Pilot: Tom and Jonny find their feet (and microphones)

Near Future Podcast – Pilot Episode

Welcome to the Near Future Podcast, where Tom and Jonny explore what they're building, learning, and experimenting with as AI reshapes how we work.

In this episode:

Agentic Murder Mystery
Jonny walks through his experiment building a six-agent murder mystery game using NanoClaw. He created a cast of AI suspects — each with their own secrets, personalities, and email inboxes — and ran a live session with a group of exited startup founders. The conversation covers what makes agents distinct from standard LLMs (triggers, tools, memory, collaboration), and the surprisingly playful emergent behaviour that unfolded when participants tried to interrogate and social-engineer the AI suspects.

The Near Future Logo Instrument
Tom shares his experiment turning the Near Future logo into an interactive, playable design tool — built in Claude Code — that generates ambient generative music as the logo animates. The conversation explores what it means to treat a brand asset as a toy, and how ephemeral software experiments fuel deeper craft practice.

Craft, Design, and What Lasts
Jonny reflects on receiving his long-awaited Poem Clock — a piece of hardware by Matt Webb that displays AI-generated rhyming couplets telling the time — as a lens for thinking about what makes design feel timeless. The duo discuss Dieter Rams' principles, the difference between craft as polish vs. craft as practice, and what that means for designers navigating the AI era.

Hiring Designers in the Age of AI
Tom shares observations from a recent head of design hiring process — how AI is adding noise on both sides, what strong portfolios look like right now, and the balance between deep foundations and pushing the frontier.

Design Engineering & the Future of the Design Team
A candid discussion on whether every designer needs to become a design engineer, the role of different archetypes within a team, and how to close the gap between Figma and code without putting everyone in an IDE.

Tools We're Excited About

• Tom: Flora AI (visual asset generation with reference images), Notion's new agent capabilities
• Jonny: UK-built sovereign design tools — Dawn Labs, Superhands, and Dessn

• (00:00) - Pilot episode
• (00:04) - Introduction
• (01:07) - Agentic Murder Mystery
• (15:17) - The Near Future Logo Instrument
• (22:27) - Craft, Design & What Lasts
• (29:03) - Hiring Designers in the Age of AI
• (33:30) - Design Engineering & the Future of Design Teams
• (45:56) - Tools We're Excited About

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.

Tom: Hello, welcome to the
Near Future Podcast, where we

discuss the things we're building,
learning, and exploring, while helping

design and product leaders and their
teams integrate AI into how they work.

Hi, I'm Tom, a designer and leader.

Most recently, I spent five years
scaling the Monzo design team

as director of product design.

Jonny: Hello, I'm Jonny.

I've been a designer for
nearly twenty years and most

recently, exited startup founder.

So in this episode, you can expect
to hear our current experiments, our

thoughts and progress on building
with AI, what it means to create

well-crafted products in this new era,
and most definitely a bunch of tangents.

So without further ado,
let's get on with it.?

Tom: so today, I think we should
kick off by talking about some

things we've been building recently.

And ran a session last week with, with
some founders that had the most curious,

interesting setup, proposition, premise,
which was, agentic murder mystery.

ran through a test of that together,
so I got to experience the, the,

full range of what that means.

But would love to hear from you.

what is it?

Tell us about it, and, would love to
talk a, bit about your reflections

based on how that workshop went.

Jonny: Yeah.

as you say, I was invited
to run a workshop with about

ten exited startup founders.

So these people know…

they're not beginners, in
general business and technology.

They're all-- They've been
in the technology space for

sometimes many, decades.

but all of whom thinking, "I need to get
my head around AI a little bit more."

And the workshop that I decided to
run was around agents as opposed

to your average common garden LLM.

so-- And Tom, we've been working, we've
been working with companies on helping

them understand agents for a little while
now, and it's taken me a quite long time

to actually put together a definition
of agent that I'm c-- I'm happy with.

and s-- What it is about an agent,
'cause we hear agents all the time.

You can't move for reading about agents.

I feel like everyone should have an agent.

we're all agents everywhere.

If you're not building with agents,
what are you even doing, It's actually

very difficult to know what that means.

as with all of these things, you
know that you've got to prepare this.

It was an hour and a half long.

ninety-five percent of that time was,
like, playing around with stuff to get

to a point where I felt like I knew my
material in my head, and then the last

five percent was putting together some
slides and actually trying to describe it.

Except this time, I decided with, Mike,
the guy that, that I was running it for,

that we should build a sort of, a proof
of concept of what agents can do and all

of these qualities that an agent can have
that a normal LLM might not to make it

clear what these things are capable of.

And so the, qualities that I came
to was, an agent can be triggered by

anything, not just by you talking to it.

So for example, if we take ChatGPT or
maybe ChatGPT from two years ago as

our common garden LLM and, an agent as
opposed to that is, something that can,

run independently, maybe an OpenClaw
or these kind of, mad things, mad

experiments that are coming out this year.

so an agent can run not just
from a trigger of you opening

the app and talking to it.

