June 10, 2025 • 51 mins
Sign up for A.I. coaching for professionals at: https://www.anetic.co
Unlock the secrets of better communication with AI! In this episode, we explore how to use AI for leadership development, enhance your communication skills, and improve user experience (UX) with AI tools. Innovator and author Ian Pilon (Cultivating Clarity) joins us to discuss the future of human-computer interaction beyond the simple chatbot. Learn how to leverage AI as a sparring partner, analyze your subconscious behaviors to become a more effective leader, and use contextual intelligence to get more from your prompts. We cover everything from building AI agents and organizing communities to the practical strategies for using large language models (LLMs) to overcome the blank page problem and prepare for difficult conversations.
Learn more about Ian Pilon:
https://www.linkedin.com/in/ianpilon/
Get FREE AI tools
pip install tool-use-ai
Connect with us
https://x.com/ToolUseAI
https://x.com/MikeBirdTech
https://x.com/IanTimotheos
00:00:00 - intro
00:01:09 - How Logistics & UX Inform AI Development
00:05:42 - The Future of UX in a World of AI
00:18:02 - Using AI as a Mirror for Better Leadership
00:33:43 - Sparring with AI: Role-Playing for Difficult Conversations
00:36:59 - The Power of Contextual Intelligence in AI
00:46:02 - How to Build a Local AI Community
Subscribe for more insights on AI tools, productivity, and communication.
Tool Use is a weekly conversation with AI experts brought to you by Anetic.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
That epiphany or the aha moment that you might not be aware of
this, but there's a subconsciousbehavior that you have and it's
not serving you or the others around you.
But I think with the right technology stack you can really
do some interesting stuff here. Welcome to episode 43 of
Toolies, the weekly conversationabout AI tools and strategies
brought to you by Anetic. I'm Mike Bird, and today we're
(00:20):
talking about how to use AI to be a better communicator.
We're joined by Ian Pelon, an innovator who sits at the
intersection of AI, human centered design and leadership.
He organizes AI agents. Waterloo and Ian's book
Cultivating Clarity is a must read for leaders and product
innovators. Ian, welcome to to use.
So happy to be here, Mike. Thank you.
Yeah, I'm, I'm pumped. And it was such a pleasure to
(00:41):
have you at one of our events. I didn't get to talk to you much
about the book, so maybe we can cover some ground on that.
But yeah, for those that don't know me, I'm in Waterloo ON
Canada, an active builder, playing around AI technology and
just having a lot of fun. So yeah, thanks for having me.
Mike, absolutely pleasure to have you here.
And that was a great event. I really hope more people in the
Waterloo Region kind of start aggregating together rather than
(01:03):
just all migrating down to Silicon Valley because
Waterloo's got the clout, but weneed to make sure we got that
home grown talent going. Would you mind giving us a
little bit of your background? Yeah.
So in the last decade I've been into UX research and design and
then over the last couple years moving really heavily on the
product side around AI. But before that, logistics,
(01:26):
which is kind of an interesting thing that's sort of resurfaced
in my ability to see patterns related to using AI and, and
flow design, especially around agentic systems.
And thinking about the relationships of, you know,
supply chains. Funny enough, in some of the
(01:46):
principles that I've, I had in logistics before I got into
technology, a lot of these ideasaround what I started seeing in
in tech were practices I've seenin logistics before I got into
technology. So that's always been cool
because I guess Japanese philosophy, I used to work for a
company that supported Toyota supply chain and they all
(02:09):
applied lean methodologies and use things like Kaizen and all
these Japanese ideas around efficient systems design, right.
Just in time, logistics, all this stuff when I ended up
getting into technology was, wassurprisingly relevant or, or
just familiar. Sorry, because I was like, well,
(02:31):
I've seen this over here in, in,in logistics.
But then it's funny that some people in the tech side of the
world, and I guess in software tech weren't familiar of where
the origins of some of these ideas came from.
So, so it was kind of cool. It was familiar to go from the
physical world of logistics intosoftware design and seeing a lot
(02:55):
of similarities and trying to just have work flows that are,
are products that are helping somebody achieve something
somewhere in a supply chain. So I love the mental models that
I took from logistics and, and sort of the, the systems of
efficiency in that world. And, and, and realized, you
know, in the UX, if you understand the flow and what
(03:16):
people are trying to do, it's very transferable into building
actual products and services that people use versus not use.
So yeah, UX has been a huge influence into then bridging
into AI, the AI world now and why I'm sort of obsessed about
the customer relationship between the technology shift and
(03:40):
use users using tools or customers to getting jobs done.
It's just a really fascinating intersection, right?
That it doesn't go away. The fact that we're tool use
humans and just happens to be the tools that we're creating or
getting really, really powerful.So that excites me and I go,
well, how's that going to, how'sthat going to affect me?
(04:01):
How's that going to affect the work I do, the people I work
with, and then obviously my children.
So I'm very particular about sort of trying to anticipate how
this new thing might be used in the future and start to learn
more about it. I love learning.
So I go all in. And then if there's like, if
it's something currently like AIand agents, which is so new,
(04:22):
like you're not going to learn about that in school or in
college and university because some of the stuff is just
bleeding edge, right? You're not going to learn about
memory and agentic flows. It's just not in the curriculum
yet. Maybe for some real time
schooling stuff, but that's where throwing events is like,
hey, who else is sort of at the bleeding edge of where things
(04:42):
are going like yourself and and then going, how are you using
these things? What are you, you know, what's
your perspective on it? And that helps me just learn by
osmosis. Absolutely.
And it's one common theme that keeps coming up where people
with different types of backgrounds are kind of
migrating to the AI world, Bringthose different experiences with
them. Last week we had on Michael
Tiffany and he brought up how the concept of a desktop
(05:04):
directories being folders and files on your desktop is just
representation of how people used to work.
But now you're getting people entering the workforce.
And like you said, there's there's no historic knowledge of
AI ages. This is a brand new thing and
they're bringing all these different mental models into it
and being able to solve problemsin unique ways.
And everyone is starting to be able to be enabled with this
technology and have their outputamplified that.
(05:25):
It's getting really exciting just seeing a whole new paradigm
being formed rather than just the next gradual iteration.
So 1 area that I'd love to hear some thoughts on from you are
just the UX of current AI tools.The chatbot is everywhere.
There's always a little empty text field that you go to in
order to type and kind of createthis engagement.
But where do you see either the current state or the future
projection of how humans are going to interact with AI?
(05:47):
Yeah. The to answer the question about
where do I think UX is going or where it is it right now, I
think we can't go without talking about the GPT moment.
Like anybody who is a user that got onto Chachi BT and first had
their interaction experience with LLMS, There's no question
it was, it was pivotal. And in doing that for, for
(06:11):
whatever somebody's experience was with that type of
interaction. To me, my, my like aha moment
was that I could finally use English to talk to a computer.
I didn't have to know code. And that's been always sort of
like this friction of me gettingthe output from machines that
(06:32):
I've wanted. So in doing so, when I had my
ChatGPT moment, it was really, Ithought it was a revolutionary
change. It was like, Oh my God, I've
been waiting for this forever. I can finally talk to a computer
and have it talk back to me. And then and then even then
really start to push it to go, well, how, how far can I go with
this? Can I get it to build an app?
(06:54):
And right away I was like, like trying to go, OK, well, I don't
understand Python, but I understand some principles about
what an application is a recipe.So can I get it to do something
basic local on my machine and then see if I could really be
communicating with the computer right?
So now it was like, OK, can you build me something?
(07:16):
OK, so, so then Fast forward into the UX paradigm of, of what
I started using it with for myself personally.
First was all right, I now have a translator real time that I
can talk with a computer and getit to build something for me.
I've always wanted that. And then it showed up.
So in around that time, I was also super interested in
(07:39):
thinking about, well, this is a,this is a massive shift in tool
use because I don't have to learn sort of like Python.
I mean, other people can. And this does not discredit or
take away people that have learned the hard way.
It's, it's, but for me, I did never, I never learned how to
(07:59):
write or learn those language, how to write them to talk to a
computer. So now I had a translator.
Basically, that's how I think about it.
My perspective is that I have a translator.
I can not only ask it to go communicate, I don't actually
care what language it uses. I care about the outcome that
I'm trying to get. So an example might be, I mean,
(08:23):
I'm also into sound design or orplaying with music.
So can I write a simple app thatuses the microphone on my
computer to pick up my acoustic environment and give me a
visual, visual output? And sure enough, you can one
shot these things and I was justlike, Oh my God, this is crazy
because you know that hello world feedback used to take
forever. Even when I was in web design to
(08:46):
build a web page just took took a bit of work to get a hello
world up. Now you're talking about like
instantly, which is absolutely groundbreaking.
So that excite me led me to going a little bit to think
about right away the future. And to answer your question,
where does UX go? Well, I realized that if anybody
(09:06):
can build instantly, now there'sa whole world of how we used to
work that gets that just changesbecause now, like if you think
that we're building in software often tools for other people
right away, I'm like, well, people will just build the tools
(09:27):
themselves. If you can speak English.
I mean, you don't, you know, youknow, you can now solve a lot of
your own problems. Now things are moving really
fast. But I was kind of like this
might make my role in what I do change.
So how do I stay relevant in thefuture if what I used to do is
help inform engineers on on why we're building tools?
(09:49):
Everybody is on some journey. And in UX, what I used to do is
understand where they fall in this journey.
So the tools that we built are sometimes helping them solve
tasks that are in the milliseconds, seconds or minutes
sort of time scale. I'm bored.
I'm going to go on Twitter. I'm no longer bored, right?
(10:12):
The job to be done there is verysmall.
It's minute, but it is a thing. And if you don't fulfill it,
they're going to do it doing something else.
So that's where you can insert certain things like social
media, but as ultimately you go up to more complex products and
services, you're getting into typically things where people
(10:34):
are trying to get tasks done. Sometimes it can go from minutes
to hours to days to weeks, basically a bit of a gradient
here, but you're in a sense making zone.
And this is where a lot of B to B plays are and stuff that it's
like, no, I'm using this, this tool is helping me move forward
on some project. And it's a much bigger thing
(10:56):
that you need to do. And in this range, you often
have complex systems and there'sa lot of engineering and product
involved and you end up buildingcool products and services.
But this now hyper shrinks down,right?
Because it just used to be, there was no other way to do it.
Before you had to hire a UX practitioner.
