EPISODE 7 (Part 2): AI Strategy: The 9 Pillar AI SWIFTER Framework. A Comprehensive Guide for CEOs & Business Owners

Speaker: 00:00:00

If you're a business owner, a CEO, or an

executive, are you feeling the familiar

2

: 00:00:05

unease around all the AI hype, the sense

that you need to be doing something,

3

: 00:00:10

yet the fear of making the wrong

potentially costly move can be daunting.

4

: 00:00:16

You're not alone.

5

: 00:00:17

I speak with CEOs and business owners

regularly who wrestle with this exact

6

: 00:00:23

tension, the risk of inaction versus

the peril of going down a wrong path.

7

: 00:00:29

Or even complete failure

of the AI initiative.

8

: 00:00:34

That pressure resonates with you.

9

: 00:00:35

Then you, me out for the next few

minutes, because in this episode I'm

10

: 00:00:39

going to lay out nine fundamental pillars.

11

: 00:00:44

Think of them as your strategic blueprint

for navigating the complexities of

12

: 00:00:49

ai, allowing you to make confident,

impactful decisions, not just chase

13

: 00:00:55

the latest sexy trend or AI hype.

14

: 00:01:00

Stick with me and discover all

the nine pillars in detail because

15

: 00:01:04

leaving one out may just put your AI

initiative down the path to failure.

16

: 00:01:11

I promise you clarity and

confidence is within your reach.

17

: 00:01:33

Now for those of you who joined me

last time in episode six, part one,

18

: 00:01:37

you'll recall we took a deep dive into

the executive mindset surrounding ai.

19

: 00:01:43

We unpacked what internal

struggles many of you are facing.

20

: 00:01:49

The tug of war between the fear of being

left behind and the very real concerns

21

: 00:01:54

of investing down a dead end path.

22

: 00:01:58

We explored what failure means and

actually looks like some AI projects

23

: 00:02:03

seemingly succeed from a project

perspective, but fail when implemented

24

: 00:02:08

in the real day-to-day business.

25

: 00:02:10

And you'll recall that we identified

eight often overlooked factors that

26

: 00:02:15

can lead to AI initiatives failing

before they even truly begin.

27

: 00:02:21

If you happen to miss conversation.

28

: 00:02:24

Don't worry.

29

: 00:02:24

This episode stands entirely on its own.

30

: 00:02:27

However, if you're curious to delve

deeper into those potential pitfalls,

31

: 00:02:33

I'll include a link in the show notes.

32

: 00:02:35

But today, our focus shifts,

we're moving beyond that initial

33

: 00:02:40

apprehension and stepping firmly

into the path of proactive strategy.

34

: 00:02:45

We're transitioning from

guesswork to decision making.

35

: 00:02:51

To guide us, I want to introduce you to

what I call the AI swifter Framework.

36

: 00:02:59

Now, this isn't just another

trendy acronym to add to your list.

37

: 00:03:04

This is a structured approach, a

set of guiding principles designed

38

: 00:03:10

to bring clarity, confidence.

39

: 00:03:13

A clear sense of direction to your

AI journey, regardless of where

40

: 00:03:17

your organization currently stands.

41

: 00:03:21

However, before we dive into the

specifics of AI swifter, there's

42

: 00:03:25

a more foundational element.

43

: 00:03:27

We need to discuss something that

underpins every single aspect

44

: 00:03:31

of successfully integrating

AI into your business.

45

: 00:03:37

That process is called discovery.

46

: 00:03:40

It's the ongoing radar

that guides everything.

47

: 00:03:45

Think of discovery as your always

on radar in the rapidly evolving

48

: 00:03:50

world of artificial intelligence.

49

: 00:03:53

It's not a one-time task, it's an

ongoing discipline, a continuous

50

: 00:03:58

scanning of the horizon.

51

: 00:04:00

Why is this so crucial?

52

: 00:04:02

Because without this constant process.

53

: 00:04:05

Discovery.

54

: 00:04:06

Your AI decisions risk

becoming purely reactive.

55

: 00:04:11

You might find yourself chasing the

latest headlines, blindly following vendor

56

: 00:04:16

pitches, or even worse, being swayed

by consultants, pushing solutions that

57

: 00:04:22

aren't truly aligned with your needs.

58

: 00:04:25

Discovery provides you with a

space to observe, to thoughtfully

59

: 00:04:29

analyze, and to lead with a sense

of calm and informed perspective.

60

: 00:04:36

It eliminates the fear of missing out

because discovery provides concrete

61

: 00:04:40

information and turns feelings of

overwhelm into calm, reassurance, and

62

: 00:04:47

shows up real opportunities of ai.

63

: 00:04:52

Ultimately, it ensures you are

well prepared when AI genuinely

64

: 00:04:57

merges as as the right solution for

your unique business challenges.

65

: 00:05:04

All the angles to the

framework I'll cover today.

66

: 00:05:08

How do you implement

this discovery process?

67

: 00:05:10

It doesn't need to be overly complex.

68

: 00:05:13

Establish a simple, consistent

system for regularly scanning

69

: 00:05:17

industry news, keeping tabs on your

internal initiatives, monitoring

70

: 00:05:24

your moves, and staying abreast of.

71

: 00:05:27

Pay particular attention to emerging

disruptors in your market and keep

72

: 00:05:32

an eye on new technologies, AI

models, and emerging methodologies.

73

: 00:05:38

For now, simply gather this

information, file it away.

74

: 00:05:43

The initial goal is just

to learn and understand.

75

: 00:05:49

Maybe you can assign a small team

or even rotate the responsibility

76

: 00:05:53

to champion this discovery process.

77

: 00:05:56

Perhaps create a shared

document or a simple dashboard.

78

: 00:06:00

The key is consistency.

79

: 00:06:02

It needs to become an ingrained

habit within your organization.

80

: 00:06:07

What exactly are you looking

for in your discovering scans?

81

: 00:06:11

Keep an eye out on the following.

82

: 00:06:13

Number one, emerging

AI tools and platforms.

83

: 00:06:16

Number two, innovative AI-driven business

models that are gaining traction.

84

: 00:06:21

Number three, practical and impactful

AI use cases relevant to your industry.

85

: 00:06:28

Number four, the ever-evolving

ethical considerations surrounding ai.