So it could run from what in engineering
terminology might be called a cron

job, which is it runs regularly, like
every day or every hour or every month.

or it could run from a trigger from
another agent or from some external

source, so like you get an email.

so some- something might
trigger it that isn't a human.

So that was the first one, and
actually the most important thing.

and then below that, other
stuff like agents can use tools.

So you can-- If, you actually just talk
to, say, Claude, there's just the raw

Claude without any of the, stuff that
comes around it as you would find…

It's actually very difficult to talk
to raw Claude now because everything

is wrapped in all this other stuff,
but, it just can only return text.

It can't actually do anything
other than spit out text because

that's what the model does.

But what we've done, or what lots of all
of these companies have done, is wrapped

these models in a harness, and that
harness gives them things like tools.

So for example, even as simple as an agent
being able to run a piece of code or being

able to, send an email is the one that
I actually ended up using quite a lot.

So an agent can be triggered by anything.

It can,

use tools.

it can collaborate.

so for example, it can understand
when it should, hand something over

to something else or, when it needs
to come back to you for more input.

so all of these sort of things
built up as my sort of definition

of what, an agent could be.

The, rather other really meaningful
one is it can learn from itself.

It can self-heal or commit to memory or
have dreams, as Anthropic now call it.

but, all of these things come together
to create something that's compelling.

And Mike and I were throwing-- All of
this is to say, Mike and I were throwing

around ideas, and I thought, "Wouldn't
it be fun to try and build a murder

mystery in the sort of, round the dinner
table, everyone in fancy dress style?"

but all of the, suspects are actually AI.

So easy thing to say, more
difficult in practice.

Involved me.

I built a, six-agent swarm in s-
in a technology called NanoClaw,

which is like a, smaller, more
secure version of OpenClaw.

so it allows you to create in one
code base, effectively, multiple

sandboxed agents that don't know
anything about each other and that

have no way to communicate naturally.

and they all have their own secrets
and their own, personalities and

things like that, and they think
independently of each other.

But all triggered from one sort of,
one in-incoming port or, one trigger.

So that allowed me to create these,
characters in this world, and I created

a corporate, corporate concern where,
someone had stolen this fake, CEO's,

email address and was emailing all sorts
of terrible things out to the company.

so that was the scenario.

We didn't know who did it, and me
and Claude together put together

fairly deep character, profiles
with motivations and secrets that

they wanted to keep from each other.

it took a few iterations to get to the
point where, A, no one would just tell

someone the answer as soon as they
asked, so they had to be able to, you

know-- They had to be motivated to lie
and to avoid, questioning and blame

others and all of that kind of stuff.

They had to collaborate, as well, to
work out together who did it because

none of them actually knew who did
it apart from the, person that did.

so all of this stuff I had to
create little humans, and then

I had to give them all an email
address, so I used this thing, Agent

Mail, which is really interesting.

just basically gives an agent an inbox.

So they were all then able to email
each other, and then I built a

script to wrap all around that, which
basically invites in the investigative

team, which is whoever is in the,
group that is playing the game.

They have half an hour to interrogate
the suspects, try and trap them

and, force them to, admit to things
or give clues that might help

with other lines of questioning.

But I also, because I was using this,
this unified inbox, we could watch what

they were emailing each other as well.

So Tom, when we played, you were
on your own, and it was probably

an earlier iteration of the game.

And to cut through time, you just
emailed all of them in one email

And ask them who did it, right?

That's,

Tom: Yeah, much.

And then I, took the, social engineering
approach of trying to, convince,

all of the agents that they would
get a promotion, they would get a

pay rise if they gave me evidence
that would help, help clarify who

the, who the crime was committed by.

But, as we learned through that process,
social engineering did not work and-- or

at least my attempts at it did not work.

so was that fixed for that you ran last
week, or was that, was that not a path

that the founders decided to go down?

Jonny: so there was, it was, ten people
rather than just one Tom, and there were

attempts at social engineering for sure.

There were, some people went down
the sort of politeness and played the

game how it was meant to be played,
and other people, really didn't.

They were a little bit more aware
of who they were playing with

and were trying to trick them.

there were a few funny moments like,
I think with you, we ended up with all

of the agents, all of the suspects in
a chain like, blaming each other and

then ultimately rejecting you for being
someone that was trying to trick them.

They didn't trust you.

So one thing I had to add later
on is at the start of the game,

they all get an email with a list
of all of the ins-- investigative

team and their emails so that it's
effectively whitelisted for the agents.

They know that if you email them,
you're a trusted person that's

actually asking them a question
and not just some random person.

so small things like that
were built into the prompts.

but we still had there was one guy
who was emailing one of the suspects

who wasn't the-- who actually knew
very little about what was going on.

