna do this online on August 13th.

It's gonna be a half-day session.

more information, this at a
URL, which is probably going

to be nearfuture.works/build.

if it's not, we'll add it in the show
notes or we'll edit this out probably.

But hopefully that's the URL.

and just before we start recording
today for the podcast, we, tried to

record a 30-second clip explaining
what the workshop was about, is about.

and so we thought we'd share some
behind the scenes thoughts on that.

Although aware we're coming up on time,
so we should probably keep this brief.

But it was super fun to, a, short
script try and sound compelling.

Although I think in a lot of ways
it sounds like an infomercial.

So if you listen to it, please
listen to it, check it out.

and bear with the, awkwardness of us
sounding like an infomercial and, stay

for the part where Jonny lists out all
of the design tools we've recently seen.

Not all of them, some of them.

because, impressive that you
managed to get through that

whole list without breaking.

Jonny: I thought I was gonna be able to do
it in one breath, but it took two breaths.

I was a bit disappointed about that.

But yeah, we're, we're-- we need auto
cues 'cause reading from a script is

surprisingly hard, I think, when you're
also trying to look compelling and

look like you're looking at the camera.

Yeah.

It ended up as two minutes as--
we were aiming for 30 seconds.

It ended up as two minutes.

But I think two minutes is pretty
good, so I don't think we should beat

ourselves up too much about that.

But yes, the, the-- do check it out and
also do check out the actual, workshop

because we'll be building design tools.

maybe design tools like
shortcuts, maybe tools that do

all sorts of different things.

And we've run this before.

It's really fun.

Genuinely f- like fun and fast-moving.

It's not watching us talk all the time.

we get hands-on really quickly,
and everyone will walk away with

something that they've built and
something they can take back to

their team and things like that.

yeah, that's on the thirteenth of August.

and it's gonna be online, and
it's gonna be half a day, and,

tickets will, be live soon.

I think that's it really, Tom, isn't it?

Tom: Yeah, I think that is, it?

Hopefully we'll be live tomorrow with
both the podcast and that website, so

Jonny: Yeah

Tom: you can sign up for.

And, this is especially suited, you
can bring others in your team along

with you, or you can just come solo.

we have found a lot of companies we've
spoken to have small design teams.

there's, two or three designers there,
and people are curious to dig into this,

but it's hard to justify doing a whole
team session when the team is small.

so if you're a solo person out there
or you're in a small team, or even

if you're in a big team but your
company is not, doing as much as you

would like to do in exploring this AI
space, then this really is for you.

So you'll check it out
and at least consider it.

and if not, then, if you do check it out
and decide not to do it, then we'd love

to any feedback or thoughts you have.

Both, for that, but also
for the podcast in general.

we have been getting some
good feedback so far.

It's been really helpful in us shaping,
our format, the stuff we're talking

about, and, and potential future
guests that we're gonna be inviting on.

So please, if you have any ideas,
if you listen to this, have some

thoughts you wanna share, us.

We have a, an email address,
in fact, that Jonny set up.

I think it is [e-mail].

So

Jonny: Yep

Tom: a note there or ping
either of us individually.

We'd love to hear it.

Jonny: You can-- It's okay to
just tell me to brush my hair and

stop wriggling around so much.

That's acceptable feedback.

Tom: Excellent

Jonny: Let's call it.

one, one in the can.

Thanks everyone for listening,
and, see you next week

Tom: Thank you