86

: 00:06:35

Crucial regulatory changes

in the AI landscape.

87

: 00:06:39

And number six, importantly pay

attention to both the successes and the

88

: 00:06:45

failures of AI implementation by others.

89

: 00:06:49

There are valuable lessons

to be learned from both.

90

: 00:06:51

So what happens if you neglect this

foundational discovery process?

91

: 00:06:56

Quite simply, you risk

becoming a victim of hype.

92

: 00:07:01

You might end up investing in

systems you don't truly need.

93

: 00:07:06

Or worse, you live with a constant

state of mental anxiety, unsure

94

: 00:07:11

of what to do, who to trust, and

ultimately missing the strategic

95

: 00:07:16

path that's right for your business.

96

: 00:07:19

So who is responsible for discovery?

97

: 00:07:22

While everyone in organization

plays a role in being aware of d.

98

: 00:07:28

Designate a few passionate individuals.

99

: 00:07:30

Perhaps those natural enthusiasts who

love technology and digital innovation.

100

: 00:07:37

They can be your dedicated radar watchers.

101

: 00:07:41

These champions can help guide the

rest of the organization, inspire

102

: 00:07:47

curiosity, and foster a culture of

AI awareness within your workspace.

103

: 00:07:55

Discovery is what provides you

with that crucial breathing room in

104

: 00:07:59

the often turbulent waters of ai.

105

: 00:08:04

It transforms potential chaos

into confident decision making,

106

: 00:08:09

and it's a discipline that will

serve you exceptionally well in

107

: 00:08:13

every AI related conversations and

decisions you make moving forward.

108

: 00:08:20

You'll be informed, you'll speak with

conviction and your choices will be

109

: 00:08:25

rooted in objective understanding.

110

: 00:08:28

So, all right, now that we've

established the critical importance

111

: 00:08:31

of having your radar constantly

scanning, let's delve into the heart

112

: 00:08:37

of our discussion today, the nine

pillars of the AI Swifter framework.

113

: 00:08:44

Each of these pillars is intentionally

designed to help you build AI solutions

114

: 00:08:50

that deliver genuine sustainable

value, creating a tangible shift

115

: 00:08:56

into your in your business, and

potentially unlocking opportunities

116

: 00:09:00

you haven't even considered yet.

117

: 00:09:01

It's important to understand

that these pillars aren't

118

: 00:09:06

necessarily a linear process.

119

: 00:09:07

Rather, they represent the

essential foundational elements.

120

: 00:09:12

That should inform, inform every

AI initiative you undertake

121

: 00:09:17

within your organization.

122

: 00:09:19

So starting with the A in ai swifter

alignment at its core alignment is

123

: 00:09:27

about ensuring your AI efforts are

directed at solving the right problems

124

: 00:09:33

for the right people and in the right

way, ensuring it truly serves your

125

: 00:09:39

overarching business objectives.

126

: 00:09:41

Stakeholder needs.

127

: 00:09:45

Often AI initiatives are technology led,

going down a path propelled by hype,

128

: 00:09:51

implementing solutions not aligned with

the business goals or stakeholder needs.

129

: 00:09:59

The alignment component

tackles this challenge head-on.

130

: 00:10:04

When implemented, you are 100% certain.

131

: 00:10:08

That you are building, that what you are

building is needed in your organization

132

: 00:10:13

and is most likely to move the performance

needle if implemented correctly.

133

: 00:10:18

Why is this so critical?

134

: 00:10:20

Because AI that isn't strategically

aligned risks becoming a mere sideshow

135

: 00:10:28

flashy technology that doesn't

address any meaningful business

136

: 00:10:32

challenges or provide real value.

137

: 00:10:35

It's a tail waacking the dog analogy.

138

: 00:10:37

How is the alignment pillar implemented?

139

: 00:10:41

The process of alignment begins with

identifying your stakeholders and

140

: 00:10:46

understanding how they play their

part in your organization's ecosystem.

141

: 00:10:53

As I've already highlighted in previous

episodes, it's helpful to categorize

142

: 00:10:58

stakeholders into three key groups.

143

: 00:11:03

First, those you serve,

as you will remember.

144

: 00:11:06

Those are your shareholders, your

customers, or the communities you impact.

145

: 00:11:14

Second, your enablers, your

staff, your internal teams, your

146

: 00:11:19

partners, and the key vendors.

147

: 00:11:24

And third, your influences.

148

: 00:11:27

These are your board of directors,

regulatory bodies, and your investors.

149

: 00:11:36

Each shareholder group has their

unique needs and aspirations that

150

: 00:11:40

can be fulfilled by AI solutions,

but more importantly, you must

151

: 00:11:45

understand how solving problems of one

stakeholder group impacts the other

152

: 00:11:50

two groups that we just spoke about.

153

: 00:11:55

So, for example, to implement

an AI client portal.

154

: 00:11:59

You'd need to provide staff with

tools to upload information and

155

: 00:12:04

step in where AI can't provide the

information required by your clients.

156

: 00:12:11

Going even further, you need AI

driven performance dashboards

157

: 00:12:15

showing the influencers, in other

words, the executives that the

158

: 00:12:20

AI solution is paying off and is

positively driving business growth.

159

: 00:12:27

Cost reduction.

160

: 00:12:28

What do you need to do to get

alignment of AI solutions to

161

: 00:12:32

stakeholders and business needs?

162

: 00:12:36

You effectively understand

and achieve alignment.

163

: 00:12:41

You need to start with stakeholders,

categorization I just told you about.

164

: 00:12:47

Identify stakeholder needs through

interviews and focus groups or

165

: 00:12:51

surveys to gather direct insights.

166

: 00:12:56

Carry out stakeholder journey

mapping to visualize the experience

167

: 00:13:00

you are aiming to improve.

168

: 00:13:03

Note I said stakeholders,

not just customers.

169

: 00:13:06

You need to factor all three stakeholder

groups we've just covered earlier.

170

: 00:13:12

So map out stakeholder needs, problems

and aspirations on an impact effort matrix

171

: 00:13:19

to help prioritize initiatives that will

give your business the greatest results.

172

: 00:13:24

The most optimal investment

of time and capital.

173

: 00:13:30

So what if you skip this crucial

step of alignment, you might

174

: 00:13:35

very well end up building a

technically brilliant AI solution.