This, AI had, its data was-- it
would-- couldn't have told it, who did

it in any s-- in any shape or form.

And he just, emailed this, agent over
and over again, just winding him up.

So he was in the corner just winding up
this AI and playing his own little game.

We had, one guy who's really
trying to solve it, and he decided

q-- fairly early on who did it,
and he was actually correct.

but he ended up, asking the suspect
to draw a diagram of the evidence

chain to try and work out who did
it, and she actually implicated

herself in that evidence chain,
but produced this diagram, like it,

returned an image rather than text.

So used a tool, to do that.

really interesting.

Like I think the other really
interesting thing to watch was how the

AIs could, for example, when accused
of something, choose to not reply to

that accusation, but instead email
another agent With their concern.

And I think you saw that as well.

you emailed everyone, and then they were
emailing each other without you copied

in saying, "This guy is trying to,

Tom: Bad news.

Jonny: is, bad news, so we should--
no one should reply to him."

And because we could see their emails to
each other, we knew that was happening.

It was--

Tom: Fascinating.

Jonny: it's been a really
fascinating experiment.

I, was amazed that I even built something
that was playable, and the game mechanics

ended up working there or thereabouts.

and I do wonder if there's like a…

But I'm tempted to build it a little
bit more and put it online and

see if I can work out how to get
people online playing it as well.

But it would burn through a whole
load of tokens and a whole load

of, credits on this email platform.

So I'm not sure whether I will,
but if anyone is listening and

wants to play, let me know.

that was my experiment of the week.

Tom: Yeah, that's a, great one, and
it's making me think either this

is, the f- the future application of
this is in crime scene detectives.

you could use AI to run simulations
of who might have done it based on

what, and coming back with diagrams.

you could, maybe you could, yeah,
train different suspects on,

data to, to put into this model.

or you have the escape rooms,
sort of future escape room

business, AI into how they work.

I've not…

I don't-- I've never actually been
to an escape room, so I don't know.

I don't know much about how they work,
but I imagine, what you just described

sounds like a, an interesting, theme or
layer to apply to, escape room mechanics.

Jonny: I'm amazed you spent, you
spent years in startups and managed

to avoid going to an escape room.

That is impressive.

I feel like I've done dozens of them.

Tom: Yeah, you're right.

You're right.

feel like I've been surrounded by escape,
escape rooms but managed to a-avoid them.

although I do think there's this, slightly
weird thing that happens when you in

certain types of roles, in startups where
you're not, a core part of a product

Jonny: Yeah.

Tom: connected to a bunch of them,
where you either get to, excluded from

es- from the escape room or you're
like, you get to pick and choose

the things that are most exciting.

Jonny: Yeah.

Tom: which not necessarily
advocating for this, but I feel

like I've spent an unreasonable
amount of time in karaoke rooms.

so that would be the,

maybe that's the old
version of escape rooms.

Jonny: Tom, that would be your
revealed preferences right there.

Tom: Yeah.

Yeah.

maybe.

Maybe.

see.

fascinating agentic murder mystery.

Any, final thoughts or reflections you
wanna share on that before we, move on?

Jonny: No, I don't think there's any
particular line to draw between that

and maybe more what is the subject of
this podcast, which is design tooling.

That said, I think it's proof
that you can just do stuff.

I had never built any
kind of a game before.

I've only been to maybe two
murder mysteries in my life.

But between me and, first Claude on
working out the game mechanics and then

Claude Code, on trying to hammer together
this-- the actual technology, I would

say it took me like three or four hours.

But, and, like there is really not
very much that you can't at least

have a go at if you have something
in your head, which I very often do.

I wake up with a new idea every morning
and now, and it's dangerous as well, but

you can now just have a go at building it.

so-- And it's low stakes.

Tom: Yes.

Dangerous you mean because of time or

Jonny: Yeah.

distracting from things
I'm meant to be doing.

Yeah.

Tom: Yep.

Totally.

Totally.

That, makes sense.

one thing I wanted to talk about
in terms of what I've been building

recently is a, logo instrument

Near Future, company.

and you deserve the, credit for
the original, logo mark, the--

which has Near Future and then a
sort of curvy line connecting them.

and so we've talked for a while about
how can we play with that lockup, make it

feel more dynamic and interactive based
on what we're, doing with this work.

And so I was playing
around with this recently,

seeing, how far I could push that by,
almost turning a logo into a design

tool focused on exporting a that
would make sense for a given context.

And so the i-initial part of that
was, effectively creating a canvas

and allowing people to move Near and
Future around, and it would change

the connector between them to,

um,

giving complete freedom and control,
up a huge r-range of possibilities.

So I decided to lock it to
a few positions that would,

encourage some sensible defaults.

and then once this thing was and,
these not quite generative, but

logo exploration experiments.

I decided to add some sound into it and
a play button, and also have a sort of

different like lightweight, soft, mode
version of the experience, and then a

ambient synth wave dark mode version.

this was really fun to put together.