They would do a bunch of research and then they would
(11:16):
come back and talk to engineers and build a prototype and then
put it in front of people and then wait for feedback.
And you know, this, that that could take weeks, but
nonetheless, the user's goals and needs don't change.
We're all struggling to make progress towards goals, whatever
those might be. They're just different for
everybody. But we always have to take into
consideration of the things thatwe're trying to build.
(11:38):
Who's it serving? It's not for ourselves.
Sometimes it can be. But I just happened to be doing
this for a long time. So I got really good at
understanding on going to do thediscovery research to figure out
what people are trying to do. Is it, you know, is it a short
term interaction experience? Well, then that's what I call in
the book is low customer contextand and in, you know, thinking
(12:04):
about how I was going to play inthe future.
I also have to think about like how people are interacting with
tools. The the human computer
interaction paradigm is just evolving just like language.
So I think you mentioned the, the story about like the the
first stages of computers being,you know, a mental model of, you
(12:25):
know, folders and trash cans. You, you worded it a bit
differently, but it's basically like you, we gave users a mental
model to get away from the, let's say the command line
interface where, you know, all of a sudden when somebody could
look at a screen and have a visual metaphor that, oh, if I
want to delete a file, I can drag it to a recycling bin.
(12:46):
That was very powerful to introduce this technology to a
whole world of new users that would have never been able to do
write some code to get it to do that.
So that democratization is part of the interaction of how humans
are using the tools. And now we're going to shift
into the ability for people to speak plain language to solve
(13:09):
some of their problems, which isinteresting.
We're all at the very cusp of this happening, but ultimately I
think this this goes to where we're all striving to become
other versions of ourselves, even if it's subconscious.
So some people are, they're veryforward-looking and they do have
goals and they know who they're trying to become.
(13:30):
Other people. It's just subconscious.
They actually don't know it. But regardless, they're going
towards becoming something else,whether you're conscious about
it or not. So if you can sort of think
about the spectrum of humans moving through time, there's
tools in that flow. And there's no question that
(13:51):
software is just hyper embedded in everything we do now.
You can't get out in the world without being in the middle of a
software experience. So those are all like hyper
micro interactions. You're on Twitter to Oh my God,
I'm at work and I've got all allkinds of products and tools that
help me get my job done so that I can get paid.
But at the end of the day, we'reall striving to become some
(14:13):
other version of ourselves. And and so on that scale, I like
to think about context as how I think about the potential change
of experiences between humans and computers.
And, and that's why I wrote the book because I was like, well, I
(14:35):
think I know what context is, but I actually want to dig a bit
deeper. When that ChatGPT moment hit, I
was realizing that it was a paradigm shift.
All of a sudden I seen a old world where there was all this
work and activity into shaping somebody else's experience.
And all of a sudden it gets democratized.
And now a lot of that work is not needed.
And you can just not in all scenarios, but in some you could
(14:58):
just clearly see that, OK, there's not going to be a need
for that. They can do it themselves.
So I'm excited about this idea that it's it's going to change.
And do you want to participate in the change of how tools are
used for humans? So you can sit on the sidelines
and not do anything or you can actively participate in trying
(15:21):
to imagine and play that. I think there's a word in an era
of play. Same like sort of like the
mother of all demos. Do you know that that demo, the
human interaction demo back fromwhatever the 60s and that was
when like the first experience of the mouse was sort of
portrayed to the world. Those ideas are we're we're at
(15:41):
that now, I think with AI. So to be honest, I don't think
there's a solid answer on like where do I think the UX is
going? I just know, and you're right,
chat was this first thing, but it's kind of wild.
I started playing around with these ideas about like, OK, can
I have generative UI respond to content?
(16:01):
So now I'll take content and sort of like, hey, what does
this look like as an animation? What does this look like as a
slide deck? What is it?
So I think you're going to get into almost a hyper personalized
way that each of us wants to getthe information back as a tool
(16:22):
in, in the medium that we we love.
Like, so I love watching YouTubevideos to learn stuff other
people don't they, they might love to just read and go deep on
a white paper. Those two extreme differences in
the media is, is, is huge. And if somebody has the option
now to switch, you know, a toggle or even not even switch a
(16:46):
toggle, talk to a system. And if it learns your
preferences over time, it can start serving you the, the
content, not only in a paper or a YouTube video or generate a
movie for you, put you in the movie, which was kind of really,
really cool. So it's, it's just it's, it's,
(17:07):
it's groundbreaking. Everything's changing in real
time. Yeah.
And I mean, it's crazy how we'reseeing this, this rapid
evolution where now I interact with my computer mostly with
voice. I'm always prompting with voice.
I know that's not for everybody.I work from home, I don't have
to worry about other coworkers and it just allows me an
increased bandwidth of input. And Carpathy actually came up
with a tweet recently about how high bandwidth video consumption
(17:29):
is and how you can really get more granular information.
And kind of, you're allude to some people prefer reading, but
what if in that reading instead of just static diagrams, you had
an animation? And what if you could interact
with in bidirectional communication to ask you
questions? The ability for us to really get
a tailored experience to exactlythe output, the outcome that we
want that you're going to is just getting closer and closer.
VO 3 came out and all of a sudden you can make these
(17:50):
incredibly realistic videos. As soon as that applies in like
almost instant dynamic wait, we're really going to be able to
learn and just accelerate our interaction so much faster.
But what one thing I want to pull it back to when we met you,
you told me this cool story about how a lot of the times you
can make improvements based on certain metrics.
But as soon as you get into a leadership role, what are those
metrics? How do you make things better in
(18:12):
a measurable way? Would you mind talking with
that? And just how how you're able to
work with communication in the leadership aspect.
So being a UX practitioner for so long, I had gotten really
good at finding friction in process.
And this happened way before I got to tech in logistics, as I
mentioned, I used to time, you know, when we would move certain
(18:35):
freight to a different lane and if you would shave off even
seconds on some of these work flows, you were saving a lot of
money because it was all hyper connected to other things in the
in the supply chain. You know, save a second here 300
and 65360 days a year. It it means that this truck can
get in here faster and get out there faster.
(18:55):
It all adds up. So, you know, thinking about
that in the context of software was easy for me because I was
like, OK, well, just the, this interaction point is where
there's, there's friction. How to, what are some ideas that
we can solve it? But I think I, I got really good
(19:18):
at go finding the friction. That's what I'm really good at.
And then from there, you can have multiple perspectives on
how to solve that problem, whichis always good.
Whether you're solo or on a team.
You can, you can say, do we all agree that this is the problem?
And then, and then you're set upfor success to deliver a
solution. When I got into managing people,
(19:41):
then I had no way to measure my this, this idea about, well, can
you lead a high performers or a high performance team?
All of a sudden you're in this fuzzy land about what, what
constitutes performance is that,you know, how many updates go
out, how many push updates are sent out.
(20:04):
So it's just really tricky way of trying to describe even,
well, you, you start to get intothe ambiguous terms, right?
So you're like, well, what does that mean?
Rather than the, you know, an abandoned shopping cart
experience, It's very measurable.
We're not getting people to get out through the checkout
experience in a, in a software. It's very a binary.
(20:25):
In the human world, it's, it's psychology and you've got a lot
of fuzzy logic in there. So when I was leading A-Team, I
wasn't sure how it was performing.
I was kind of like, I don't know, they won't tell me, right.
So you have this power dynamic as well between manager and
subordinate relationships. And, and I think we know this
(20:46):
subconsciously, but it's not actively talked about.
But I was curious. I love self development.
I read a lot of books and at that time I was working on a
project and there was an executive coach from Vancouver
who I I, I expressed this to, I said, yeah, Daryl, man, like
I've been leading this team for a year.
I have no idea how I'm doing. And that was starting to really
(21:10):
gnaw at me. It was kind of like, I don't
know, right? And, and, and that, that was,
you know, I was a fish out of water, an experience that I was
not normally used to. And he said, well, you got to
read this one book. It was called Multipliers.
And he said, you don't even haveto read the whole book, Ian just
(21:30):
read 1 chapter about accidental,accidental diminishing behavior.
So I hadn't even finished this call.
I was on this virtual call with him because he's in Vancouver
and I had already checked out the shopping cart in Amazon.
The book was on the, you know, hit buy.
I love that I could do that. I was like when you, when
(21:50):
somebody inspires you that you were like, man, this guy, I
really trust his really value his opinion.
If somebody tells me like, you know, like that, you should read
this. Those are the books I read,
right? I used to buy whatever interests
me. Now it's like if there's people
that I really follow and I'm like, that person is somebody I
admire and they wreck make a recommendation.
(22:12):
It's like I will take take it seriously.
So I did and, and I read it, youknow, cuz it showed up at my
house the next day. I read it and I had my light
bulb moment about trying to interpret this fuzzy logic, this
uncertainty that I had around, how do you know if you're a good
leader? Well, what does that even mean?
(22:34):
I had other books on leadership before, but this one was, was
really good because it said, well, all of a sudden you could
classify what a bad leader lookslike, right?
People that diminish others and they're not even aware they're
doing it. And it's a subconscious thing.
You, you, you don't always know that you're doing it.
Plus you have a lot of multiplying behaviors and people
(22:59):
don't even don't know what thoseare.
So you want to keep doing those and not doing the other thing.
So I had a couple of couple of them that jumped out at me like
the idea guy, you know, so if you're leading A-Team and
they're trying to share their ideas with you and you're like,
you have a, oh, I got this idea too.
And you're, it depends on the person you're leading or
managing, But they're all, everybody's unique, right?
(23:22):
Depending on the state of your career, you want more autonomy
or less or more agency or less. Some people need more help on
the support side. Some people want like, just
leave me alone, get it on my way.
And you have to figure that stuff out.
So I was able to read the book, get the epiphanies like, oh, I
might be doing that. I might be doing this.
(23:43):
All these other ones. I definitely know I don't do
that. So it was really cool.
And then from there, this was also the time that catchy PT had
hit. So I was like, I was wondering
about these new interaction paradigm, Speaking of, you know,
how UX might change. And I started thinking about
(24:04):
recording myself to try and see if it could catch these
diminishing behaviours. Now this is obviously a very
sensitive area and rightfully so.
I'm super big on privacy by design.
So I tried to to mock up these ideas at at home, working with
the local tools to see if it waspossible to record a
(24:30):
transcription. I have one with my my daughters
and try to get the the LLM to make inferences almost in real
time on the communication between two parties to figure
out if you could have enough context to detect a diminishing
behavior in myself. So think of it like a mirror.