175

: 00:13:39

You may even build it in record

time and super efficiently

176

: 00:13:44

within budget estimates, however.

177

: 00:13:47

You will build an AI solution

that ultimately no one needs or

178

: 00:13:51

even worse, no one actually uses.

179

: 00:13:55

You'll most likely end up building

a solution looking for a problem.

180

: 00:14:00

It's therefore no surprise why more than

85% of AI initiatives actually fail.

181

: 00:14:08

Alignment is a North Star guiding your

AI led transformation based on the needs

182

: 00:14:14

of your business and your stakeholders.

183

: 00:14:18

So who's responsible for

the alignment pillar?

184

: 00:14:22

While your leadership team and business

analysts will likely take the lead in

185

: 00:14:26

this area, it's absolutely vital to

involve end users early in the process.

186

: 00:14:33

Listen attentively to feedback from

across your entire organization.

187

: 00:14:38

So let's talk about innovation.

188

: 00:14:40

The first I in the AI swifter

framework, in our case, innovation

189

: 00:14:44

isn't that shiny buzzword heavy.

190

: 00:14:47

Outta this world kind of innovation

that's all about being first to market.

191

: 00:14:52

What we're interested in here

is purposeful innovation.

192

: 00:14:56

Nothing groundbreaking, but

something new to your organization.

193

: 00:15:00

Innovation that's rooted in business

value and solves something real,

194

: 00:15:05

not just chasers the latest trends.

195

: 00:15:09

Why does this matter?

196

: 00:15:11

Because without a structured approach to

innovation, you're left with two extremes.

197

: 00:15:17

Either your businesses or you waste

time, spinning wheels on disconnected

198

: 00:15:24

side projects that really go nowhere.

199

: 00:15:29

So how do you build innovation

that actually works?

200

: 00:15:33

You have to start by creating

safe spaces to test ideas.

201

: 00:15:40

No one will want to innovate in

an environment where they get

202

: 00:15:43

fired for trying out new things.

203

: 00:15:46

Fail.

204

: 00:15:47

Ultimately encourage bottom up

input, not just top down vision.

205

: 00:15:53

Find ways to capture the ideas from

the frontline, from people who see the

206

: 00:15:59

opportunities and the problems while

serving stakeholders on a daily basis.

207

: 00:16:06

You can run small pilots with

room to fail, followed by

208

: 00:16:10

reflection and adaptation.

209

: 00:16:12

That's how you get real

insights and shift.

210

: 00:16:18

What could the innovation

process look like?

211

: 00:16:23

Remember the discovery

process I spoke about.

212

: 00:16:27

This is an ongoing process and it'll

give you new ideas inspired by AI led

213

: 00:16:33

business models and new technologies

that show you what's possible.

214

: 00:16:39

Discovery is the foundation to

innovation because it gives you

215

: 00:16:44

ideas of possibilities that you

won't even have thought about.

216

: 00:16:48

Don't stick to your industry.

217

: 00:16:49

Explore new AI technologies and business

models emerging in other industries

218

: 00:16:55

that has got nothing to do with yours.

219

: 00:16:57

So innovation starts with brainstorming

in a safe space where ideas from all

220

: 00:17:03

levels of the company can flow freely.

221

: 00:17:06

No hierarchy.

222

: 00:17:07

No politics and no egos.

223

: 00:17:12

If you're brainstorming online,

use whiteboard tools to capture

224

: 00:17:15

ideas from remote teams.

225

: 00:17:18

Make it a fun and creative process.

226

: 00:17:23

Remember, innovation shouldn't

just be for the sake of innovation.

227

: 00:17:28

You are innovating to solve a pressing

problem or delivering on the needs

228

: 00:17:34

of one of your stakeholder groups.

229

: 00:17:36

Not just customers.

230

: 00:17:37

Remember also that by going through

the alignment pillar, you'll be finding

231

: 00:17:43

innovative ways to solve a problem that

is surfaced as a priority, one that will

232

: 00:17:49

deliver the highest return on capital and

time invested ideas don't only come up in

233

: 00:17:55

well structured brainstorming sessions.

234

: 00:17:58

They are spontaneous.

235

: 00:18:00

It can come from anywhere.

236

: 00:18:01

Maybe when one of your team is

dealing with an irate customer.

237

: 00:18:05

Maybe when your employee is frustrated

by a complex and unnecessary process.

238

: 00:18:15

Or maybe when managers can't monitor

the performance of a business, or

239

: 00:18:22

maybe it comes to you as the leader

in your business when you are in.

240

: 00:18:33

You really need to develop a mechanism

to capture ideas where they strike.

241

: 00:18:39

This is in addition to the

formalized brainstorming process.

242

: 00:18:44

For example, keep suggestion

boxes around your office.

243

: 00:18:50

Develop a dedicated email

address to capture ideas from

244

: 00:18:54

across the organization or.

245

: 00:18:57

Develop a web form to capture

these innovative ideas.

246

: 00:19:01

Reward people that come

up with great ideas

247

: 00:19:06

after ideation.

248

: 00:19:08

The idea must translate into AI

prototypes that can be tested.

249

: 00:19:14

Of course, only if this makes sense

and passes the should we do it test.

250

: 00:19:21

Maybe you can run a design sprint around

a specific and well-defined problem.

251

: 00:19:25

Host internal innovation

contests like hackathons.

252

: 00:19:30

Alternatively, you can prototype

quickly, share the solution with real

253

: 00:19:33

users and listen to their feedback.

254

: 00:19:36

Ask them how the prototype can be

improved to make their lives better.

255

: 00:19:42

There are many resources and books

out there on the innovation process,

256

: 00:19:47

so I won't cover it here, really,

but what if you skip the step of

257

: 00:19:50

innovation in the AI swifter framework?

258

: 00:19:54

When innovation isn't guided by your

strategic objectives and stakeholder

259

: 00:19:58

needs, you'll end up chasing novelty

and hype instead of progress.

260

: 00:20:05

You pursue innovation for

innovation's sake rather than

261

: 00:20:09

for improving your business and

how you serve your stakeholders.

262

: 00:20:13

When going down the wrong type

of innovation, you burn time,

263

: 00:20:17

budget, and goodwill on things

that don't really move the needle.