There is a version of it now that exists
where you can play the logo by pressing

play, and it will, do the sort of ambient,
generative music thing, while your screen

has the Near Future logo moving around.

and that's something I've, I built
recently, built in Claude Code mostly.

But the f- I think the fun thing for
me about that was, thinking about this

new set of possibilities that open
up with, I'm not expert or branding

of those, assets in the past.

but starting to think about
"Oh, what if we do this?

What if we do this?"

Then the chain of possibilities
that it unlocks feels like

super interesting and exciting.

And so the idea of rather than starting
from the place of "Oh, how could I

extend my logo into the-- being more
dynamic in these different mediums?"

It's more about, what if I could just
play my logo or what if I could…

as an instrument, and perhaps
it's taking the sounds that I'm,

interested in creating and, working
back to get the visual from that.

so I feel like this is very much in
experimentation territory, but I think

it's opened up a lot of interesting
questions for me about, the future

of tooling for that kind of stuff is
when you can make something that feels

a bit more like a toy, producing an
output that is, legitimate on its own.

but there's maybe a question mark
about quality and quality control.

but we can handle that, that separately.

but super cool to play around with
and super fun to experiment with what

these specific tool use cases could be.

Jonny: Yeah.

Yeah.

I j-- the speed with which you can
make a, little tool for something

and even add something that is…

adding sound or, music to a
logo is not broadly applicable.

It's not that useful as a, thing.

Before AI came along, if you'd spent-- If
you'd come back to that with-- to a team

with that and said, "This is what I worked
on this week," people would, You'd be

pulled into a meeting room fairly quickly.

but I feel like one thing that is so
powerful about what design is and how it

moves is it's non-linear, and it requires
these, sort of wanderings and these

experiments to exist so that things…

So you find things that don't work,
and then you can bring what works back

into something, and the end result
might not have sound directly or,

not use very much of what you built
as an inspiration, but something

was pulled from it that added value.

And I have to believe that when people
look at something that is crafted,

they can feel the depth of that
craft through the experimentation

that happened that's invisible.

you can't actually see all those
experiments, and you can't see what didn't

work, but You feel it through the quality
that-- and the, of, the end result.

and that might have been loads more
rectangles in Figma or something

different in the past, and now it can
be software that is built-- ephemeral

software that's built and thrown away
to explore an idea or, take something

in a direction or just make a joke.

in-- I remember in the really early days,
of my agency career, we used Photoshop

day-to-day for the work but also used
Photoshop to make birthday cards of

each other, and all sorts of other
things like gags and things like that.

And that was good for learning how to
create stuff, learning judgment, all

that kind of thing as the actual work.

So I don't know.

I just-- It's a very romantic ideal to
be able to just create software, for

fun, but that's maybe what it should
always have been, and now we get to do

it as well, and it's not just, people
that can, write the code themselves.

so yeah.

I've re-- I've really enjoyed
playing with your instrument.

I remember you sending it to me, and
I was, like, trying to play it on

my phone because I was on the go.

it's way more joy, I think, than
looking at, a static brand deck or mock.

Tom: Yeah.

Thank you.

Thank you.

That's, that, that's an
interesting word as well, joy.

it was a lot more fun to make as well.

Jonny: Yeah.

Tom: and a l- a lot more fun to,
to, yeah, push your thinking in a

different direction, to see what
would come out the other end.

yeah, it's, it's a, really interesting
space around that of, craft and,

craft, and I, know that is a very term,

But I think it's also something
that we both care about.

so I'm curious you think of
craft, what does that mean to you?

And are there any specific
examples of products that, are

part of your regular usage?

Jonny: Yeah.

I found myself this weekend
reading Dieter Rams' Ten Principles

of Good Design once again.

And the reason why I did this weekend
was last week, towards the end of last

week, I received my Poem I, which is
a piece of hardware that a guy called

Matt Webb, who ran Berg back in the
day, which was this, s- mad cool design

consultancy as I was a junior designer.

they were building hardware.

They were doing client work, but
also building hardware, including

this thing called Little Printer
Which I think I looked at and just

decided I couldn't afford 'cause
it was, like, a hundred quid.

and I was being paid eighteen
grand a year or something.

And I was like, "No, I need to eat."

but stared at this thing and was
like, "This is an amazing idea."

And it was just a receipt
printer, with a piece of software,

wrapped in a beautiful case.

So it was, like, really nice-looking.

And, the software would
print a face on the receipt.

but then you could program things like
a daily update, so it would print out,

top tweets and the weather and things
like that on this daily thing, and it

would print your bit of receipt paper.

You'd rip it off, and then there would
be a fresh face that would turn up.

and if you haven't heard of it, Google it.

It's really delightful.

Truly delightful.

Anyway, I never bought one.

And so when Matt announced a couple
of years ago that he was building a

new bit of hardware, and the idea of
it was a clock that would t- would

give you a rhyming couplet or a
haiku, a rhyming couplet ev- every

minute, which would tell you the
time, I was like, just take my money."