(24:50):
That's what I was trying to build originally was this mirror
and to see if I could catch diminishing behavior.
So I had prototyped that and I think I have AI do have a post
on LinkedIn and Twitter a while ago when I, when I had it
realized that Oh my God, LLMS can do this given the right
(25:12):
context, they can do a lot of stuff.
So I was like, well, wow, this is really, really cool.
So I could just have this self monitoring myself.
And I remember also telling, I think my brother was like, like,
'cause him as well as other people will know that if I go in
your house and you have Amazon on or you like Google, you have
(25:35):
these listening devices. I'm like, they're all listening
and I have nothing that is sharing that data back to me.
So I I say things like Amazon and Google know me better than I
know myself. And I'm like, well, I want to
build a private by design personal sidekick that can
monitor myself and and then see if that could be useful, right?
(26:00):
Could I, could I do what the book did, which was give me that
epiphany or the aha moment that you might not be aware of this,
but there's a subconscious behavior that you have and it's
not serving you or the others around you.
So as I, as I mentioned, a bit of a sensitive area, but I think
with the right technology stack,you can really do some
(26:22):
interesting stuff here, right? Without going too deep into
that. I mean, there, there is PII
scrubbing tech that you can use zero knowledge proofs, all kinds
of weird things that people are doing to keep data private.
Now also because I was in UX forso long, I just got used to
(26:42):
whenever I have a customer discovery call, I would ask, are
you OK if I record this? And if they say no, I'm like,
OK, cool. And I'm back to the old school
scratch and notes and trying to remember everything.
But for most people there were, there were like, yeah, cool.
As long as I tell them what the data is being used for, right?
This data will not be used outside of this company or so on
(27:04):
and so forth. So that's what I'm doing with
Nero Kick or the leadership development is if I'm not
leading A-Team right now. But if I was, it'd be like, hey,
I've got a sidekick or a coach and it's a mirroring device.
It's not, you know, listening toyou, but it can delete the data
instantly. Do you want me to turn this on,
(27:26):
On or off, Yes or no? I'm trying to become a better
leader managing you and I don't know what I don't know.
So does that answer your question?
That's sort of the the technology that I'm playing with
right now and having a lot of fun because it's forcing me to
learn what the limitations of are, what the limitations are of
using these new technologies right there.
(27:49):
And it gives me a chance to playwith them at the same time.
Yeah. And I think that's a really
interesting approach because when we're just unconsciously
having a conversation or just, or just, you know, engaging a
regular life, we're not always being analytical about
ourselves. So having a device that's
actually like being introspective and trying to
analyze what you're doing does offer a lot of real insights.
(28:09):
I'm curious about for the more active mindset.
If you are putting together an e-mail or some type of external
communication, a lot of people will resort to using AI to
either help with the blank page problem or kind of proofread.
To summarize, have you found success using AI to help with
your direct communications, or does it always end up being slop
and you have to rewrite it anyways?
Well, that's a good question. I think there's two things in
(28:30):
there. The blank page problem is just
that's a problem on its own. That's just beautiful that LLM
solve for you was like, and I think I get this from remix
culture and music. It's always good to have
something to jump off rather than starting with the blank
slate. Inspiration doesn't always come
to you instantly. So if you're in a creative role
and somebody's like, OK, I need this design tomorrow, you may
(28:53):
not be inspired. So that's when you're often then
going what we're looking at other things to find inspiration
in. So yeah, the jump off point for
the blank start problem is absolutely huge.
I think everybody's found utility in that as far as using
it to correct your, your writingor not even correct it help
(29:17):
augment your your writing. I used it instantly to write my
book. And so I'm not a writer and I
was looking at hiring a ghost writer.
So in my book, Bob Mesta was oneof the people I interviewed in
my book. And what a what a, what a gift
to, to talk to Bob Mesta. And he told me he had hired a
(29:41):
ghost writer for one of his, told me the price.
And I'm like, yeah, that is justnot in my, my wheelhouse.
So I'm going to use AI and I have no, you know, no shames
about it. I was like, I I'm not a writer,
but I will figure out I do what I do is I, I went and grabbed
all the story arcing frameworks and plots there are known to
(30:02):
man. And then I was like, OK, if I
have 9 interviews, this is what was in my book.
I want to get different perspectives.
And we're talking about context in the framing of product,
developing products and services.
So I, I had a big mural board and I, I visualized the story
arc of how I could put this intoan overarching story.
(30:26):
So that's where I used AIS at that time.
I think it was Gemini. I wasn't ChatGPT, but I was
like, you know, OK, I've got these nine different interviews
and how do I have an arc and a story that cohesively puts it
together? And, you know, it's basically a
sparring partner. And while you could try in this
and you can, you can, you know, come out with a framework where
(30:50):
you can, you can go at it so many different ways.
So then I, you know, I use it asa sparring partner and go back
and forth until I landed on something that I was happy with,
not only with just AI, watched alot of videos on storytelling
and then and then from there, yeah.
So it was like, OK, I'm going tocome up with this crazy intro
(31:11):
and this hook and this is what I've got and sparred with the AI
back and forth. And then it's really back and
forth. So what about this?
No, I don't like this. And I and I fine tune it, right.
OK, cut this out. Don't like how it's sounding
like that. And you could you could at that
time, it was so obvious when youcould see like AIS would would
(31:32):
use the word delve into something over and over and
over, you know, But by doing this, I was able to see, I was
able to really find the some of the common patterns that I would
I was seeing in the LLM while working with it.
And then going, OK, as I keep going to write my book, if it
was a ghost writer, because evena human ghost writer is is not
(31:56):
going to be your voice, right. It's like it's trying the best
to it can be to interpret the story in your voice, but it's
still somebody else. So I just ran with that.
I was like, OK, well let me see what I can do a lot anything.
It took me 8 months even with with an AI and I don't even know
how long it takes with the humanghost writer.
But I had spent almost every evening going to the library and
(32:22):
just having space to, to spar with this thing back and forth.
And you know, I had my overarching map of where I think
I wanted to go with the story and then just brute force
whittled it down like, OK, got to get this first section of the
book to be the hook before it gets into the the interviews and
(32:42):
then have the ACT, the outro. Oh, and, and back to that map
that I sort of had as a screen share up.
That's that's that's I guess a visual way that I try to capture
the essence of the book. And so although that's not
writing all the by, by the time I was done the book, I was able
to go, OK, put the book into an LLM and go, you know, if you had
(33:08):
to give me a summary of the book, what is its essence and
things like that. So that's kind of cool.
And everyone never would have been able to do that before.
Now I'm like sparring with the entire context of a book.
And then I that led to help me build a a flow diagram that that
thinks about the end customer's experience and how context
(33:29):
matters so much in shaping that.I really like the visualization
or just the concept of sparring with the AI when you're not just
going a back and forth to do AQ and A, but you're trying to get
the critical thinking out of it.Acknowledge blind spots, point
of gaps. Have you ever used AI in either
like a a role-playing capacity or something to help you prepare
for a difficult conversation where you're going to say assume
(33:50):
the role of X, which I have to deal with shortly.
Help me prepare for this. Sort of so, so today actually,
and it's for a, it's for a project that I can't disclose,
but nonetheless, it's trying to,yeah.
(34:10):
It's trying to use synthetic, let's say digital twins or
profiles of roles and trying to see if you can have
conversations with these almost simulated versions of like you
and I or somebody who, let's saya podcaster, you and I both do
podcasts. So I have a method of building
(34:33):
psychographic profiles synthetically and they're quite
rich, very rich actually, actually.
And then if I was to build a psychographic profile of a
podcaster, in theory you have a mini model, mental model or a
mini model of this roles, somewhat of this role's
(34:54):
identity. So I just threw it into Gemini
this morning. It wasn't a podcaster, but it
was a different role. And I, and I did what you said.
So I got it to play out a scenario because I'm actually
rapid prototyping something before we get into a a much
larger flow of it. It's like, I just want to see
how well it can role play with me if I gave it a body of
(35:17):
context and I said, you need to role play as a teacher and the
person you're going to be talking to is this identity,
let's just say podcaster. But that's a very rich
description of what a podcaster is, not just surface level
stuff. Back to the the context map that
I have where we say, you know, what is a podcaster ultimately
(35:38):
trying to become, right? So there's a bunch of
psychology, psychology around why we all do what we do.
And there's a bunch of stuff that you do, Mike, you know,
using your tools that you do, you know, scheduling to get
somebody set up to a call, all of these micro interactions that
finally lead to you eventually publishing it.
All of that wrapped up into a ina nice context model and then
(36:01):
spar or go back and forth with an LALLM to say, OK, hold this
in context. I got it to role play as a
teacher. And I want you to assume you're
talking to me, a podcaster. I am.
But in this scenario that I did this morning, it wasn't a
podcaster. And I'm like, OK, let's see if
it had. It can answer a question that I
(36:23):
had. And then I had the actual
contextual profile, psychological profile of this
role in a very specific thing about this role.
And then I asked the LLMA question to see if it can find
the needle in the haystack, which is like, hey, what are my
interests in this? And it should go through that
context that we gave it now before it, they never could do
(36:47):
this. They're all doing it now and it
got really good. It was like, and it caught
exactly what I was looking for. I'm like, OK, so I could
actually just spar with it in a chat cycle right now and it
would work probably really, really well.
One thing I'd kind of like to just pivot to you slightly is
the the idea of your contextual intelligence they cover in the
book is there. Have you found there's been an
(37:07):
influence on the way you use AI based on that?
Yeah, I think it's all. It's all I think about it.
You can't you can't get away with it Once you see Jeff Jonas.
Jeff Jonas was doing this back in the day in the way way in
just the machine learning days before large language models.
He has a great video of it. Maybe I'll share a link with you
(37:29):
later that you can that you can share with people.
And Jeff Jonas was where I really latched onto the initial
idea of how powerful it is. He has a YouTube video where,
but because he's so good at analogies too, he's like, look,
I have this puzzle and I laid out on the table at my house and
he's got a bunch of students that are working as a team to
(37:51):
put the puzzle together and theydon't know that I've actually
put some. He bought 2 puzzles.
He doesn't tell them this, but he puts duplicates in there and
he does some other crazy tricks to see how these people respond
to having duplicate information.So he's, he solved the entity
resolution problem back, way back.