264

: 00:20:25

Okay, so who should be

responsible for the process of

265

: 00:20:28

innovation in your organization?

266

: 00:20:30

Naturally, the r and d

team comes to mind, right?

267

: 00:20:35

If you run a small business,

you'd think it's the job of the

268

: 00:20:38

CEO and the other executives.

269

: 00:20:42

However, innovation should be

everyone's job from the front line.

270

: 00:20:49

To leadership because the best

ideas often come from where

271

: 00:20:53

the problems are felt the most.

272

: 00:20:57

In business, innovation

can't happen in a lab.

273

: 00:21:01

Ideas can come from anywhere, and

you have to have a system to capture

274

: 00:21:06

and quickly evaluate new ideas.

275

: 00:21:13

Okay, so let's now turn to S

in the AI swifter framework.

276

: 00:21:17

S stands for skills.

277

: 00:21:19

People, most people will be

forgiven for thinking that

278

: 00:21:22

AI is a technology challenge.

279

: 00:21:25

We associate AI with data methods

like large language models or buying

280

: 00:21:31

and deploying super fast computers

or installing internet of things,

281

: 00:21:35

sensors that hook up to AI systems.

282

: 00:21:40

But make no mistake, AI

is powered by people.

283

: 00:21:46

People build it.

284

: 00:21:48

Deploy it, question it, use it, and

clean up after it when things go wrong.

285

: 00:21:59

So why focus on people?

286

: 00:22:00

Because people make AI happen

and it's essential that you

287

: 00:22:04

have the right mix of skills.

288

: 00:22:08

You definitely need deep technical skills,

but you also need skills that motivates

289

: 00:22:14

and empowers humans to use the AI systems.

290

: 00:22:18

As you already know, this is often

referred to as change management skills.

291

: 00:22:23

You also need project management

skills, people who can teach others

292

: 00:22:28

how to use AI systems, experts who

understand risks and the ethics of ai.

293

: 00:22:35

Process improvement specialists,

et cetera, et cetera, et cetera.

294

: 00:22:40

So how do you get these skills on board?

295

: 00:22:43

Train up your people, staff and

executives hire new skills if you are

296

: 00:22:49

able to attract them in a competitive

market, outsource or bring contractors

297

: 00:22:55

on board to fill a temporary demand.

298

: 00:22:58

How do you approach the skills and people

component of the AI swifter framework?

299

: 00:23:05

You could start with a simple

capability map, taking stock of the

300

: 00:23:10

skills you already have on board.

301

: 00:23:13

Remember, you may have an AI

expert hobbyist working in a

302

: 00:23:18

completely non-AI related business

function, totally being wasted.

303

: 00:23:23

When considering your AI goals, you

have to find these people and make sure

304

: 00:23:29

they are using their skills effectively.

305

: 00:23:33

After all, it's a win-win.

306

: 00:23:36

They will be much happier in the job

while you have just acquired an AI

307

: 00:23:41

expert or inspiring expert without

the pain of hiring or training.

308

: 00:23:49

Once you know what skills you have

on board, now you can identify the

309

: 00:23:54

gaps, gaps, gaps based on where

you are today against what AI

310

: 00:23:59

initiatives will be in place in future.

311

: 00:24:03

This can only happen once you know what

AI solutions you are going to prioritize

312

: 00:24:08

and implement, and this comes from work

you've already done in the alignment and

313

: 00:24:14

innovation pillars we've already covered.

314

: 00:24:17

You'll need to figure out how to build

or acquire these skills in a market

315

: 00:24:23

with a demand for ai, technical and

soft skills currently outweigh supply.

316

: 00:24:29

Yes, this can be challenging and you

will need to find new ways to attract

317

: 00:24:34

top talent away from your competitors.

318

: 00:24:39

The Skills and People Pillar is

all about building a team that

319

: 00:24:43

not only understands the tech,

but also the context it lives in.

320

: 00:24:48

What happens if you don't develop

the skills and people pillar

321

: 00:24:53

of the AI Swift framework?

322

: 00:24:57

You may end up building

or buying a system.

323

: 00:25:00

No one is equipped to implement in

the business implementation stalls.

324

: 00:25:06

Risks escalate and the expected results

don't materialize without the right

325

: 00:25:13

people, it makes little sense to embark

on an ambitious AI system build things

326

: 00:25:19

will just come to a grinding halt until

you can get those skills on board.

327

: 00:25:25

In whichever way you need

to get them on board.

328

: 00:25:29

If you forgot everything I just told you

about skills, remember this one thing.

329

: 00:25:34

You may already have an AI expert in your

company that you don't even know about.

330

: 00:25:40

People pursuing AI qualifications

in their private life building

331

: 00:25:45

hobby AI projects and side hustles.

332

: 00:25:49

Find them and develop

an AI career for them.

333

: 00:25:54

So who owns the skills and people

pillar of the AI swifter framework?

334

: 00:25:59

Again, it's not only the

responsibility of the HR department

335

: 00:26:03

to find and develop AI skills.

336

: 00:26:06

This really is a team sport.

337

: 00:26:09

Anyone leading people play a role here,

whether it's HR department heads, or the

338

: 00:26:16

Learning and Development Department, or

its employees recommending their friends.

339

: 00:26:22

Possibly also their family.

340

: 00:26:26

Let's now turn to the W in the AI Swifter

Framework, workflows and Adoption.

341

: 00:26:33

This pillar is all about the use of AI

systems that are built and deployed.

342

: 00:26:39

It's about embedding the AI systems,

IT workflows, and adoption by users

343

: 00:26:45

across all stakeholder groups.

344

: 00:26:50

In my experience, this is often the

most overlooked part of AI initiatives.

345

: 00:26:56

'cause even the smartest tech in the

world is useless if no one uses it.

346

: 00:27:03

Executives and decision makers

often get caught up in promise of

347

: 00:27:07

technology thinking it'll deliver

the promised results even if the

348

: 00:27:12

system is right for the business.

349

: 00:27:15

Implemented with precision, it makes

no difference to the business if

350

: 00:27:21

such systems are not actually used.

351

: 00:27:25

So think about this.

352

: 00:27:27

How could an organization

end up going through the cost

353

: 00:27:30

and effort of implementing a

system that's not even used?

354

: 00:27:35

What's going on here?

355

: 00:27:39

This situation can easily happen

when AI initiative start with a tech.

