Hey Megan, let's do a podcast. Great idea. What

should we talk about?

Today we are talking about the Design and Implementation

Enablers section of the AI Implementation

Canvas. Megan, one of the things that I've heard you talk about with

the AI Implementation Canvas is that it's a capability

builder as much as a practical tool for a single

implementation. It strikes me that this section of the canvas

embodies that capability-building aspect more than any of

the other ones. Yeah, I think I agree with you 100%. Well, no,

"I think I agree with you 100%"? I totally agree with you 100%.

The design of the

Canvas is such that you use the Canvas for every implementation

in an organization, and that

means you might be using it a lot if you have a lot of things

going on. And the more that you use it, and the more the team

uses it, the more natural it's going to become to raise these

questions and have these conversations. In the early days of gen

AI release at Torrance Learning, our pilot teams... I think

we had six teams at one point each using a Canvas. And

that built in

early on the practice and the routine of thinking

around all these questions, these 14 planning dimensions, so

that the team in future implementations can be arriving

at each of these conversations faster. They're like, ah, I've been here before, done

that. But when we get into design and implementation

enablers, that's really about how do we start from

pilots, make thoughtful selections, and then

scale the projects that make sense scaling. So this is

one in which we're using the Canvas a lot

and really, really leveling up the entire game. Design

and Implementation Enablers is just one of the four sections.

There's Strategic Foundations, there's Technology & Experience

Infrastructure, there's also Human-Centered Adoption & Change, which will be another

conversation. That'll be a good one. Yeah, looking forward to that.

You mentioned pilots. I mean, so many pilots when it

comes to anything AI related. I've heard even

the term "random acts of AI." So how do organizations

pilot things efficiently and how do they figure out what

pilot projects make sense to move forward? This

is done in a number of different ways, but I think

some of the most informed approaches I've seen

have been almost like new product development.

And there are a number of opportunities to get

engaged, and they bring in.... One of the companies that I

researched for the book had a very structured

process by which they had a bunch of different pilots. Anybody could

submit an idea, and then ideas were vetted with some

criteria that were consistent across the organization so

that everybody's idea got a look,

but also that perspective and that feedback from that consistent

rubric. And then things that passed that first look got a little bit of time

and a little bit of funding to create a

prototype and to experiment a little bit. They also got support from the

organization. Can you imagine an organization of thousands

of people and everybody's got an idea about how to use AI a different way?

It's really hard from a governance perspective and keeping track of where is

our data and what's going on. So how do you resource that and provide each

one of those teams with some support? Requires a lot of transparency and

communication. So when you make it

okay to experiment, provide structured avenues and

support for experimentation, you then

bring all of those little random pockets in the organization that don't know

what the other part is doing, brings them together and

allows them to have a much more structured and organized

approach to having a lot of innovation going on at once.

Yeah. Another planning dimension you have here is

scaling and integration. And when I think about

scaling, you can think of pilots as like,

they could be small scale, but they also could be really large scale.

So it's sort of a layer on top of that even

and just another dimension to look at. So some AI projects and organizations

are going to be really big. They're going to like upend the way that systems

and processes work across your org, or

that there might be others that are going to be super small, right? Yeah. And

I think once a project has been greenlighted to

scale, that's not a "send the link to everybody!"

kind of moment, right? "Yay!"

That sounds dangerous.

But because

you're right, it's dangerous at that kind of

pivot point from pilot to scale, right? If you can go to pilot,

generally it may use dummy data, it may be a small portion of the

organization, it may not have

direct ... it may have a lot of humans sitting around watching it, making sure

it's been paying attention. So as we move from pilot to

scale, then we start looking at

IT and security involvement. We start looking

at stress testing and volume

testing for products. We may look at,

for lack of a better word, kill switch. How do we very

efficiently get the signal that this thing has gone awry and turn it

off before it causes more damage? And that's actually

some of the considerations around projects that we've got. How

do we monitor, by a human, maybe even

supported with AI, but how do we monitor these things and then make sure that

we turn them off if things aren't

going well? And what are the rules for that? So there's a lot

of considerations around scale.

And then I think there's also, like, how do we manage? How do we manage

this once it's huge and big and everybody's using it in an organization?