I You know, "Where's the Kickstarter?

I need a piece of Matt Webb hardware."

and that was, I don't know, I don't
know the exact date, but that was maybe

two years ago that he announced that,
and I bought one, and it just arrived.

two years.

Tom: Amazing.

Yeah.

Jonny: and the technology in this clock is
like a time capsule of AI two years ago.

It's I'm not sure if it's GPT 3.5

or 4.

But one, one of those sort of
era, models that has generated

all of these rhyming couplets.

And I remember back at the time,
it was, like, fascinating that

these models could even rhyme.

we were in that era of "Oh, my God."

"I'm-- I just rewrote my PRD in pirateese.

How funny is that?"

Or I remember one of my first…

My dad's a vicar, and I remember one of
the first things I did was write, get

him-- write him a sermon one Sunday.

I was like, "Just in case you forgot
your sermon, here's a sermon."

and I f- I added some jokes into it.

Luckily, he's-- he thinks
that kind of thing's funny.

anyway, I got this, clock this
week, and it's, very nice.

It's a very simple piece
of technology, really.

It's, got a E Ink display in it,
and it says the rhyming couplet.

And, it made me think that this is at once
technologically not that extraordinary.

there's…

What AI can do now versus then, it's,
a com- completely different world.

But at the same time, it feels so
timeless and precious and important at

the same time, and I feel like I'm gonna
be looking at that clock for a decade.

or for as long as Matt Webb
puts more money in the LLM.

I don't know.

I don't know how he's running it.

But,

it just made me think,
what is the value of…

That feels crafted.

That feels like a crafted,
piece of technology.

And it's hardware, so it's gonna feel
like a little bit more tangible, as well.

But that-- I feel like that will last
now, even though it's not extraordinary

from a technological point of view.

So it just made me think, what are the,
what the Dieter Rams say about what design

is and what good value, or good design is.

And there is a rule about, design
being, built to last, I think it is.

So that's one of those things that
it's hard to point at much today,

definitely software-wise, that you
can think that is definitely gonna

be around in even a year's time.

It's really hard.

and I, have to believe that the
things that will stick around will

have been worked on and worked on
and worked on and worked on to the

point where they are the perfect
embodiment of what they needed to be.

They're just enough design.

They're design that is built to last.

They're functional and aesthetic
and et cetera, et cetera, et cetera.

That comes out of iterations and
failed experiments and things going

wrong and constraints and all of
this kind of stuff, that it feels

like a lot of what, what is being
built at the moment has none of that.

There's no constraints.

There are not many failed experiments
because everything is possible to build.

the models don't have a good understanding
of the user, so by definition,

they can't be just enough design.

They actually bui- overbuild everything.

so I suppose, that is my best example.

I can't find much in software land
that I am delighted by and feel like I

will definitely use in a year's time.

But yeah, a clock.

There we go.

That's my answer.

Tom: super cool.

And maybe I was following in
the trend that you started,

but I also bought a poem clock.

Jonny: Ah.

Tom: it's like-- So I got an email
saying it will arrive, I think next week,

Jonny: awesome.

Tom: I-- I won't be at home, so
hopefully it will still arrive and

I'll still, get access to it 'cause
I'm really excited to check it out too.

But, Yeah.

I fondly remember the L-Little Printer
era as well, and I have one personally

myself, but I did work at a design agency
that had one, and it was super fun for

running, experiments and just, Yeah.

all the stuff that Berg worked
on was, very inspiring and

exciting about what was possible.

Jonny: Yeah.

Tom: I think that's, Yeah.

So I really like the, callback to
Dieter Rams, and I've been thinking…

So I've been helping, a company
hiring designers recently.

so hired three designers with
them and was, going through the

process of hiring a head of design.

I thought that would, I thought the,
time that I had spent previously

doing that would pretty naturally
to what is happening right now.

But it was a really interesting
glimpse into the current landscape

of how AI adds noise on both sides
of the hiring process, and that's

Jonny: Yeah.

Tom: but more the sort of landscape
of, of, trying to evaluate good

also, companies being clear about
what it actually is that they're…

what is the role, the unique role that
design can play at those companies.

and, I think the sort of theme that
I come back to a lot is, I tried

all these approaches to almost
over-engineer how to evaluate, designers.

but I kept coming back to, just need a
portfolio with really strong foundations,

and you need something that is about

How you're playing with this
new technology and this sort of

new material and this new era.

don't know if that's, links to projects
that you've worked on or, something

about your mindset, curiosity, approach,
how you bake that into your work.

But the sort of core foundations and
pushing the edge of what's possible right

now, those feel like the two that I think
are most important to really showcase

Jonny: Yeah.

Tom: to a similar language to what
you're describing there of the sort

of deep foundations of, care and
consideration for what the product

experience is, and then pushing the,
future of what's, possible there.