356

: 00:27:46

Leaders are often mesmerized

by technology and it's promise.

357

: 00:27:51

You buy it and implement it, but

not because there was a need for it,

358

: 00:27:55

but because you believed that the

promise sold to you by the vendor.

359

: 00:28:03

Go back to episode one where I talked

about an AI chat, Bott example.

360

: 00:28:07

The bank had a predominantly

older customer base who didn't

361

: 00:28:11

want to use the chatbot.

362

: 00:28:14

They prefer speaking to a real person.

363

: 00:28:18

Also, in that example, when chatbots

start giving wrong answers, staff

364

: 00:28:24

will obviously discourage their

customers from using these chatbots.

365

: 00:28:29

A key missing component here is the

alignment step in our AI swifter

366

: 00:28:34

framework, which was ignored.

367

: 00:28:36

Really, the AI system implemented without

stakeholder needs assessment risks.

368

: 00:28:45

Why does adoption matter so much?

369

: 00:28:47

Because shelfware doesn't deliver

any benefits, let alone return on

370

: 00:28:52

investment adoption of AI systems.

371

: 00:28:56

In other words, AI systems that

are actually used by the target

372

: 00:29:00

stakeholder groups is what turns

potential and promise into performance.

373

: 00:29:07

So how do you get it right?

374

: 00:29:10

Simple focus on the alignment pillar

in the AI swifter framework, and

375

: 00:29:14

make sure the AI system is addressing

a genuine desire or aspiration

376

: 00:29:19

of a specific stakeholder group.

377

: 00:29:22

Then design it with a user in

mind and not for them without

378

: 00:29:27

any involvement from them.

379

: 00:29:30

You've gotta watch what people

actually do, not just what they

380

: 00:29:34

say they do when using a system.

381

: 00:29:39

Are there too many steps in

the user journey that gets

382

: 00:29:42

them to abandon the AI system?

383

: 00:29:46

Is the user interface

clunky and confusing?

384

: 00:29:51

Are the colors on the app

or webpage overpowering?

385

: 00:29:56

All of these things really do matter.

386

: 00:29:59

You must co-create workflows that

fit into the reality of your users.

387

: 00:30:04

You start with a prototype,

but then get feedback from as

388

: 00:30:08

many actual users as possible.

389

: 00:30:10

Look out for problems, listen to

their feedback and suggestions.

390

: 00:30:16

Ask them how the AI

solutions could be improved.

391

: 00:30:19

Iterate, test, get feedback, and repeat

the cycle until you get a system which

392

: 00:30:25

users are unanimously happy with.

393

: 00:30:31

Of course, you can't please everyone

and naturally some users will still

394

: 00:30:36

dislike certain features of workflow.

395

: 00:30:38

That's just part of the process.

396

: 00:30:42

What are some ways or

practices to ensure adoption?

397

: 00:30:48

Let's just explore a few

tools and techniques.

398

: 00:30:51

These are only examples, really.

399

: 00:30:53

Number one, journey mapping and testing.

400

: 00:30:55

Number two, shadowing

actual users on the ground.

401

: 00:30:58

See how they use pilot systems,

learn what works and what doesn't.

402

: 00:31:03

And number three, prototyping tools in

real environments with real users, not

403

: 00:31:09

just in boardrooms or innovation labs.

404

: 00:31:14

In some cases, even in the AI

system, is beneficial to the user

405

: 00:31:18

and co-created with their input.

406

: 00:31:20

Users may avoid using it deliberately

because of a fear that the AI system

407

: 00:31:25

is going to take away their job.

408

: 00:31:28

Some may just resist change

because they like doing things

409

: 00:31:32

either have always been done.

410

: 00:31:35

What if you don't focus on the workflow

and adoption pillar of the AI framework?

411

: 00:31:41

You won't get the intended stakeholders

using the system, a wasted investment

412

: 00:31:45

of time, money, and effort.

413

: 00:31:48

You can lose trust for the

users when a system is forced

414

: 00:31:51

on them without their input.

415

: 00:31:54

People work around systems

instead of working with it.

416

: 00:31:58

People resist the change, and AI

becomes just another failed project.

417

: 00:32:07

The who should lead this

workflow and adoption.

418

: 00:32:10

Ideally, you give this responsibility

to a digital transformation

419

: 00:32:15

lead or a change manager.

420

: 00:32:18

However, success depends on close

collaboration with operational leaders

421

: 00:32:23

to align workflows with day-to-day

realities with HR people, teams for

422

: 00:32:30

using training, training, change,

communication, and behavioral change.

423

: 00:32:36

Frontline staff for systems that

are intended to be used by customers

424

: 00:32:41

and other stakeholders like joint

venture partners or suppliers

425

: 00:32:47

IT or product managers to ensure

the technology is implemented

426

: 00:32:52

with usability in mind.

427

: 00:32:55

Change management champions in

every business unit can certainly

428

: 00:32:59

go a long way to encouraging

widespread adoption of the AI system.

429

: 00:33:06

Let's now move on to the second I in

AI swifter framework infrastructure.

430

: 00:33:14

This is in effect the nuts

and bolts of the AI system.

431

: 00:33:18

Your infrastructure is the foundation

and it includes a component

432

: 00:33:23

such as computer resources.

433

: 00:33:25

You need high performance computers that

can scale as AI technologies advance.

434

: 00:33:32

A more practical solution could

be to use cloud computing with

435

: 00:33:37

platforms like AWS Azure and Google.

436

: 00:33:41

They offer on demand access to vast

computer resources, which you can

437

: 00:33:45

scale up or down as you require

without having to invest in buying

438

: 00:33:51

and building the system yourself.

439

: 00:33:55

Many highly regulated industries will

often need to invest in their own service.

440

: 00:34:00

Where they can keep the data locally and

have the computer power on tap instead

441

: 00:34:06

of sharing it with others in the cloud,

442

: 00:34:10

you'll need data storage

and management solutions.

443

: 00:34:14

AI models learn from a vast amount

of data that need to be extracted,

444

: 00:34:19

cleaned, and stored in a systematic

way so that it can be retrieved easily.

445

: 00:34:27

Data, after all is the

lifeblood of AI systems.

446

: 00:34:32

Whether it's your data or external

data available in public libraries.