He was the chief scientist at IBM before developing his own
(38:16):
venture. But he had described in one of
his videos that so that puzzle metaphor, he says.
Well, in machine learning we started doing that and we
realized when we're trying to catch bad guys that bad actors
or in large data sets, big data,there was a problem in computing
with entity resolution and it's called the entity resolution
(38:39):
problem. A computer cannot distinguish
between some of these things. So he started building models
for them to to understand context around entities so that
it can resolve these issues. So if there's two like birds in
the system, is there a junior and a senior, sometimes father's
name, their sons the exact same name, but their postal code is
(39:00):
different or their number, theirphone number is a bit different.
Is that because you accidentallyput in the wrong number?
Or is it because literally there's a senior and a junior
and it's just not explicitly describing the data?
So this wreaks havoc on big datasystems.
But anyways, he ended up developing models to, to give
these machine learning algorithms the ability to have
(39:23):
to start to form contextual nuance around these entity
resolution problems. And then learned that the more
context you could give these systems, they actually
computationally can get faster the more context they give them
because they're anyway. So, so in this video, he shows
that is, it's a very human story.
(39:43):
It's beautiful because it's justa bunch of people putting a
puzzle piece together. And then, you know, it's, it's
intuitive for us to think that most people start with a
boundary space around a puzzle and you build and you work your
way in. And when you're doing that, when
you start, you're really slow because you're just putting the
puzzle together. And then as you get more space
(40:06):
coming in on the puzzle, it's get, you get faster to do the
middle part of the puzzle than when you start it.
So that's a really cool phenomenon when you're like, OK,
everything's going to be slow atthe beginning.
It has no context. But as soon as you give it some,
the next Rep or the next cycle of trying to understand what
that next piece is, is a little bit faster.
(40:27):
So it's, it's just really beautiful pattern that you know,
the more context I have about something, the more
understanding I have and the actual the, the, it's really
what the essence is. You understand it more.
And then it happens to be usefulin computers because you can
increase the speed of which you need to go learn something.
I knew this before ChatGPT actually, but I didn't, I didn't
(40:50):
get too far down on the computational side.
I knew as a UX practitioner thatwhen I was solving problems, I
just needed to get more context so that I could take that back
to a team or or an executive. The reason why we need to look
here or do this or do that is because of this.
I was the context getter. Like I was like, go out and
(41:12):
figure out why we're doing what we're doing where people are
struggling. So I just made my job better.
I was like, if I have all the rich context, it's not biased
either. That helps settle settle a lot
of problems internally at companies is just don't use
your, you know, go get observational data and bring it
back and say, look, this isn't me.
This is the user or customer. Boom, you're not fighting
anymore. So Fast forward into then LLMS.
(41:38):
I just knew I used that pattern in my work and I was like, OK,
well, right away sparring with aeyes, I would give it large
context. And then I start to realize they
struggled because I'm trying to write a book and I'm like, OK
here, you know, I would do everything on my own first,
right? I'd have mega long documents and
(42:00):
then I would try to get it to help me make it sound better.
And it would really struggle. It would start to miss stuff.
And I was like, OK with these things are limited.
And I realized, well, I started doing research on that back in,
you know, the very early versions, they struggled with
large context. So I had to work within that
constraint. And then but once I knew it, I
was fine. I was like, oh, it can't take
(42:21):
six of my word files. It can only take one.
If I give it 6, it misses page 234 and it just gives me one and
six. Which was a funny pattern
because I seen it in UX design as well.
And it's called the peak end rule or peak and end rule.
And we know that when you're designing experiences, we humans
(42:44):
have a memory to recall things that are more of a peak or an
exciting, exciting moment of an experience or the ending of it.
They knew this in doctor's offices when they were trying to
figure out how to improve the patient experience.
You could have a long ass surgery, but you know, the
memory of the interaction pointsof that whole experiences,
(43:08):
whatever was painful and an exciting moment and then the
very end. So anyways, it's a funny
phenomena that this is in humansas well.
You don't remember everything. You're not you can't right?
You're not supposed to. It's it's not efficient.
So but you do want to remember things that either helping you
not get killed and things like that become a better version of
(43:30):
yourself, whatever. So I had applied context now
into writing the book, right with my sparring and
understanding the limitations ofAI and starting to go, OK,
where, where does this thing start to struggle and where does
it not? How much context can I give it?
How much can I not give it? And then I started figuring out,
(43:54):
well, I struggled with that actually quite a bit until I had
tried to hire an AI engineer from Toronto, MJET.
He's in the book as well. And I told him, I said, look,
man, I just want to give. I just want to give this thing
all my documents and help it help me write, you know?
And he's like, yeah, you can't do that.
He's like, it just can't. So what you need to do is chunk
(44:16):
it. Like, chunk it.
What the hell is that? He's like, oh, yeah, well, this
is how you deal with the contextproblem.
And so he gave me tricks right away to start using to reason
better with these or spar betterwith them.
Was like, I really started to feel the edges of what was
possible back. This is over a year and a half
ago. I think so.
(44:37):
So I think that answers your question.
So I was like realizing that I used to use these the power of
what context could do in my UX practitioner role.
And then you just hit that. Now when when I was trying to
reason with a is, I was like, they they won't take the entire
puzzle even if I give it to them.
And then I had to get into chunking and figuring out even
(44:59):
there you, you start to hit a problem where it chunked, but it
the way, the way that it went from 1 chunk to the next, it's
sometimes, well, there's ways around this before, but it'll
lose continuity between Page 1 and page 2.
And I was like, how do I deal with that?
(45:21):
So yeah, just, I guess just playing around with them and
then understanding context from my old work LED into how I would
find the boundary of or the constraints of working with
them. Yeah.
And I think it's a very important practice for everyone
to continually do where they're pushing the boundaries and see
what the limitations are, because it changes all the time.
Every model has different limitations, and the only way
(45:42):
you can really understand what the state is at this current
moment is by doing yourself. Last question for me, a lot of
this has been interacting, like communicating with the computer
or having the program interact with the user.
But it's very important to maintain human to human
communication, especially when we're all like locked away in
our offices doing our own thing.So as someone who's organized AI
agents Waterloo, which I love asa as a group, do you have any
(46:02):
advice for people to bring community together, organize
events, even things that Co working spaces to help promote
interperson communication? Any ideas to help make sure that
we keep the human aspect in thisindustry?
I love that because I think if anything, I know there's a lot
of fear around AI right right now, but I actually believe it's
going to reverse what mobile andsocial systems did and and
(46:25):
actually help us go be more human again.
But for tips around people that want more of that experience.
Yeah is well, find a builders club if you don't have one
around you, start a meet up group or run an event somewhere.
And I used to do that before my agents of group.
(46:46):
I ran IoT Waterloo and I would just go find any space that had
a bar or whatever and I would say, hey, I will bring people.
We just need this section of the, the meeting row.
The if they had an area, right, I'd find a space that had a
conference room or something. And I'd be like, look, I'll
(47:07):
bring in 50 people, 20 people and we're just going to talk
about this evolving technology. At that time, it was about
networked objects and you make money on the bar.
So my, my feedback to anybody who wants, you know, more human
to human interaction is, is go do it because waiting for
somebody else to do it, then you'll be waiting forever.
(47:29):
So if it doesn't already exist in your town or wherever you're
from, go do it. If that's the thing that you
really want and it's not there, then make it happen.
And even if it's five people or 10 people, I think you just
changed the format. So when you came and talked at
our event, it was, it was amazing.
A different, different structure, right?
(47:50):
Different vibe. People are there for different
reasons, but this guy Deb, who'spart of the barn here, ran a
very different format and he hadI think 1010 people, but it was
white boarding and just everybody talking about the the
agentic systems they're building, where they're
struggling, what's working, whatisn't.
(48:10):
And that was a really great interactive experience that so
much so I can't wait till he does his next one.
He calls his AI agents builders.I think mine was very casual
because I'm just doing what I did back in the when I ran AI,
sorry, IoT Waterloo, I was like,well, who out around me is doing
(48:31):
this? And then I just want to be
around them and talk to them andhear stories, right?
It's it's very high level, but he's right into the weeds.
So he's like, let's whiteboard to how are you addressing the
problem? Does anybody else in here have a
solution to that? Really, really cool.
And he does live stream data stuff.
(48:51):
So he's trying to figure out howtheir organization plays in the
future with live stream data, right.
So that's, that's a great way for people just to be
participants in, in this, this cycle because it's all real
time. Everybody's learning this real
time. All right, What a better time.
There's no better time to get out there and just participate
(49:14):
either attend events that are are actively sharing and
distributing the information. And if that's not happening,
then start one your own. It's never been easier, right?
You get on Luma and you just go,I'm going to create an event and
and don't sweat it if it's like only 5 people show up to because
those are the people you want totalk to anyways.
Like right. Whiteboard.
Whiteboard. Exactly.
(49:34):
Yeah. Whiteboard topic, only other
group I'd shout out as AI tinkerers.
I I love their events and they kind of say no investors, no
business, no suits, just people that make demos.
Yeah. And and that's a great, that's a
great play if that's your play. Like I love investors.
I love lots of different perspectives.
So, you know, and I think that'sthe, the beauty of having these
(49:56):
things because they end up becoming their own little unique
tribes. And, and if that's your, your,
your shtick, which is like no VC, no, you know, this, that
that's cool. But I think just naturally for
me, when I ran IoT Waterloo, I had a lot of business people
that were involved. It wasn't just tech.
So I try to find at least just for mine, the balance of like,
(50:16):
you know, it's, it's, it's high level.
And then storytelling can sort of help a business person see
how that might affect their their business model because
things are changing or if it's atech person.
Ian, this was a blast. Before we let you go, is there
anything you want the audience to know?
I yeah, so I guess the one thingI'm trying to get better at is
self promotion. I don't typically do this at
(50:37):
all, but it's one of the things that I need to get better on.
But the the book is not so much the thing I want to push the the
product I'm building or service.Product and service is trying to
figure out if there is a demand for people who want to become
leaders. So we are about 3 weeks away
from a closed beta release. And if anybody's interested,
(51:02):
that is an audience member of your show wants to learn how to
measure if they are becoming a better version of themself or
you know, what it actually meansto be become a leader.