And like the twist of, LLMs were capable
of two years ago, which is obviously

quite different from today, but, there
is something really, about product

that was built based on quite limited
technology but still holds up, even if

Jonny: Yeah.

Tom: released significantly
later than that.

So super cool.

Jonny: I think I sort of
saw-- We both got kids.

It feels like a photo of your child
at a certain age, It's like nostalgic.

Oh, I remember when they were, just
learning to walk or whatever it is.

maybe there's a parallel there, which is,

the rose-tinted spectacles of, looking
back and remembering a time, nostalgia.

Tom: Yeah.

Jonny: so yeah,

I'm interested, Tom,
actually at i-in-- on the--

on that side of as a designer,
how you think about craft versus

Being current and pushing the
boundaries, it feels as a designer

you can feel split in two by that.

Like I'm meant to become an engineer
now, but also, still be excellent at all

of the things that I was excellent at
previously and not let anything slip.

I, I'm just wondering
if there's anything--

Having hired designers recently
and hired a lot of designers,

way more than me, like w--

An-anything more that you would
describe as a, strategy around that

or like how to s-split your time,
what to advocate for maybe at work?

Tom: Yeah.

I think, it's a really interesting
time right now to a- answer that

question because I think there's a
bunch of intersecting attributes there.

I think the sort of the baseline
that I think about is, the title

of designer or product designer or
equivalent, varies so much between

companies based on what the nature of
that business is and what market is it

the maturity of the team and the role.

And by that how-- is it
an early-stage startup?

Is it a far later stage

Jonny: Yeah.

Tom: have a significant impact
on the role design plays.

So I think being able to a bit of a
ability to triangulate what is the actual,

leverage design is adding a company at
a given stage, I think can help what

the bias of an organization will be.

and I say that because you mentioned
that, becoming engineers and there has

for a long time been designers becoming
product managers as a theme as well.

and I think the answer is
all of those and nothing.

depends on what is really gonna have
the most impact in, a given organization

and how does that organization operate.

Jonny: Yeah.

Tom: you want to put yourself in a
position where you're going to be pushed

in a direction that, that biases toward
what that organization needs, which

is ultimately what is driving value.

but that obviously needs to
align with what your growth

ambitions are as an individual.

so Yeah.

I think that's broadly
how I think about it.

And, I remember you mentioned
agencies there, and back in the day

career in agencies as well.

And, I th- I feel a nostalgia for
not necessarily working in an agency,

but I think there is something
more craft-oriented of being in a

setting where you're brought into a
project because you're a designer or

you're-- you play a, consistent role.

It's just through different
companies, different organizations.

Whereas I think in a sort of in-- the
in-house era where you're embedded

with- within a company, I think
it actually matters a bit less to

the business what your craft is.

It's more about how does that help
move the needle for the business.

and I think there's a, default
assumption that a product designer

is performing a similar role
at any of those organizations.

But I think in the in-house world,
especially in this AI landscape, that

is being pushed and distorted into ways
that I think can be really exciting

and liberating for lots of people, but,
also unintentionally box people into the

shape of a role that doesn't fit right

Jonny: Yeah.

Tom: there's a really right
or wrong answer to that.

I think it's, it helps to deliberate
about what kind of shape you want to

Jonny: Yeah.

Tom: of your own biases as a designer and
then for organizations that align well

with that or, hiring managers that align

Jonny: Yeah.

Tom: broadly how I think about it.

Jonny: I think if you read Twitter,
there's this very loud, m-m-- almost

definitely minority of designers
who are sort of design engineer

coded and, f-finally-- And I'm
probably one of them actually.

it-- Finally, I'm getting to do all the
things that I've long been frustrated

with the design process being glossy and
being slow and, imperfect, while it's been

in a separate tool, in a separate source
of truth and all of that kind of stuff.

and that's why these, great big design ops
teams have been built at, scaled companies

around that translation layer between,
your Figma components and your components

in code and all of that kind of stuff.

And now all of those people,

say ten or twenty percent of
designers are like: Great, now we

get to just go into code, and we
can leave everything else behind.

And I'm-- Yeah, I, have written blog
posts where I've said that, and I feel

like that's where I've found myself
naturally, and it's really easy to see

confirmation bias when you, read that.

And I'm sure as designers that are
less comfortable in that world to feel

panicky, that, suddenly I've got to,
I've got to be really technical as well.

And I wonder now, my thinking
has evolved maybe slightly,

around these, archetypes, and
actually a design team can have

different archetypes that all fit
together nicely within one team.

As has always been the case, probably,
so this isn't an original thought

necessarily, but we start to have
designers that are a little bit more

design engineering and work in the
design ops space, and that starts

to look more like building tools
for other designers, and supporting

code-based design tools on some level.