447

: 00:34:37

AI also needs high speed networks

and connectivity load as systems

448

: 00:34:42

need to quickly upload and download

data to perform their functions.

449

: 00:34:46

Such data transfers must be possible

across different types of systems.

450

: 00:34:51

Most importantly, the network

over which data travels.

451

: 00:34:55

Must be secure, protecting sensitive

data and systems against hackers.

452

: 00:35:03

So the AI system will need

sophisticated security and compliance

453

: 00:35:07

infrastructure that are kept up to date.

454

: 00:35:10

Why does this infrastructure

pillar matter?

455

: 00:35:13

Because your AI initiatives

will be limited by your infras.

456

: 00:35:22

If your hardware can't cope with

the demands of AI solutions,

457

: 00:35:28

whereas if your infrastructure

fails or doesn't function properly,

458

: 00:35:33

you don't get, you don't get AI

transformation, you get AI chaos.

459

: 00:35:38

And with AI's ability to automate

things can get out of hand very quickly.

460

: 00:35:43

Putting your organization, your

customers, and other stakeholders at risk.

461

: 00:35:51

So how do you avoid that?

462

: 00:35:52

You start by assessing what you

already have, what systems can

463

: 00:35:58

integrate, what need replacing,

what's your data quality really like?

464

: 00:36:05

Remember, all those gap analysis is

based on work you've already done in

465

: 00:36:09

the alignment and innovation pillars.

466

: 00:36:12

These two pillars are foundational and

they ensure you are building an AI system.

467

: 00:36:17

Designed to meet stakeholder needs

that shift the business to new heights.

468

: 00:36:26

We won't go into the technicalities

of AI infrastructure.

469

: 00:36:29

There's already quite a bit of

information that is already out there.

470

: 00:36:32

Vendors will be pleased to inform

you about the hardware so you

471

: 00:36:36

can get information from them.

472

: 00:36:39

Maybe we can dedicate a future episode

of this topic or better get some

473

: 00:36:43

infrastructure provider guests to go

through all the infrastructure components

474

: 00:36:48

and the technology stack you will need to

consider when building your AI solution.

475

: 00:36:56

So what's the risk if you skip

this infrastructure better?

476

: 00:37:02

Well, your AI system just won't work

no matter how well they're designed.

477

: 00:37:09

System flaws and security breaches will

expose you legally and reputationally.

478

: 00:37:17

So who owns the hot

infrastructure component?

479

: 00:37:21

Fortunately, this is one area you

don't want everyone involved, just

480

: 00:37:26

IT experts, your IT team, your data

architects and your cybersecurity

481

: 00:37:32

teams will take the lead.

482

: 00:37:34

If you don't have these teams, you can

outsource or bring incredible freelancers.

483

: 00:37:40

They're not back offers.

484

: 00:37:41

They're mission critical.

485

: 00:37:42

In this case, definitely rely on

vendors, but only once you are clear on

486

: 00:37:49

what AI systems you are implementing.

487

: 00:37:52

And most importantly why as an

executive or owner of the business,

488

: 00:37:57

you don't need to understand the

nuts and bolts of infrastructure.

489

: 00:38:01

However, you need to have a grasp.

490

: 00:38:04

Of the basic infrastructure needed for

AI and keep abreast of new developments.

491

: 00:38:10

Make sure you have a strong IT

and hardware team, whether they

492

: 00:38:15

are in-house or outsourced.

493

: 00:38:17

So next we moved the F in the AI swifter

framework, frameworks and models.

494

: 00:38:22

Now we're talking about the engine under

the hood, the powerhouse of AI systems.

495

: 00:38:28

This is where the models live.

496

: 00:38:29

The logic that drives your AI

talking, we are talking about

497

: 00:38:33

algorithms, techniques, and software

tools to build and train AI models.

498

: 00:38:39

It's essentially the brain of the

AI system while infrastructure

499

: 00:38:43

was the hardware and data.

500

: 00:38:46

Here we focus on the software briefly,

we've got the following AI frameworks.

501

: 00:38:51

Number one, machine learning

frameworks and libraries, like deep

502

: 00:38:56

learning frameworks like TensorFlow.

503

: 00:38:59

PY Torch and Kara.

504

: 00:39:01

Classical machine learning libraries

like T Natural Language processing

505

: 00:39:08

libraries like hugging face transformers,

data processing frameworks for

506

: 00:39:16

model preparation like Apache Spark.

507

: 00:39:20

The core techniques used to

learn from data by AI systems

508

: 00:39:25

include supervised learning.

509

: 00:39:28

Unsupervised learning, reinforcement

learning, computer vision

510

: 00:39:33

processing, and generative models.

511

: 00:39:38

Again, we won't delve into

the technicalities of models,

512

: 00:39:42

methods, and frameworks here.

513

: 00:39:44

Remember, this is a means to an end and

not the starting point of AI systems.

514

: 00:39:50

As an executive or owner of

the business, you only have

515

: 00:39:54

to understand basic concepts.

516

: 00:39:57

Keep on top of developments.

517

: 00:39:58

Why does this pillar matter?

518

: 00:40:00

For obvious reasons, the frameworks

and models component is crucial.

519

: 00:40:04

Think of them as sophisticated

construction kits designed

520

: 00:40:07

specifically for building ai.

521

: 00:40:10

It's like a Lego brick for building a toy.

522

: 00:40:13

AI frameworks can accelerate

development because of prebuilt

523

: 00:40:16

components, libraries and function

524

: 00:40:22

complex.

525

: 00:40:24

Make AI development more accessible.

526

: 00:40:28

They also ensure you are following

standardization and best practices.

527

: 00:40:34

AI models and methods, on the other hand,

translate data into insights and actions.

528

: 00:40:41

AI methods help computers learn

patterns and relationships.

529

: 00:40:46

They also enable generalization.

530

: 00:40:49

What does this mean?

531

: 00:40:51

The AI.

532

: 00:40:53

To apply learned knowledge

to novel situations.

533

: 00:40:57

In other words, demonstrating

intelligent behavior.

534

: 00:41:02

Different AI models and methods can

address different and diverse problems.

535

: 00:41:07

So it's crucial to choose the right model

or method for your specific problem.

536

: 00:41:12

Frameworks, models and methods allow

you to explain how your AI work.