And they were struggling like I was, which is like, I don't even
know. We, we've developed a way to use
to measure qualitative data. It's like a mirror and it's
(51:24):
private by design. We're looking for beta customers
to sort of like just kick it around and try and learn if it's
a if it's useful or if it's justsomething I'm going to use
myself. I don't know.
So I think it has value. But if you know, I won't know.
So if anybody in your audience is interested in, that's
something I'd be open to an introduction or a conversation
and let them try it out. Perfect.
(51:44):
Yeah, we'll put all the information down below.
Ian, this was a pleasure. We'll talk to you soon.
Awesome. Thank you so much, Mike.
That epiphany or the aha moment that you might not be aware of
this, but there's a subconsciousbehavior that you have and it's
not serving you or the others around you.
But I think with the right technology stack you can really
do some interesting stuff here. Welcome to episode 43 of
Toolies, the weekly conversationabout AI tools and strategies
brought to you by Anetic. I'm Mike Bird, and today we're
(00:20):
talking about how to use AI to be a better communicator.
We're joined by Ian Pelon, an innovator who sits at the
intersection of AI, human centered design and leadership.
He organizes AI agents. Waterloo and Ian's book
Cultivating Clarity is a must read for leaders and product
innovators. Ian, welcome to to use.
So happy to be here, Mike. Thank you.
Yeah, I'm, I'm pumped. And it was such a pleasure to
(00:41):
have you at one of our events. I didn't get to talk to you much
about the book, so maybe we can cover some ground on that.
But yeah, for those that don't know me, I'm in Waterloo ON
Canada, an active builder, playing around AI technology and
just having a lot of fun. So yeah, thanks for having me.
Mike, absolutely pleasure to have you here.
And that was a great event. I really hope more people in the
Waterloo Region kind of start aggregating together rather than
(01:03):
just all migrating down to Silicon Valley because
Waterloo's got the clout, but weneed to make sure we got that
home grown talent going. Would you mind giving us a
little bit of your background? Yeah.
So in the last decade I've been into UX research and design and
then over the last couple years moving really heavily on the
product side around AI. But before that, logistics,
(01:26):
which is kind of an interesting thing that's sort of resurfaced
in my ability to see patterns related to using AI and, and
flow design, especially around agentic systems.
And thinking about the relationships of, you know,
supply chains. Funny enough, in some of the
(01:46):
principles that I've, I had in logistics before I got into
technology, a lot of these ideasaround what I started seeing in
in tech were practices I've seenin logistics before I got into
technology. So that's always been cool
because I guess Japanese philosophy, I used to work for a
company that supported Toyota supply chain and they all
(02:09):
applied lean methodologies and use things like Kaizen and all
these Japanese ideas around efficient systems design, right.
Just in time, logistics, all this stuff when I ended up
getting into technology was, wassurprisingly relevant or, or
just familiar. Sorry, because I was like, well,
(02:31):
I've seen this over here in, in,in logistics.
But then it's funny that some people in the tech side of the
world, and I guess in software tech weren't familiar of where
the origins of some of these ideas came from.
So, so it was kind of cool. It was familiar to go from the
physical world of logistics intosoftware design and seeing a lot
(02:55):
of similarities and trying to just have work flows that are,
are products that are helping somebody achieve something
somewhere in a supply chain. So I love the mental models that
I took from logistics and, and sort of the, the systems of
efficiency in that world. And, and, and realized, you
know, in the UX, if you understand the flow and what
(03:16):
people are trying to do, it's very transferable into building
actual products and services that people use versus not use.
So yeah, UX has been a huge influence into then bridging
into AI, the AI world now and why I'm sort of obsessed about
the customer relationship between the technology shift and
(03:40):
use users using tools or customers to getting jobs done.
It's just a really fascinating intersection, right?
That it doesn't go away. The fact that we're tool use
humans and just happens to be the tools that we're creating or
getting really, really powerful.So that excites me and I go,
well, how's that going to, how'sthat going to affect me?
(04:01):
How's that going to affect the work I do, the people I work
with, and then obviously my children.
So I'm very particular about sort of trying to anticipate how
this new thing might be used in the future and start to learn
more about it. I love learning.
So I go all in. And then if there's like, if
it's something currently like AIand agents, which is so new,
(04:22):
like you're not going to learn about that in school or in
college and university because some of the stuff is just
bleeding edge, right? You're not going to learn about
memory and agentic flows. It's just not in the curriculum
yet. Maybe for some real time
schooling stuff, but that's where throwing events is like,
hey, who else is sort of at the bleeding edge of where things
(04:42):
are going like yourself and and then going, how are you using
these things? What are you, you know, what's
your perspective on it? And that helps me just learn by
osmosis. Absolutely.
And it's one common theme that keeps coming up where people
with different types of backgrounds are kind of
migrating to the AI world, Bringthose different experiences with
them. Last week we had on Michael
Tiffany and he brought up how the concept of a desktop
(05:04):
directories being folders and files on your desktop is just
representation of how people used to work.
But now you're getting people entering the workforce.
And like you said, there's there's no historic knowledge of
AI ages. This is a brand new thing and
they're bringing all these different mental models into it
and being able to solve problemsin unique ways.
And everyone is starting to be able to be enabled with this
technology and have their outputamplified that.
(05:25):
It's getting really exciting just seeing a whole new paradigm
being formed rather than just the next gradual iteration.
So 1 area that I'd love to hear some thoughts on from you are
just the UX of current AI tools.The chatbot is everywhere.
There's always a little empty text field that you go to in
order to type and kind of createthis engagement.
But where do you see either the current state or the future
projection of how humans are going to interact with AI?
(05:47):
Yeah. The to answer the question about
where do I think UX is going or where it is it right now, I
think we can't go without talking about the GPT moment.
Like anybody who is a user that got onto Chachi BT and first had
their interaction experience with LLMS, There's no question
it was, it was pivotal. And in doing that for, for
(06:11):
whatever somebody's experience was with that type of
interaction. To me, my, my like aha moment
was that I could finally use English to talk to a computer.
I didn't have to know code. And that's been always sort of
like this friction of me gettingthe output from machines that
(06:32):
I've wanted. So in doing so, when I had my
ChatGPT moment, it was really, Ithought it was a revolutionary
change. It was like, Oh my God, I've
been waiting for this forever. I can finally talk to a computer
and have it talk back to me. And then and then even then
really start to push it to go, well, how, how far can I go with
this? Can I get it to build an app?
(06:54):
And right away I was like, like trying to go, OK, well, I don't
understand Python, but I understand some principles about
what an application is a recipe.So can I get it to do something
basic local on my machine and then see if I could really be
communicating with the computer right?
So now it was like, OK, can you build me something?
(07:16):
OK, so, so then Fast forward into the UX paradigm of, of what
I started using it with for myself personally.
First was all right, I now have a translator real time that I
can talk with a computer and getit to build something for me.
I've always wanted that. And then it showed up.
So in around that time, I was also super interested in
(07:39):
thinking about, well, this is a,this is a massive shift in tool
use because I don't have to learn sort of like Python.
I mean, other people can. And this does not discredit or
take away people that have learned the hard way.
It's, it's, but for me, I did never, I never learned how to
(07:59):
write or learn those language, how to write them to talk to a
computer. So now I had a translator.
Basically, that's how I think about it.
My perspective is that I have a translator.
I can not only ask it to go communicate, I don't actually
care what language it uses. I care about the outcome that
I'm trying to get. So an example might be, I mean,
(08:23):
I'm also into sound design or orplaying with music.
So can I write a simple app thatuses the microphone on my
computer to pick up my acoustic environment and give me a
visual, visual output? And sure enough, you can one
shot these things and I was justlike, Oh my God, this is crazy
because you know that hello world feedback used to take
forever. Even when I was in web design to
(08:46):
build a web page just took took a bit of work to get a hello
world up. Now you're talking about like
instantly, which is absolutely groundbreaking.
So that excite me led me to going a little bit to think
about right away the future. And to answer your question,
where does UX go? Well, I realized that if anybody
(09:06):
can build instantly, now there'sa whole world of how we used to
work that gets that just changesbecause now, like if you think
that we're building in software often tools for other people
right away, I'm like, well, people will just build the tools
(09:27):
themselves. If you can speak English.
I mean, you don't, you know, youknow, you can now solve a lot of
your own problems. Now things are moving really
fast. But I was kind of like this
might make my role in what I do change.
So how do I stay relevant in thefuture if what I used to do is
help inform engineers on on why we're building tools?
(09:49):
Everybody is on some journey. And in UX, what I used to do is
understand where they fall in this journey.
So the tools that we built are sometimes helping them solve
tasks that are in the milliseconds, seconds or minutes
sort of time scale. I'm bored.
I'm going to go on Twitter. I'm no longer bored, right?
(10:12):
The job to be done there is verysmall.
It's minute, but it is a thing. And if you don't fulfill it,
they're going to do it doing something else.
So that's where you can insert certain things like social
media, but as ultimately you go up to more complex products and
services, you're getting into typically things where people
(10:34):
are trying to get tasks done. Sometimes it can go from minutes
to hours to days to weeks, basically a bit of a gradient
here, but you're in a sense making zone.
And this is where a lot of B to B plays are and stuff that it's
like, no, I'm using this, this tool is helping me move forward
on some project. And it's a much bigger thing
(10:56):
that you need to do. And in this range, you often
have complex systems and there'sa lot of engineering and product
involved and you end up buildingcool products and services.
But this now hyper shrinks down,right?
Because it just used to be, there was no other way to do it.
Before you had to hire a UX practitioner.
They would do a bunch of research and then they would
(11:16):
come back and talk to engineers and build a prototype and then
put it in front of people and then wait for feedback.
And you know, this, that that could take weeks, but
nonetheless, the user's goals and needs don't change.
We're all struggling to make progress towards goals, whatever
those might be. They're just different for
everybody. But we always have to take into
consideration of the things thatwe're trying to build.
(11:38):
Who's it serving? It's not for ourselves.
Sometimes it can be. But I just happened to be doing
this for a long time. So I got really good at
understanding on going to do thediscovery research to figure out
what people are trying to do. Is it, you know, is it a short
term interaction experience? Well, then that's what I call in
the book is low customer contextand and in, you know, thinking
(12:04):
about how I was going to play inthe future.
I also have to think about like how people are interacting with
tools. The the human computer
interaction paradigm is just evolving just like language.