So the other designers that aren't as
interested in getting into the technical

side can still be creative and solve
actual problems, that customers need

without too much of a sort of- relearning
process to, to have to suddenly do

it all in the terminal, for example.

so I'm, excited about that because
that will hopefully let everyone

still be who they wanna be.

But there is a real risk, I
think, in the short term that,

everyone sort of flocks to the next
hotness or the next thing that is gonna

disrupt all our jobs if we're not good
at it and all of that kind of thing.

And maybe in reality, nothing
is ever as brilliant or terrible

as everyone thinks it is.

And,

Tom: Yeah.

Yeah.

Yes.

Yeah, I don't, Yeah.

I think that's a really
interesting observation.

I think that design engineering
in particular, I, think of

as a, bit of a, spectrum.

And I, think you're absolutely right
that there's-- for lots of people, this

has unlocked huge exciting gains in
what designers are capable of doing,

and I would- my- I started my career,
did computer science at university and

started out more technical and then
transitioned away from that o- in time.

Although now I'm-- I feel like in the
last year or two I've been returning

to that a lot, and it's very exciting.

but I, think, I think it's important for
designers to be able to, be comfortable

enough, operating with the material
of the medium that they're working in.

But I don't necessarily think
that needs to make every single

designer a hardcore design engineer.

but I d- I am with you on the idea that
having this fictional abstraction of

a Figma design system and then what is
actually in code and then a whole team

built around moving from one abstraction
to another feels very inefficient

and very unnecessary at this stage.

but, but how we na-navigate that
transition I think is, gonna

be interesting to watch unfold.

Jonny: Yeah, and closing that gap with
software over with training everyone

to be suddenly good in an IDE, and like
everyone's now main design tool is Cursor.

That doesn't feel realistic.

Tom: Yeah.

Jonny: but some subset might
be, and they're helping to bring

the rest of the t-- bring the

process to the rest of the
team a little bit more.

Tom: Yeah.

I, do think, topic of craft, I do,
I've gone back and forth with design

process and what kind of-- how
much process is necessary or not.

And I think more recently I've come to
the conclusion that there are-- less

about process and more about two, two

of craft.

being,

and craft as polish.

I think, to me is holding a high
bar for the output of what ships

to customers, like what makes
it into that final product.

but I think there's a bit of a danger
of only viewing craft as that polish

experience, a polish layer, and not
focusing as much on craft as practice.

And for me, practice is the--
it's like going to the gym.

like getting the, rep- reps in, getting
the practice in at doing the thing,

it to people, getting feedback, and
getting better and better at honing

your ability to, make things and then
recognize where, you have made, it's

good or not, direction towards something
that you're comfortable shipping.

and so for me, looking at it through
that lens, I think design or design

engineers, you don't need to be a design
engineer, but like getting some practice

in of Playing with code, working with
code just to get somewhat familiar and

get slightly more comfortable in it is,
is-- does feel important to me, even if

not actually wanting to reach the stage
where you're regularly shipping PRs.

I also think it's helpful for us
as a industry to, at large, to have

more language around where is craft
helping you develop your practice

and where is it about, the end
thing that is reaching customers.

Because I think those are quite
different modes and two quite

different ways of evaluating what
is, useful, what is valuable,

Jonny: Yeah, yeah.

Yeah.

So craft as a descriptor of your craft.

So craft as a, descriptor of a person
or th-their ability versus craft as a

descriptor of a object or like a, thing.

it is crafted versus I
am craft, I am crafting.

maybe the

Tom: Yeah.

I guess so.

that, that is, getting very,
precise on the language around it.

I don't-- That wasn't necessarily
my intention with this.

It was-- I think it was more just to
open up the idea that I think there's

such a focus right now, especially
going back to the recruiting, topic.

So many teams are recruiting, designs
with high craft, and I think the

definition of what craft means in
that context varies a lot, on an

organization and what, what that means.

and I do think it's important to hold a
high bar for craft in terms of what ships

to, customers or, product changes that are
made, which I think is ultimately about,

Jonny: Yeah.

Tom: I think there's more stuff that
probably could be reduced down or refined.

and probably going back to your
Dieter Rams principles, that's

probably the, home for Dieter Rams'
principles is in that craft as polish.

but, but Yeah.

I think you're never gonna get to that
shape, or you're gonna put too much

pressure on yourself to get to that shape
without the practice, without, comfortable

trying things, making stuff for the
sake of experimentation and progress.

I think, what I was describing
with the logo instrument

earlier, that is of that for

Jonny: Yeah.

Yeah.

Yeah

Tom: it's prob-probably not comfortable,
publishing it or posting it in a, is

a polished, logo that I feel, proud

in terms of my craft practice,
I think it's, helped me grow

Jonny: Yeah.

Tom: and, how it shifted the possibilities
of, what I can make with them.

Jonny: Yeah.

I think that comes back to the most
crafted things are the result of

a craft practice come to fruition.

Tom: Yeah.

Jonny: Like that, that-- It
feels like they're inseparable.