537

: 00:41:19

If you can explain the output,

you can trust the model.

538

: 00:41:22

Your stakeholders can have

faith in their outputs, but how

539

: 00:41:26

do you choose the right model?

540

: 00:41:29

Pick the simplest tool

that solves the problem.

541

: 00:41:32

Aim for models that are testable,

auditable, and explainable.

542

: 00:41:39

Firstly, understand the

problem and tasks at hand.

543

: 00:41:42

Consider what are you trying

to achieve with the ai?

544

: 00:41:47

What type of data do you have?

545

: 00:41:49

What are the inputs and

outputs requirements?

546

: 00:41:52

How accurate do you need the model to

be, et cetera, et cetera, et cetera.

547

: 00:41:57

Then explore available model types,

like classical machine learning,

548

: 00:42:01

deep learning, natural language

processing, and time series models.

549

: 00:42:06

Once you've chosen the model,

consider what data sets are needed.

550

: 00:42:11

Consider how you prepare the data

and how you deal with missing data.

551

: 00:42:18

Then go back to the infrastructure pillar

and decide whether you have that computer

552

: 00:42:23

power to run these complex models.

553

: 00:42:27

Remember, you can buy or

rent computer power, as I've

554

: 00:42:31

already spoken about before,

555

: 00:42:35

as an executive or business owner.

556

: 00:42:37

You need a working knowledge

of frameworks, models,

557

: 00:42:40

and methods used for ai.

558

: 00:42:43

More importantly, you must understand

the limitations and risks because

559

: 00:42:49

you have an oversight responsibility

on what your team produces, whether

560

: 00:42:55

they're in-house or outsourced.

561

: 00:42:59

As a leader, you have to educate

yourself and make sure you stay

562

: 00:43:05

up to date on AI mechanics.

563

: 00:43:08

Nothing too deep here.

564

: 00:43:10

A high level understanding

that's all you really need.

565

: 00:43:15

You must ensure there are governance

mechanisms in place to evaluate

566

: 00:43:20

the accuracy of the models.

567

: 00:43:22

What's even more important is to assess

the risks of bias, transparency of the

568

: 00:43:29

models and maintainability, meaning

the output remains relevant over time.

569

: 00:43:35

The model doesn't become corrupt as

it gets more data and learns on it.

570

: 00:43:41

While your framework, models and

method must be smart, they must

571

: 00:43:45

also be ethical and responsible.

572

: 00:43:50

What if you choose a framework, method, or

model poorly, you will invite unintended

573

: 00:43:55

consequences like biases, discrimination,

and losses that your stakeholders suffer.

574

: 00:44:03

You can't have a brain that is corrupt or

manipulative, so I stress again, strong

575

: 00:44:07

governance around AI is so, so crucial.

576

: 00:44:11

You've got a bold challenge, mechanisms

and auditability into the AI solution.

577

: 00:44:18

The governance mechanism must anticipate

and quickly identify problems and weak

578

: 00:44:23

points and rapidly put a stop to AI system

deployment if the risk assessment fail.

579

: 00:44:32

Okay, so let's explore

who's involved here.

580

: 00:44:35

Data scientists and AI experts

naturally, but you also need to involve

581

: 00:44:40

wider teams like compliance officers,

legal teams, ethical working groups.

582

: 00:44:46

It's a cross disciplinary effort,

not a solo data science show.

583

: 00:44:52

Remember, the most important

controlling mechanism is.

584

: 00:44:58

That's going to stop disruptive AI

systems from being deployed and that

585

: 00:45:03

keeps check on outputs regularly,

making sure outputs are not corrupt and

586

: 00:45:09

don't lead to unintended consequences.

587

: 00:45:12

Now we come to the T in the

AI Swifter framework, the T

588

: 00:45:17

tracking and learning pillar.

589

: 00:45:19

AI isn't a launch event done and dusted.

590

: 00:45:22

It's a continuous improvement effort.

591

: 00:45:25

Success doesn't happen at a go live.

592

: 00:45:27

It happens over time through

learning and adaptation.

593

: 00:45:31

AI is developing rapidly and you as the

executive or owner and your team must

594

: 00:45:38

always be on a steep learning curve.

595

: 00:45:41

You need to know about new

developments, but you also need

596

: 00:45:44

to understand new regulations.

597

: 00:45:47

Putting guardrails on

AI system developments,

598

: 00:45:53

this is critical because if you

don't measure, you don't learn.

599

: 00:45:57

And if you don't learn, you

can't scale or course correct.

600

: 00:46:01

If you don't keep up with developments,

you will be outcompeted by those

601

: 00:46:06

that follow the latest trends.

602

: 00:46:08

If you don't keep up with guardrails, you

may end up on the wrong side of the law.

603

: 00:46:13

So

604

: 00:46:16

how do you build tracking and

learning into the process?

605

: 00:46:21

I'll define KPIs, key

performance indicators.

606

: 00:46:25

Track usage data.

607

: 00:46:26

Talk to stakeholders, watch

for outliers and edge cases.

608

: 00:46:30

Keep an eye on latest developments,

some tools that can help

609

: 00:46:35

us tracking and learning.

610

: 00:46:36

Include dashboards, internal audits,

surveys, and bullet database to

611

: 00:46:42

keep track of latest developments.

612

: 00:46:45

Remember that discovery is a foundational

pillar that is relevant at every

613

: 00:46:50

pillar of the AI swifter framework.

614

: 00:46:53

Use what you learn to

continuously improve.

615

: 00:46:56

What happens if you don't get this

tracking and learning component,

616

: 00:46:59

or simply you are rescaling

something that isn't working?

617

: 00:47:04

Or worse?

618

: 00:47:05

Thinking something works

when it doesn't really work.

619

: 00:47:09

Your AI system becomes irrelevant

over time and you lose your

620

: 00:47:12

competitive positioning.

621

: 00:47:14

You could also end up on

the wrong side of the law.

622

: 00:47:17

Definitely something you

don't want to get into.

623

: 00:47:21

The who leads this pillar?

624

: 00:47:22

Naturally, your analytics team

play a crucial role, but tracking

625

: 00:47:27

and learning must be embedded at

every level in your organization.

626

: 00:47:33

This is essential feedback loop making

you confident that your AI system is

627

: 00:47:39

always on the right path and following

cutting edge development curve.