So I think you mentioned the, the story about like the the
first stages of computers being,you know, a mental model of, you
(12:25):
know, folders and trash cans. You, you worded it a bit
differently, but it's basically like you, we gave users a mental
model to get away from the, let's say the command line
interface where, you know, all of a sudden when somebody could
look at a screen and have a visual metaphor that, oh, if I
want to delete a file, I can drag it to a recycling bin.
(12:46):
That was very powerful to introduce this technology to a
whole world of new users that would have never been able to do
write some code to get it to do that.
So that democratization is part of the interaction of how humans
are using the tools. And now we're going to shift
into the ability for people to speak plain language to solve
(13:09):
some of their problems, which isinteresting.
We're all at the very cusp of this happening, but ultimately I
think this this goes to where we're all striving to become
other versions of ourselves, even if it's subconscious.
So some people are, they're veryforward-looking and they do have
goals and they know who they're trying to become.
(13:30):
Other people. It's just subconscious.
They actually don't know it. But regardless, they're going
towards becoming something else,whether you're conscious about
it or not. So if you can sort of think
about the spectrum of humans moving through time, there's
tools in that flow. And there's no question that
(13:51):
software is just hyper embedded in everything we do now.
You can't get out in the world without being in the middle of a
software experience. So those are all like hyper
micro interactions. You're on Twitter to Oh my God,
I'm at work and I've got all allkinds of products and tools that
help me get my job done so that I can get paid.
But at the end of the day, we'reall striving to become some
(14:13):
other version of ourselves. And and so on that scale, I like
to think about context as how I think about the potential change
of experiences between humans and computers.
And, and that's why I wrote the book because I was like, well, I
(14:35):
think I know what context is, but I actually want to dig a bit
deeper. When that ChatGPT moment hit, I
was realizing that it was a paradigm shift.
All of a sudden I seen a old world where there was all this
work and activity into shaping somebody else's experience.
And all of a sudden it gets democratized.
And now a lot of that work is not needed.
And you can just not in all scenarios, but in some you could
(14:58):
just clearly see that, OK, there's not going to be a need
for that. They can do it themselves.
So I'm excited about this idea that it's it's going to change.
And do you want to participate in the change of how tools are
used for humans? So you can sit on the sidelines
and not do anything or you can actively participate in trying
(15:21):
to imagine and play that. I think there's a word in an era
of play. Same like sort of like the
mother of all demos. Do you know that that demo, the
human interaction demo back fromwhatever the 60s and that was
when like the first experience of the mouse was sort of
portrayed to the world. Those ideas are we're we're at
(15:41):
that now, I think with AI. So to be honest, I don't think
there's a solid answer on like where do I think the UX is
going? I just know, and you're right,
chat was this first thing, but it's kind of wild.
I started playing around with these ideas about like, OK, can
I have generative UI respond to content?
(16:01):
So now I'll take content and sort of like, hey, what does
this look like as an animation? What does this look like as a
slide deck? What is it?
So I think you're going to get into almost a hyper personalized
way that each of us wants to getthe information back as a tool
(16:22):
in, in the medium that we we love.
Like, so I love watching YouTubevideos to learn stuff other
people don't they, they might love to just read and go deep on
a white paper. Those two extreme differences in
the media is, is, is huge. And if somebody has the option
now to switch, you know, a toggle or even not even switch a
(16:46):
toggle, talk to a system. And if it learns your
preferences over time, it can start serving you the, the
content, not only in a paper or a YouTube video or generate a
movie for you, put you in the movie, which was kind of really,
really cool. So it's, it's just it's, it's,
(17:07):
it's groundbreaking. Everything's changing in real
time. Yeah.
And I mean, it's crazy how we'reseeing this, this rapid
evolution where now I interact with my computer mostly with
voice. I'm always prompting with voice.
I know that's not for everybody.I work from home, I don't have
to worry about other coworkers and it just allows me an
increased bandwidth of input. And Carpathy actually came up
with a tweet recently about how high bandwidth video consumption
(17:29):
is and how you can really get more granular information.
And kind of, you're allude to some people prefer reading, but
what if in that reading instead of just static diagrams, you had
an animation? And what if you could interact
with in bidirectional communication to ask you
questions? The ability for us to really get
a tailored experience to exactlythe output, the outcome that we
want that you're going to is just getting closer and closer.
VO 3 came out and all of a sudden you can make these
(17:50):
incredibly realistic videos. As soon as that applies in like
almost instant dynamic wait, we're really going to be able to
learn and just accelerate our interaction so much faster.
But what one thing I want to pull it back to when we met you,
you told me this cool story about how a lot of the times you
can make improvements based on certain metrics.
But as soon as you get into a leadership role, what are those
metrics? How do you make things better in
(18:12):
a measurable way? Would you mind talking with
that? And just how how you're able to
work with communication in the leadership aspect.
So being a UX practitioner for so long, I had gotten really
good at finding friction in process.
And this happened way before I got to tech in logistics, as I
mentioned, I used to time, you know, when we would move certain
(18:35):
freight to a different lane and if you would shave off even
seconds on some of these work flows, you were saving a lot of
money because it was all hyper connected to other things in the
in the supply chain. You know, save a second here 300
and 65360 days a year. It it means that this truck can
get in here faster and get out there faster.
(18:55):
It all adds up. So, you know, thinking about
that in the context of software was easy for me because I was
like, OK, well, just the, this interaction point is where
there's, there's friction. How to, what are some ideas that
we can solve it? But I think I, I got really good
(19:18):
at go finding the friction. That's what I'm really good at.
And then from there, you can have multiple perspectives on
how to solve that problem, whichis always good.
Whether you're solo or on a team.
You can, you can say, do we all agree that this is the problem?
And then, and then you're set upfor success to deliver a
solution. When I got into managing people,
(19:41):
then I had no way to measure my this, this idea about, well, can
you lead a high performers or a high performance team?
All of a sudden you're in this fuzzy land about what, what
constitutes performance is that,you know, how many updates go
out, how many push updates are sent out.
(20:04):
So it's just really tricky way of trying to describe even,
well, you, you start to get intothe ambiguous terms, right?
So you're like, well, what does that mean?
Rather than the, you know, an abandoned shopping cart
experience, It's very measurable.
We're not getting people to get out through the checkout
experience in a, in a software. It's very a binary.
(20:25):
In the human world, it's, it's psychology and you've got a lot
of fuzzy logic in there. So when I was leading A-Team, I
wasn't sure how it was performing.
I was kind of like, I don't know, they won't tell me, right.
So you have this power dynamic as well between manager and
subordinate relationships. And, and I think we know this
(20:46):
subconsciously, but it's not actively talked about.
But I was curious. I love self development.
I read a lot of books and at that time I was working on a
project and there was an executive coach from Vancouver
who I I, I expressed this to, I said, yeah, Daryl, man, like
I've been leading this team for a year.
I have no idea how I'm doing. And that was starting to really
(21:10):
gnaw at me. It was kind of like, I don't
know, right? And, and, and that, that was,
you know, I was a fish out of water, an experience that I was
not normally used to. And he said, well, you got to
read this one book. It was called Multipliers.
And he said, you don't even haveto read the whole book, Ian just
(21:30):
read 1 chapter about accidental,accidental diminishing behavior.
So I hadn't even finished this call.
I was on this virtual call with him because he's in Vancouver
and I had already checked out the shopping cart in Amazon.
The book was on the, you know, hit buy.
I love that I could do that. I was like when you, when
(21:50):
somebody inspires you that you were like, man, this guy, I
really trust his really value his opinion.
If somebody tells me like, you know, like that, you should read
this. Those are the books I read,
right? I used to buy whatever interests
me. Now it's like if there's people
that I really follow and I'm like, that person is somebody I
admire and they wreck make a recommendation.
(22:12):
It's like I will take take it seriously.
So I did and, and I read it, youknow, cuz it showed up at my
house the next day. I read it and I had my light
bulb moment about trying to interpret this fuzzy logic, this
uncertainty that I had around, how do you know if you're a good
leader? Well, what does that even mean?
(22:34):
I had other books on leadership before, but this one was, was
really good because it said, well, all of a sudden you could
classify what a bad leader lookslike, right?
People that diminish others and they're not even aware they're
doing it. And it's a subconscious thing.
You, you, you don't always know that you're doing it.
Plus you have a lot of multiplying behaviors and people
(22:59):
don't even don't know what thoseare.
So you want to keep doing those and not doing the other thing.
So I had a couple of couple of them that jumped out at me like
the idea guy, you know, so if you're leading A-Team and
they're trying to share their ideas with you and you're like,
you have a, oh, I got this idea too.
And you're, it depends on the person you're leading or
managing, But they're all, everybody's unique, right?
(23:22):
Depending on the state of your career, you want more autonomy
or less or more agency or less. Some people need more help on
the support side. Some people want like, just
leave me alone, get it on my way.
And you have to figure that stuff out.
So I was able to read the book, get the epiphanies like, oh, I
might be doing that. I might be doing this.
(23:43):
All these other ones. I definitely know I don't do
that. So it was really cool.
And then from there, this was also the time that catchy PT had
hit. So I was like, I was wondering
about these new interaction paradigm, Speaking of, you know,
how UX might change. And I started thinking about
(24:04):
recording myself to try and see if it could catch these
diminishing behaviours. Now this is obviously a very
sensitive area and rightfully so.
I'm super big on privacy by design.
So I tried to to mock up these ideas at at home, working with
the local tools to see if it waspossible to record a
(24:30):
transcription. I have one with my my daughters
and try to get the the LLM to make inferences almost in real
time on the communication between two parties to figure
out if you could have enough context to detect a diminishing
behavior in myself. So think of it like a mirror.
(24:50):
That's what I was trying to build originally was this mirror
and to see if I could catch diminishing behavior.
So I had prototyped that and I think I have AI do have a post
on LinkedIn and Twitter a while ago when I, when I had it
realized that Oh my God, LLMS can do this given the right
(25:12):
context, they can do a lot of stuff.
So I was like, well, wow, this is really, really cool.
So I could just have this self monitoring myself.
And I remember also telling, I think my brother was like, like,
'cause him as well as other people will know that if I go in
your house and you have Amazon on or you like Google, you have
(25:35):
these listening devices. I'm like, they're all listening
and I have nothing that is sharing that data back to me.
So I I say things like Amazon and Google know me better than I
know myself. And I'm like, well, I want to
build a private by design personal sidekick that can
monitor myself and and then see if that could be useful, right?