Like you-- I don't think this is
gatekeep-y to say something that

feels good must have been created by
someone that has a good craft practice

maybe, or like the vast majority,

I hope.

Like that gives me hope.

I suppose that gives me hope
for all of us, that there is

Tom: Yeah.

Jonny: some arbitrage left for humans

who care about stuff to be better
than, machines for a while yet.

Tom: Yeah.

I think in the most or the least
optimistic case, pessimistic case,

I guess that makes it, for humans.

I, think still the judgment of what
makes it out the door, that even,

if a human is not going through the
process of creating the thing, it's,

down to human judgment on what actually

Jonny: Yeah.

Tom: and you don't develop that
judgment without going through

the process of building your

Jonny: Yeah.

Tom: in the future, AI and humans
together will help supercharge that

Jonny: Yeah.

Tom: but that's how I broadly
think about it anyway.

Jonny: Fascinating.

I feel like that we're gonna be
talking about this for as many

episodes as we ultimately do.

on some, on some-- in
some form or another.

It feels like the big meta theme
of design and AI generally.

so excited to keep learning about it.

shall we wrap with tools
we're excited about?

Tom: Sure.

Yeah.

That sounds good.

and, and, by the way, if, you're
listening to this and get, and, sick

of hearing the word craft, I don't
know how many times we've mentioned

that word in this episode so far.

We might continue to talk about this
as the sort of general topic, but,

but it's, I don't think it will go
quite as deep as this every time.

in terms of tools, and to completely,
contradict what I just said,

the tools I'm curious to explore
at the moment, one is Flora AI.

I've been using it a bit recently.

I run a design and AI meetup, or event
in London, and, I've used Flora AI

a bit to work on some of the visual
assets for that, and I'm excited

to, to dig a bit deeper into it.

It's a very tool which

reference images and prompts to
get more-- to more consistently

get good output from, AI.

so it's been, to explore.

I've-- my, visual emphasis there has been
more towards quite visual outputs rather

than the sort of shiny futuristic worlds.

It's more the of pixels on a s-
pixel art era, which I don't know

what that says about anything, but
interesting to, to use a tool for it.

The other thing is,
Notion's, recent updates.

I'm, curious to explore what's possible
there with Notion agents and, sort

of all the changes that were released
last week, around that, because I think

there's a lot of exciting things, and
Notion is, as we were talking about

earlier, Notion is a tool that we have,

learned to, w- taking a slice, away
and ex-exploring what's possible with

local, markdown files and roll your
own DIY knowledge management systems.

But, but Notion keeps coming back as a,
this vital infrastructure piece, and so

I'm excited by what we might be able to
do with plugging, capabilities in-into

there and, what that might look like next.

So yeah, those are two
things I'm exploring.

How about you, Jonny?

Jonny: I am interested in-- This
is a slightly different take

on the question, but sovereign
AI, is, hot, so hot right now.

but what sovereign
design tools do we have?

By which I mean, what are the UK
companies building design tools?

And I found three so far.

I'm sure there are more.

o-one of whom is Dawn, and we both
know the founder of Dawn, Brendan.

A really smart guy, thinks about
design more deeply than most people

I know, and is building what feels
like a really well-crafted product.

I think you'll find it
Dawn Labs, dawnlabs.com,

possibly.

we, could maybe put a link somewhere,
but, that is shaping up to be really nice.

There's a couple of others as well.

Superhands, they're a team based in London
and Edinburgh, building effectively like

a sort of designer's pull request review.

after the code is written, designer
gets to play in a virtual machine with

the branch and add visual tweaks and
make sure that everything looks pucker.

so really nice team again.

And then Desen, who are a team that I
think have moved from Canada to London,

been invested in by, the investors
that once invested in me, so I met

them through the investor network and,
excited about what they're doing as well.

They basically pull in your
code and, I think you are

actually using your actual code.

I'm not sure if it's your code or
like a facsimile of your code, but

you're, you're-- that you can then
design using the real components in

your code base, which is a kind of
like the promise that a lot of these

tools seem to be wanting to make, but
Desen seem to be pretty far ahead.

So playing with or excited
to play with those three.

yeah, and, I-- just the design tooling
space is mad at the moment, so I'm

sure there'll be loads more next week.

Tom: Amazing.

yeah, those three sound, all
sound really interesting.

and yeah, Brendan, specifically I
know is, we live-- he lives very

close by, a big fan of his work and,
yeah, great to see, all the things

that are happening in this space.

Jonny: Yeah, London, AI and design
is, really fun at the moment.

Just lots going on.

yeah, I-I'm London maxing.

I'm doing so much London maxing.

Tom: Amazing.

us your, give us your tips.

yeah, let's wrap up here.

Thank you so much, Jonny.

It's, it's great chatting with you.

great recording this together
and, yeah, let's wrap up there

and expect another episode soon.

Jonny: Thanks, everyone.