628

: 00:47:43

Now we come to the penultimate

pillar e Ethics and control.

629

: 00:47:49

AI has power.

630

: 00:47:51

And the saying goes,

power needs guardrails.

631

: 00:47:55

I've already touched on this when

explaining other pillars, but let's zoom

632

: 00:47:59

in on the vital pillar of the AI swifter

framework, which is ethics and controls.

633

: 00:48:06

Why do ethics matter so much in ai?

634

: 00:48:09

'cause one ethical failure and wipe

out years of brand equity and get you

635

: 00:48:15

into hot waters with regulators or

face lengthy and expensive lawsuit.

636

: 00:48:21

Trust takes times to build

and seconds to destroy.

637

: 00:48:27

So how do you embed ethics?

638

: 00:48:29

Start with a clear ethical

framework and principles.

639

: 00:48:32

Define your company's core

ethical values and translate

640

: 00:48:36

them into ai ethical principles.

641

: 00:48:39

These might include fairness,

transparency, always doing the

642

: 00:48:43

right thing, accountability.

643

: 00:48:47

Privacy and security.

644

: 00:48:49

Develop AI ethics policies based on your

values, regulation, and best practice.

645

: 00:48:57

Establish an AI ethics committee or

an officer if you're a small business.

646

: 00:49:01

Hiring in an ex AI expert that

comes in every now and again to

647

: 00:49:06

make sure everything is on track.

648

: 00:49:07

From an ethics perspective, I've

already mentioned the importance of

649

: 00:49:11

a strong governance framework, which

must be developed and documented.

650

: 00:49:17

You need a structured approach with

clear roles and responsibilities

651

: 00:49:22

and decision making process for AI

development, deployment, and monitoring.

652

: 00:49:29

Embed ethical considerations

at every stage of the process.

653

: 00:49:34

Conduct regular audits and

reviews for ethical risks, biases,

654

: 00:49:39

unintended consequences, and

compliance with laws and policies.

655

: 00:49:43

So what happens if you ignore this?

656

: 00:49:46

You open the door to unintended

consequences that results in lawsuit,

657

: 00:49:52

media backlash and regulatory

heat, which I've already covered,

658

: 00:49:59

and who's responsible

for ethics and controls.

659

: 00:50:03

It's gotta be embedded in every

pillar of the AI source framework.

660

: 00:50:08

Assigning responsibility for.

661

: 00:50:14

Outcome to specific stages

and roles throughout the AI

662

: 00:50:19

implementation lifecycle.

663

: 00:50:22

Risk teams and legal and compliance

will play an important role, but

664

: 00:50:27

here is where leaders must step up.

665

: 00:50:29

They set the tone at the top, they

lead by example, and they have to have

666

: 00:50:34

a strong governance framework that

anticipates unintended consequences.

667

: 00:50:40

Surfaces them rapidly deals with

them, so that risks are minimized.

668

: 00:50:47

And finally, we come to the R in the AI

Swifter Framework, results and Scale.

669

: 00:50:53

This is where the rubber meets the road.

670

: 00:50:55

Proof of concept is nice, but

proof of value, that's where

671

: 00:51:00

real transformation begins.

672

: 00:51:02

If you go through all the

effort of building an AI system,

673

: 00:51:06

it's got to deliver real.

674

: 00:51:08

Tangible results, not

just today, but ongoing.

675

: 00:51:12

AI solutions must be sustainable.

676

: 00:51:14

You also want these solutions to

scale so that they give you more

677

: 00:51:19

return on investment over time,

results and scale is the end game

678

: 00:51:24

because AI has to scale to matter.

679

: 00:51:28

It can't stay stuck in

one pilot after another.

680

: 00:51:32

Remember, pilot purgatory discussed in.

681

: 00:51:37

So how do you scale with confidence?

682

: 00:51:39

Modularize your solutions.

683

: 00:51:42

Make sure the modules are standardized

and accessible to other teams.

684

: 00:51:47

Remember, build AI like Lego blocks.

685

: 00:51:51

Once time and effort goes into innovating

and building one component, make sure

686

: 00:51:56

it's available and can be used by other

teams who are building other AI solutions.

687

: 00:52:03

Now you can develop and deploy AI

systems fast because you don't need to

688

: 00:52:07

go back to the drawing board every time.

689

: 00:52:10

You can just take what's

working and build on it.

690

: 00:52:14

This gives you a huge head start.

691

: 00:52:17

The modules must be hosted on a

digital platform that developers

692

: 00:52:21

can get controlled access to.

693

: 00:52:24

However, document everything,

how it works, how it's deployed,

694

: 00:52:28

what worked, what didn't work.

695

: 00:52:31

Toolkits that others in

the organization can reuse.

696

: 00:52:37

Here are some of the practical

assets you should be developing.

697

: 00:52:41

Internal written playbooks, videos

that guides videos that gives guidance

698

: 00:52:48

on the assets and how to use it.

699

: 00:52:51

A digital platform with reusable

modules of code governance

700

: 00:52:56

frameworks for scale and oversight.

701

: 00:53:00

What have you skipped this?

702

: 00:53:03

Well, quite simply, we'll be stuck

in pilot purgatory celebrating small

703

: 00:53:08

wins, but never changing the game.

704

: 00:53:12

Who drives scale?

705

: 00:53:14

Again, this is everyone's responsibility.

706

: 00:53:17

Everyone involved has to develop

things that can be reused.

707

: 00:53:22

Then document what they have done to make

life easier for someone who will use the

708

: 00:53:28

AI related asset, develop instructional

videos and modules, delivery team, program

709

: 00:53:34

managers, and executive sponsors must

drive scalability in everything they do.

710

: 00:53:40

So that's the AI Swift framework, a

grounded, structured way to navigate

711

: 00:53:44

AI in your business, not just with

hype, but with real practical clarity.

712

: 00:53:51

We are building an AI readiness

scorecard to help you assess yourself

713

: 00:53:55

across the nine pillars so you know

where to start, what to strengthen,

714

: 00:54:01

and where the real risks are.

715

: 00:54:04

I include a link in the show

notes to express your interest in

716

: 00:54:09

completing this when it is ready.

717

: 00:54:11

I'm your host, Jay Tick.

718

: 00:54:13

This is impact with digital.