(26:00):
Could I, could I do what the book did, which was give me that
epiphany or the aha moment that you might not be aware of this,
but there's a subconscious behavior that you have and it's
not serving you or the others around you.
So as I, as I mentioned, a bit of a sensitive area, but I think
with the right technology stack,you can really do some
(26:22):
interesting stuff here, right? Without going too deep into
that. I mean, there, there is PII
scrubbing tech that you can use zero knowledge proofs, all kinds
of weird things that people are doing to keep data private.
Now also because I was in UX forso long, I just got used to
(26:42):
whenever I have a customer discovery call, I would ask, are
you OK if I record this? And if they say no, I'm like,
OK, cool. And I'm back to the old school
scratch and notes and trying to remember everything.
But for most people there were, there were like, yeah, cool.
As long as I tell them what the data is being used for, right?
This data will not be used outside of this company or so on
(27:04):
and so forth. So that's what I'm doing with
Nero Kick or the leadership development is if I'm not
leading A-Team right now. But if I was, it'd be like, hey,
I've got a sidekick or a coach and it's a mirroring device.
It's not, you know, listening toyou, but it can delete the data
instantly. Do you want me to turn this on,
(27:26):
On or off, Yes or no? I'm trying to become a better
leader managing you and I don't know what I don't know.
So does that answer your question?
That's sort of the the technology that I'm playing with
right now and having a lot of fun because it's forcing me to
learn what the limitations of are, what the limitations are of
using these new technologies right there.
(27:49):
And it gives me a chance to playwith them at the same time.
Yeah. And I think that's a really
interesting approach because when we're just unconsciously
having a conversation or just, or just, you know, engaging a
regular life, we're not always being analytical about
ourselves. So having a device that's
actually like being introspective and trying to
analyze what you're doing does offer a lot of real insights.
(28:09):
I'm curious about for the more active mindset.
If you are putting together an e-mail or some type of external
communication, a lot of people will resort to using AI to
either help with the blank page problem or kind of proofread.
To summarize, have you found success using AI to help with
your direct communications, or does it always end up being slop
and you have to rewrite it anyways?
Well, that's a good question. I think there's two things in
(28:30):
there. The blank page problem is just
that's a problem on its own. That's just beautiful that LLM
solve for you was like, and I think I get this from remix
culture and music. It's always good to have
something to jump off rather than starting with the blank
slate. Inspiration doesn't always come
to you instantly. So if you're in a creative role
and somebody's like, OK, I need this design tomorrow, you may
(28:53):
not be inspired. So that's when you're often then
going what we're looking at other things to find inspiration
in. So yeah, the jump off point for
the blank start problem is absolutely huge.
I think everybody's found utility in that as far as using
it to correct your, your writingor not even correct it help
(29:17):
augment your your writing. I used it instantly to write my
book. And so I'm not a writer and I
was looking at hiring a ghost writer.
So in my book, Bob Mesta was oneof the people I interviewed in
my book. And what a what a, what a gift
to, to talk to Bob Mesta. And he told me he had hired a
(29:41):
ghost writer for one of his, told me the price.
And I'm like, yeah, that is justnot in my, my wheelhouse.
So I'm going to use AI and I have no, you know, no shames
about it. I was like, I I'm not a writer,
but I will figure out I do what I do is I, I went and grabbed
all the story arcing frameworks and plots there are known to
(30:02):
man. And then I was like, OK, if I
have 9 interviews, this is what was in my book.
I want to get different perspectives.
And we're talking about context in the framing of product,
developing products and services.
So I, I had a big mural board and I, I visualized the story
arc of how I could put this intoan overarching story.
(30:26):
So that's where I used AIS at that time.
I think it was Gemini. I wasn't ChatGPT, but I was
like, you know, OK, I've got these nine different interviews
and how do I have an arc and a story that cohesively puts it
together? And, you know, it's basically a
sparring partner. And while you could try in this
and you can, you can, you know, come out with a framework where
(30:50):
you can, you can go at it so many different ways.
So then I, you know, I use it asa sparring partner and go back
and forth until I landed on something that I was happy with,
not only with just AI, watched alot of videos on storytelling
and then and then from there, yeah.
So it was like, OK, I'm going tocome up with this crazy intro
(31:11):
and this hook and this is what I've got and sparred with the AI
back and forth. And then it's really back and
forth. So what about this?
No, I don't like this. And I and I fine tune it, right.
OK, cut this out. Don't like how it's sounding
like that. And you could you could at that
time, it was so obvious when youcould see like AIS would would
(31:32):
use the word delve into something over and over and
over, you know, But by doing this, I was able to see, I was
able to really find the some of the common patterns that I would
I was seeing in the LLM while working with it.
And then going, OK, as I keep going to write my book, if it
was a ghost writer, because evena human ghost writer is is not
(31:56):
going to be your voice, right. It's like it's trying the best
to it can be to interpret the story in your voice, but it's
still somebody else. So I just ran with that.
I was like, OK, well let me see what I can do a lot anything.
It took me 8 months even with with an AI and I don't even know
how long it takes with the humanghost writer.
But I had spent almost every evening going to the library and
(32:22):
just having space to, to spar with this thing back and forth.
And you know, I had my overarching map of where I think
I wanted to go with the story and then just brute force
whittled it down like, OK, got to get this first section of the
book to be the hook before it gets into the the interviews and
(32:42):
then have the ACT, the outro. Oh, and, and back to that map
that I sort of had as a screen share up.
That's that's that's I guess a visual way that I try to capture
the essence of the book. And so although that's not
writing all the by, by the time I was done the book, I was able
to go, OK, put the book into an LLM and go, you know, if you had
(33:08):
to give me a summary of the book, what is its essence and
things like that. So that's kind of cool.
And everyone never would have been able to do that before.
Now I'm like sparring with the entire context of a book.
And then I that led to help me build a a flow diagram that that
thinks about the end customer's experience and how context
(33:29):
matters so much in shaping that.I really like the visualization
or just the concept of sparring with the AI when you're not just
going a back and forth to do AQ and A, but you're trying to get
the critical thinking out of it.Acknowledge blind spots, point
of gaps. Have you ever used AI in either
like a a role-playing capacity or something to help you prepare
for a difficult conversation where you're going to say assume
(33:50):
the role of X, which I have to deal with shortly.
Help me prepare for this. Sort of so, so today actually,
and it's for a, it's for a project that I can't disclose,
but nonetheless, it's trying to,yeah.
(34:10):
It's trying to use synthetic, let's say digital twins or
profiles of roles and trying to see if you can have
conversations with these almost simulated versions of like you
and I or somebody who, let's saya podcaster, you and I both do
podcasts. So I have a method of building
(34:33):
psychographic profiles synthetically and they're quite
rich, very rich actually, actually.
And then if I was to build a psychographic profile of a
podcaster, in theory you have a mini model, mental model or a
mini model of this roles, somewhat of this role's
(34:54):
identity. So I just threw it into Gemini
this morning. It wasn't a podcaster, but it
was a different role. And I, and I did what you said.
So I got it to play out a scenario because I'm actually
rapid prototyping something before we get into a a much
larger flow of it. It's like, I just want to see
how well it can role play with me if I gave it a body of
(35:17):
context and I said, you need to role play as a teacher and the
person you're going to be talking to is this identity,
let's just say podcaster. But that's a very rich
description of what a podcaster is, not just surface level
stuff. Back to the the context map that
I have where we say, you know, what is a podcaster ultimately
(35:38):
trying to become, right? So there's a bunch of
psychology, psychology around why we all do what we do.
And there's a bunch of stuff that you do, Mike, you know,
using your tools that you do, you know, scheduling to get
somebody set up to a call, all of these micro interactions that
finally lead to you eventually publishing it.
All of that wrapped up into a ina nice context model and then
(36:01):
spar or go back and forth with an LALLM to say, OK, hold this
in context. I got it to role play as a
teacher. And I want you to assume you're
talking to me, a podcaster. I am.
But in this scenario that I did this morning, it wasn't a
podcaster. And I'm like, OK, let's see if
it had. It can answer a question that I
(36:23):
had. And then I had the actual
contextual profile, psychological profile of this
role in a very specific thing about this role.
And then I asked the LLMA question to see if it can find
the needle in the haystack, which is like, hey, what are my
interests in this? And it should go through that
context that we gave it now before it, they never could do
(36:47):
this. They're all doing it now and it
got really good. It was like, and it caught
exactly what I was looking for. I'm like, OK, so I could
actually just spar with it in a chat cycle right now and it
would work probably really, really well.
One thing I'd kind of like to just pivot to you slightly is
the the idea of your contextual intelligence they cover in the
book is there. Have you found there's been an
(37:07):
influence on the way you use AI based on that?
Yeah, I think it's all. It's all I think about it.
You can't you can't get away with it Once you see Jeff Jonas.
Jeff Jonas was doing this back in the day in the way way in
just the machine learning days before large language models.
He has a great video of it. Maybe I'll share a link with you
(37:29):
later that you can that you can share with people.
And Jeff Jonas was where I really latched onto the initial
idea of how powerful it is. He has a YouTube video where,
but because he's so good at analogies too, he's like, look,
I have this puzzle and I laid out on the table at my house and
he's got a bunch of students that are working as a team to
(37:51):
put the puzzle together and theydon't know that I've actually
put some. He bought 2 puzzles.
He doesn't tell them this, but he puts duplicates in there and
he does some other crazy tricks to see how these people respond
to having duplicate information.So he's, he solved the entity
resolution problem back, way back.
He was the chief scientist at IBM before developing his own
(38:16):
venture. But he had described in one of
his videos that so that puzzle metaphor, he says.
Well, in machine learning we started doing that and we
realized when we're trying to catch bad guys that bad actors
or in large data sets, big data,there was a problem in computing
with entity resolution and it's called the entity resolution
(38:39):
problem. A computer cannot distinguish
between some of these things. So he started building models
for them to to understand context around entities so that
it can resolve these issues. So if there's two like birds in
the system, is there a junior and a senior, sometimes father's
name, their sons the exact same name, but their postal code is
(39:00):
different or their number, theirphone number is a bit different.
Is that because you accidentallyput in the wrong number?
Or is it because literally there's a senior and a junior
and it's just not explicitly describing the data?
So this wreaks havoc on big datasystems.