So it's a worthy

time to have a slowdown, a pause, and a lot of conversation.

You used a word when you were talking there. You said the word

signal. And that makes me think about measurement and impact

as well. I know we've talked in a lot of other episodes

that we've done about the importance of measurement. I'm

sure there's aspects that we want to measure here too, right?

Yeah. One of the things we want to be looking at as we design pilots

and then move into scale is what does success look like?

How will we know when we have it? And

what does success look like for people, for performance, for the organization

overall? And

this gets tied back to our business goals. What are we trying to achieve

here? And are we achieving this? And what happens

when we do? Right, so say a goal is to reduce a particular

process, reduce the speed it takes to complete that. If we

reduce the speed it takes to complete that, what do we do with that extra

time? So there's all sorts of

considerations here. And some of these things are going to be easy to measure, and

some of these things are not going to be easy to measure. But one of

the things that learning professionals bring to the table is

this focus on measurement and impact. And

both in the activity measurements — "are people using it?" The

effectiveness measurements — "are they using it well and using it

to change their work and outcome metrics?" Like, did it

achieve those desired outcomes? So there's lots of

opportunities here for us to apply the skills that we have already in

learning design and that

we could

borrow from our learning products into these AI products

and projects. When I'm

thinking about takeaways here, the first thing that comes to

mind is that this is almost about a

transition in a way, like we're talking about maybe

pilots to scaling and almost a maturity that's

taking place in this section of the quadrant. Would you say that's right? I

do. Yeah, I agree. And it's even — I think there's

both maturity in moving from pilot to implementation with a rigorous

measurement, but also maturity in even having those structures

in the organization in the first place. Mhm, and having those conversations. What's

one more thing you've got for us? You know,

one of the things we talk about is like, "where are the people in each

of these sections?" And the

people here are — what I love

about the, especially the experimentation and pilots perspective,

is there's innovation and a

divergence-of-thinking opportunity to get

a lot of people involved and excited and an

opportunity for sideline skill development around that.

"How do I make a business case?" "How do I present my case?" "How do

I team with other people across the organization to be on this

part of this project team?" And "how do I connect in with the purpose of

my organization by participating in this innovation

effort?" So there's a lot there and

a lot of really good conversation, I think,

to be had. Yeah. I mean,

pilots need people. Pilots need people.

And feedback, which we can measure. Go figure.

All right, Megan, how'd that one go? That was a lot of fun to

talk about. It was — I love, you know

me, I love a good hackathon. So I get very

excited about the experimentation end of things and

the prototyping, like, "hey, can we build this thing?" And I know, like, personally,

sometimes I get frustrated when you hit what

feels like organizational friction

to scaling. That friction is there

for a really good reason. And so

that's important. But it is definitely a gear shift and a

speed shift that's important but sometimes hard for

individuals to internalize.

Instead of friction, I thought you were — you mentioned "antifragile" the

other day, and it almost

made me think of like, I thought your brain

or my brain was starting to go towards, like, "I hit the organizational

rigidity." Yes!

But yet from an antifragile — right. So let's back up,

right? The concept of "antifragile," or "anti-fragility." And we'll

put the link to the book in the show notes because it's a concept —

I'm super fascinated, right? So one of the

core themes here is that things that are fragile,

when you push up against them, they break.

Your pilots might be fragile. Right. It's a prototype. It's

not built out and robust. Things that are rigid,

when you push up against them, they also break. They push back

or they stop. That's not healthy either. And then

things that are antifragile, when you push up against them, they actually get

stronger. And it helps the team get stronger, it helps the organization and the process

gets stronger. So that pushback from the organization — at

scale, at governance, at bringing the security people and

looking at the wider impacts — absolutely makes an

idea stronger. This is Meg Fairchild and

Megan Torrance, and this has been a podcast from

TorranceLearning. Tangents is the official podcast of

TorranceLearning. (As though we have an unofficial one.) Tangents

is hosted by Meg Fairchild and Megan Torrance. It's

produced by Dean Castile and Meg Fairchild, engineered and

edited by Dean Castile, with original music also

by Dean Castile. This episode was fact checked

by Meg Fairchild.