But anyways, he ended up developing models to, to give
these machine learning algorithms the ability to have
(39:23):
to start to form contextual nuance around these entity
resolution problems. And then learned that the more
context you could give these systems, they actually
computationally can get faster the more context they give them
because they're anyway. So, so in this video, he shows
that is, it's a very human story.
(39:43):
It's beautiful because it's justa bunch of people putting a
puzzle piece together. And then, you know, it's, it's
intuitive for us to think that most people start with a
boundary space around a puzzle and you build and you work your
way in. And when you're doing that, when
you start, you're really slow because you're just putting the
puzzle together. And then as you get more space
(40:06):
coming in on the puzzle, it's get, you get faster to do the
middle part of the puzzle than when you start it.
So that's a really cool phenomenon when you're like, OK,
everything's going to be slow atthe beginning.
It has no context. But as soon as you give it some,
the next Rep or the next cycle of trying to understand what
that next piece is, is a little bit faster.
(40:27):
So it's, it's just really beautiful pattern that you know,
the more context I have about something, the more
understanding I have and the actual the, the, it's really
what the essence is. You understand it more.
And then it happens to be usefulin computers because you can
increase the speed of which you need to go learn something.
I knew this before ChatGPT actually, but I didn't, I didn't
(40:50):
get too far down on the computational side.
I knew as a UX practitioner thatwhen I was solving problems, I
just needed to get more context so that I could take that back
to a team or or an executive. The reason why we need to look
here or do this or do that is because of this.
I was the context getter. Like I was like, go out and
(41:12):
figure out why we're doing what we're doing where people are
struggling. So I just made my job better.
I was like, if I have all the rich context, it's not biased
either. That helps settle settle a lot
of problems internally at companies is just don't use
your, you know, go get observational data and bring it
back and say, look, this isn't me.
This is the user or customer. Boom, you're not fighting
anymore. So Fast forward into then LLMS.
(41:38):
I just knew I used that pattern in my work and I was like, OK,
well, right away sparring with aeyes, I would give it large
context. And then I start to realize they
struggled because I'm trying to write a book and I'm like, OK
here, you know, I would do everything on my own first,
right? I'd have mega long documents and
(42:00):
then I would try to get it to help me make it sound better.
And it would really struggle. It would start to miss stuff.
And I was like, OK with these things are limited.
And I realized, well, I started doing research on that back in,
you know, the very early versions, they struggled with
large context. So I had to work within that
constraint. And then but once I knew it, I
was fine. I was like, oh, it can't take
(42:21):
six of my word files. It can only take one.
If I give it 6, it misses page 234 and it just gives me one and
six. Which was a funny pattern
because I seen it in UX design as well.
And it's called the peak end rule or peak and end rule.
And we know that when you're designing experiences, we humans
(42:44):
have a memory to recall things that are more of a peak or an
exciting, exciting moment of an experience or the ending of it.
They knew this in doctor's offices when they were trying to
figure out how to improve the patient experience.
You could have a long ass surgery, but you know, the
memory of the interaction pointsof that whole experiences,
(43:08):
whatever was painful and an exciting moment and then the
very end. So anyways, it's a funny
phenomena that this is in humansas well.
You don't remember everything. You're not you can't right?
You're not supposed to. It's it's not efficient.
So but you do want to remember things that either helping you
not get killed and things like that become a better version of
(43:30):
yourself, whatever. So I had applied context now
into writing the book, right with my sparring and
understanding the limitations ofAI and starting to go, OK,
where, where does this thing start to struggle and where does
it not? How much context can I give it?
How much can I not give it? And then I started figuring out,
(43:54):
well, I struggled with that actually quite a bit until I had
tried to hire an AI engineer from Toronto, MJET.
He's in the book as well. And I told him, I said, look,
man, I just want to give. I just want to give this thing
all my documents and help it help me write, you know?
And he's like, yeah, you can't do that.
He's like, it just can't. So what you need to do is chunk
(44:16):
it. Like, chunk it.
What the hell is that? He's like, oh, yeah, well, this
is how you deal with the contextproblem.
And so he gave me tricks right away to start using to reason
better with these or spar betterwith them.
Was like, I really started to feel the edges of what was
possible back. This is over a year and a half
ago. I think so.
(44:37):
So I think that answers your question.
So I was like realizing that I used to use these the power of
what context could do in my UX practitioner role.
And then you just hit that. Now when when I was trying to
reason with a is, I was like, they they won't take the entire
puzzle even if I give it to them.
And then I had to get into chunking and figuring out even
(44:59):
there you, you start to hit a problem where it chunked, but it
the way, the way that it went from 1 chunk to the next, it's
sometimes, well, there's ways around this before, but it'll
lose continuity between Page 1 and page 2.
And I was like, how do I deal with that?
(45:21):
So yeah, just, I guess just playing around with them and
then understanding context from my old work LED into how I would
find the boundary of or the constraints of working with
them. Yeah.
And I think it's a very important practice for everyone
to continually do where they're pushing the boundaries and see
what the limitations are, because it changes all the time.
Every model has different limitations, and the only way
(45:42):
you can really understand what the state is at this current
moment is by doing yourself. Last question for me, a lot of
this has been interacting, like communicating with the computer
or having the program interact with the user.
But it's very important to maintain human to human
communication, especially when we're all like locked away in
our offices doing our own thing.So as someone who's organized AI
agents Waterloo, which I love asa as a group, do you have any
(46:02):
advice for people to bring community together, organize
events, even things that Co working spaces to help promote
interperson communication? Any ideas to help make sure that
we keep the human aspect in thisindustry?
I love that because I think if anything, I know there's a lot
of fear around AI right right now, but I actually believe it's
going to reverse what mobile andsocial systems did and and
(46:25):
actually help us go be more human again.
But for tips around people that want more of that experience.
Yeah is well, find a builders club if you don't have one
around you, start a meet up group or run an event somewhere.
And I used to do that before my agents of group.
(46:46):
I ran IoT Waterloo and I would just go find any space that had
a bar or whatever and I would say, hey, I will bring people.
We just need this section of the, the meeting row.
The if they had an area, right, I'd find a space that had a
conference room or something. And I'd be like, look, I'll
(47:07):
bring in 50 people, 20 people and we're just going to talk
about this evolving technology. At that time, it was about
networked objects and you make money on the bar.
So my, my feedback to anybody who wants, you know, more human
to human interaction is, is go do it because waiting for
somebody else to do it, then you'll be waiting forever.
(47:29):
So if it doesn't already exist in your town or wherever you're
from, go do it. If that's the thing that you
really want and it's not there, then make it happen.
And even if it's five people or 10 people, I think you just
changed the format. So when you came and talked at
our event, it was, it was amazing.
A different, different structure, right?
(47:50):
Different vibe. People are there for different
reasons, but this guy Deb, who'spart of the barn here, ran a
very different format and he hadI think 1010 people, but it was
white boarding and just everybody talking about the the
agentic systems they're building, where they're
struggling, what's working, whatisn't.
(48:10):
And that was a really great interactive experience that so
much so I can't wait till he does his next one.
He calls his AI agents builders.I think mine was very casual
because I'm just doing what I did back in the when I ran AI,
sorry, IoT Waterloo, I was like,well, who out around me is doing
(48:31):
this? And then I just want to be
around them and talk to them andhear stories, right?
It's it's very high level, but he's right into the weeds.
So he's like, let's whiteboard to how are you addressing the
problem? Does anybody else in here have a
solution to that? Really, really cool.
And he does live stream data stuff.
(48:51):
So he's trying to figure out howtheir organization plays in the
future with live stream data, right.
So that's, that's a great way for people just to be
participants in, in this, this cycle because it's all real
time. Everybody's learning this real
time. All right, What a better time.
There's no better time to get out there and just participate
(49:14):
either attend events that are are actively sharing and
distributing the information. And if that's not happening,
then start one your own. It's never been easier, right?
You get on Luma and you just go,I'm going to create an event and
and don't sweat it if it's like only 5 people show up to because
those are the people you want totalk to anyways.
Like right. Whiteboard.
Whiteboard. Exactly.
(49:34):
Yeah. Whiteboard topic, only other
group I'd shout out as AI tinkerers.
I I love their events and they kind of say no investors, no
business, no suits, just people that make demos.
Yeah. And and that's a great, that's a
great play if that's your play. Like I love investors.
I love lots of different perspectives.
So, you know, and I think that'sthe, the beauty of having these
(49:56):
things because they end up becoming their own little unique
tribes. And, and if that's your, your,
your shtick, which is like no VC, no, you know, this, that
that's cool. But I think just naturally for
me, when I ran IoT Waterloo, I had a lot of business people
that were involved. It wasn't just tech.
So I try to find at least just for mine, the balance of like,
(50:16):
you know, it's, it's, it's high level.
And then storytelling can sort of help a business person see
how that might affect their their business model because
things are changing or if it's atech person.
Ian, this was a blast. Before we let you go, is there
anything you want the audience to know?
I yeah, so I guess the one thingI'm trying to get better at is
self promotion. I don't typically do this at
(50:37):
all, but it's one of the things that I need to get better on.
But the the book is not so much the thing I want to push the the
product I'm building or service.Product and service is trying to
figure out if there is a demand for people who want to become
leaders. So we are about 3 weeks away
from a closed beta release. And if anybody's interested,
(51:02):
that is an audience member of your show wants to learn how to
measure if they are becoming a better version of themself or
you know, what it actually meansto be become a leader.
And they were struggling like I was, which is like, I don't even
know. We, we've developed a way to use
to measure qualitative data. It's like a mirror and it's
(51:24):
private by design. We're looking for beta customers
to sort of like just kick it around and try and learn if it's
a if it's useful or if it's justsomething I'm going to use
myself. I don't know.
So I think it has value. But if you know, I won't know.
So if anybody in your audience is interested in, that's
something I'd be open to an introduction or a conversation
and let them try it out. Perfect.
(51:44):
Yeah, we'll put all the information down below.
Ian, this was a pleasure. We'll talk to you soon.
Awesome. Thank you so much, Mike.
Popular Podcasts
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!
NFL Daily with Gregg Rosenthal
Gregg Rosenthal and a rotating crew of elite NFL Media co-hosts, including Patrick Claybon, Colleen Wolfe, Steve Wyche, Nick Shook and Jourdan Rodrigue of The Athletic get you caught up daily on all the NFL news and analysis you need to be smarter and funnier than your friends.
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.